Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
I \ I I Uf\L.nUL.
IC/T^/33IUTCERNAL REPORT
CLimited distribution)
International Atomic Energy Agency
and
United Nations Educational Scientific and Cultural Organization
INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS
ELECTRICAL BEHAVIOUR 0? LAKGMUIR FILMS:
A R E V I E W *
V.K. Agarwal
International Centre for Theoretical Physics, Trieste, Italy.
MIRAMARE - TRIESTE
May
* To "be submitted for publication.
ABSTRACT
The electrical behaviour of thin films obtained by a variety of
processes, e.g. thermal evaporation in vacuum, have been extensively studied.
But the study of organic mono- and multilayer films obtained by the Blodgett-to
Langmuir technique (commonly referredAas Langmuir films) has gained considerable
momentum only during the past decades. Unlike in evaporated films, the
striking features of these organic films are their controllable thicknesses
down to one monolayer (™ 25 A) and the possibility of obtaining them free
from holes and conducting imperfections .The aim of this paper is to describe
the film deposition techniques, some of the .properties of the films so obtained
and to review their electrical behaviour. It has been intended here to make
this review a comprehensive and up-to-date source of information for the
workers vho are either already engaged in this field or planning to adopt
Langmuir films for future investigations. Nevertheless, for greater details,
to suit one's particular interest, original papers must be consulted.
In this survey, emphasis has also been laid on the possible problems
worth further study to get more insight into the basic properties of theselatter
films. Further, since the^ possess some interesting electrical properties,
this paper may prove useful in the assessment of our depth of knowledge about
them and in reducing the existing gap between basic research and technological
applications. Their potential usefulness in developing devices has also been
discussed. It is very likely that further exploitation of Langmuir films should
prove to be one of the productive enterprises for the advancement of thin
film technology.
-1-
I. INTRODUCTION
Insulating thin films in the thickness range 100 A to 20 ,000 A have
been a subject of varied interests among the scientific community because of
their potential applied significance for developing devices, such as, optical,
magnetic, electronic, etc. Some of the unusual electrical properties
possessed by thin films which are unlike those of bulk materials, led to
think about their technological applications and,consequently, interest in thin
film studies grew rapidly. Earlier studies did not prove to be very inspiring
because the films obtained always suffered with the presence of pin holes,
stacking faults and other impurities, etc., and hence the results were not
reproducible. In the last few decades many sophisticated methods have become
available for the production and examination in thin films and reproducibility
of the results could be achieved and controlled to a greater extent. Never-
theless, the unknown nature of inherent defects in a wide variety of thin
film systems still complicates the interpretation of many experimental data
and thus limits their use in the devices.
In a recent bibliographical survey on breakdown conduction in thin
films the author found that the major subject of investigation have been
the films prepared by thermal evaporation under vacuum or such like methods.
It was thus realized that the Langmuir films, which are the subject of the
present paper, have remaine.d less known among the workers of this field. In
fact, films formed by oil on water surface vere known as early as 17T1* when, •2)
Benjamin Franklin reported their presence. Lord Rayleigh was the first who
proposed that these films were only one molecule thick. Rayleigh's hypothesis
of monomolecular spreading thus gave birth to a new "branch of science involving
3) Mthe study of physics and chemistry of surfaces . Subsequently, Langmuir
introduced the simple technique of transferring these monolayers onto a
substrate by moving the latter across the film-covered surface. This technique
of transferring the monolayers onto the solid surface by repeated dipping and
withdrawal process was later modified by Blodgett and Langmuir * and has
become the basis of a new field of thin film investigations. This novel
technique of "building up" mono- and multilayer films is commonly known as the
Blodgett-Langmuir (henceforth BL ) technique,and the films thus formed are
many times referred to as Langmuir films.
Using this simple technique, it is now possible to obtain monomolecular
spreading of a number of organic substances, such as, fatty acids and their
soaps, branched and unsaturated fatty acids, esters, dyes and many proteins,
etc. It appears that this is the only technique "by which it is possible to
deposit films on smooth substrates which are uniform, structurally oriented
-2-
and whose thicknesses are accurately known and controllable. It is now
well established that films which are largely free from defects and other
conducting imperfections can "be obtained and a wide range of electrical
characteristics can be studied in a variety of organic materials in thin
film form. However, we shall here be concerned only with "solid" monolayers
in which the molecules are closely packed and which can be deposited on solid
surfaces. Among the many substances which form a stable monolayer, the
straight chain fatty acids and their metal soaps are used for most electrical
measurements. This is, probably, because the fatty acids precipitated by
calcium or barium ions in the solvent were used initially ' * and the
technique was perfected for them.
Although, in principle, any substance forming an insoluble;stable and
condensed raonolayer can be used for building up multilayers, it is not
necessarily so. For example, the molecules containing a hydrophilic group
such as -OH or -COOH at one end of the chain and a hydrophobic group such
as CH, at the other have been found to form good monolayers, the molecules
being arranged almost vertically on the water surface and becoming close
packed when the monolayer is compressed. But not all such substances behave
this way, If the molecule is too short, the Van der Waal forces along their
lengths would be insufficient to hold them together as a "solid" film and,for
long molecules, the buckling of the molecules themselves may give rise to a
crumpled film on compression. Therefore, only the molecules of -inter-
mediate chain length can be used to obtain an oriented close-packed solid
monolayer on the water surface. However, there are no clear cut criteria
for the choice of substances forming good Langmuir films. Reference should bev)
made to the excellent text of Gaines for the choice of material,and a
recent review of Srivastava is a comprehensive source of information for
beginners in this field. In addition, the reader is referred to several moreQ)-lM
reviews and texts that have appeared before and cover the vast subject
on such films.
Langmuir films were successfully prepared as early as 1935 "but 3Ome
of their dielectric properties studied in around the same period did not yield
reproducible results. Renewed attempts were made in the Sixties but it was
not until the late Sixties that reliable results could be obtained. This is
because many further advances in the original B-L technique have "been made
during the last decade to obtain a good film for reliable .electrical measure-
ments. A handful of experimental data on electrical behaviour of Langmuir
films during the last few years yielded some very interesting electrical
characteristics which will be discussed exhaustively in this paper. There
-3-
is no '..-ther useful survey available in the literature covering an up-to-date
account of their electrical behaviour, Even the books and reviews published
recently or in the past on the whole subject of thin films do not mention the
work on Langmuir films. The one review devoted to Langmuir films carries
only limited ini'ormation about their electrical characteristics.
Essentially, the two crucial factors governing thin film properties are
their uniform thicknesses and structural perfection. Particularly, all the
electrical properties are very much sensitive to these parameters and thus,for
a judicious experimental study, one requires films of accurately known thick-
ness and free from gross defects. The Langmuir films, which concern us in
the present paper, are known to fulfill these requirements to a far greater
extent than other types of thin films. For instance, Mann and Kulin • have
recently shown through their electrical studies that the films obtained by
the modified B-L technique were free from even a small fraction of holes and
conducting imperfections. Secondly, the uniform thicknesses of these films
are accurately known,as measured by standard optical methods like Tolansky's
multiple beam interferometry, * and may be controlled and reproduced
successfully to a lower limit of one molecule thick (about 25 A ). In addition
to many other features described in the text of the paper, these two constitute
a sound basis for studying these films. Control of these two parameters in
other film systems is notoriously difficult to achieve.
To fulfill the aim of this paper, the film deposition technique of
Elodgett and Langmuir will "be discussed in detail. The recently proposed1 o \
evaporation technique to obtain such organic films down to one molecule
thick and with better purity than the evaporant, will also be described. The
physical properties of these films, which are relevant in the context of their
electrical behaviour, are described in the next section. Each of the following
sections covers a particular aspect .of the electrical behaviour and carries
a detailed account of the experimental and theoretical work (if any). To
make this survey more substantial, it has been thought desirable to discuss
briefly the existing theories about conduction mechanisms, "breakdown
behaviour, etc.,in the respective sections. This vould certainly prove useful
for the better understanding of the subject and may further help the researchers
to select out some suitable problems for their studies on Langmuir films.
Nevertheless, efforts have also been made to point out some of the possibilities
for future investigations on such films which require the specific attention
of the workers.
-k-
For the very fact that many reviews and "books have already been published
on the -whole range of electrical measurements in thin films, this paper is
devoted to Langmuir films only. However, to adopt a broad viewpoint, some
pertinent references have been cited just to make a comparison between the
results on Langmuir films and on other film systems. To summarize for a
wider readership and to make the survey an up-to-date source of information,
all the possible references since 1937 (when the earliest work on electrical
properties was reported) have been provided.
In the end, some space has been reserved to discuss briefly the future
potential of these films in the development of devices. In spite of the fact
that further electrical studies on Langmuir films would prove useful for a
"better understanding of the underlying electrical phenomenon in thin films,
it is also very likely that the use of such ordered film systems in developing
devices should lead to a major "breakthrough for the advancement of thin film
technology.
II. FILM DEPOSITION TECHNIQUES
1. Blod^ett-Lanamuir technique
This technique requires a very -simple apparatus consisting of a
long narrow trough T (Fig.l) usually made of inert materials such as teflon*
perspex, etc. Metals such as copper or aluminium are not used "because they
may contaminate the solution with metallic ions whose presence would stop the
deposition. The inside of the trough is heavily waxed. Waxing of the
trough is necessary in order to obtain a higher water level at the edges,
which ensures that there is no leakage of the monolayer past the barriers.
The barriers, rectangular in shape, such as glass slides, must also be
waxed all around to render them hydrophobic. The trough is first levelled
and then filled to the brim with doubly distilled or deionized water (specific
resistance ~ 6 x 10 ohm cm) having a low concentration of barium by adding
3 x 10 ^ M barium chloride or such like. For easier deposition, the pH of
the solution is adjusted to greater than 6 by the addition of a specific
quantity of an alkaline substance. For example, KHCO- at a concentration of,
-5-
k x 10~ M , thus holding the pH at 7-2. This results in smooth deposition
of Y-type films. The surface of the liquid is now thoroughly cleaned by
sweeping the waxed Carrier B across the long edges of the trough and finally
leaving it resting on the trough end (Fig.la). Another Carrier, A ,rests at
the other end of the trough which is used to hold the vaxed silk thread S
placed upon the water surface as shown between A and E . The thread is
vaxed so that it does not vet and sink into the vater and it is carefully made
to touch the vater surface at every point to avoid any possibility of leakage
of the spread monolayer. The technique here is specifically described for
barium stearate films and must be followed similarly for all other substances
eligible to form Langmuir films.
Stearic acid (or other material to be deposited) dissolved in benzene
(concentration = 3 x 10 by weight) is now placed by the micrometer syringe
on the vater surface near D. in the form of a small droplet vhich spreads
spontaneously and pushes the thread outwards to take the shape as shown in
Fig.lb. (if the first drop does not spread quickly, it is an indication
that the surface is not perfectly clean.) As soon as the spreading is
complete, the thread is fastened to the trough edges by two small clips C .
At this stage the divalent barium ions present in the solution undergo surface
reaction with stearic acid molecules to form the barium stearate soap
and the solvent (benzene) slovly evaporates leaving behind a monolayer on the
surface, its boundary being confined by the thread. If a second drop is put
near D after the spreading is complete, it would not spread and remain as
residue vhich is an indication that the available surface has been covered
with the monolayer. The stearate molecules at this stage remain standing
upright with their reactive carboxyl groups touching the water surface but
some of them also tend to bend over the water surface due to the relatively
large space available for them. A Email drop of oleic acid called "piston oil"
is now placed on the surface at P which thus exerts a constant lateral
compression of 30 dynes/cm on the spread monolayer. It has been found that
this much surface pressure ( <v 30 dynes/cm) is sufficient to compress a mono-
layer of stearic acid into a solid phase and the molecules remain relatively
closely packed. The thread boundary nov separates the whole surface area
between barriers A and B into two regions with a stearic acid monolayer
on one side of the thread and oleic acid on the other, which press each other
to give an equilibrium shape to the thread (shown in Fig.lc). The transfer
of monolayer onto the"conditioned" slide may nov be accomplished by moving the
dipper near D up and down repeatedly across the monolayer covered surface.
"Conditioning"of the slide here means to render it hydrophobic initially. This
••• nn be achieved simply by rubbing the slide surface with a waxy solid, e.g.
ferric stearate has been found satisfactory. It has "been found that a'pure
aluminium or1 silver film deposited on the cleaned slide by thermal evaporation
also behaves as a hydrophobic surface for good multilayer deposition.
The monolayer is thus transferred on the "conditioned" slide on both
its downward and upward journey, and the process may "be continued till the
thread S moves forward through an area to attain the shape shown in Fig.Id.
Every time the thread moves forward through an area equal to that of the slide
on which the film is deposited and the thread motion takes place because of
the constant compression provided by the "piston oil". This visible motion
of the thread is a very clear cut indication that the layers are being trans-
ferred on the slide every time on dipping or vithdrawal of the latter. The
film thus built up is termed Y-type and contains an even number of layersaccurarely
outside water. Film thickness can then be calculated/by counting repeatedly
the number of monolayers and multiplying it with the known monolayer thickness.
To obtain an odd number of monolayers outside the water surface, it is necessary
first to dip the slide in the solution and then to spread the stearic acid in
the stage shown in Fig. lls. In this case, the first layer will be transferred •
when the slide is moved upwards and finally an odd number of layers will be
obtained. Even to obtain one monolayer, this process has been found satisfactory
by the author in his studies with these films. To obtain X type of films,
it is necessary to make the solution more strongly alkaline (pH = 9) and then
the transfer would follow only during the down trips of the slide. It is
necessary to maintain the ambient temperature at 20 - 22 C during experi-
mentation and the substrate should be raised or dipped across the monolayer
very slowly and smoothly. To obtain a large number of layers on a solid
substrate, i.e. thicker filons , the author has found that it is not convenient
to use a trough of large dimensions and thus a large surface area for monolayer
spreading. This is due to the perfect cleaning of the surface becoming more
cumbersome as well as the molecules developing some tendancy to collapse. More
satisfactory results have been obtained by using a trough of lesser dimensions19)
{say 29 x 20 x 6 cm, as used by Nathoo yi) and using a,fresh solution after
every deposition cycle is completed. It is also advisable to
prepare the solution once in a large quantity to avoid any possible changes
in its pH value, etc., which may affect the nature of monolayer deposition.
The technique described above appears attractively simple in principle,
however,it requires one's scrupulous attention and experimental skill to obtain
structurally stable and smooth filma. It is therefore necessary to mention
some of the important precautions to be taken during experimentation,and these
-7-
follow. For necessary details references should further "be made to the
text "book of Gaines and original papers of Blodgett and Langmuir
One can also make use of commercially available highly sophisticated film
balances for film deposition.
a) List of precautions
i)The wax and chemicals used in the experimentation must "be of highest
grade quality and of extreme purity. Chemicals with E. Merck grade have been
found satisfactory. For still higher purity, the solvents (benzenej etc.)
should be fractionally distilled.
ii)The water used should be good quality conductivity water. For best
results deionized water should be treated with KMnO, (10 m) and NaOK (10 m)
and after heating it for about 10 hours at 70°C, it must be doubly distilled
in a quartz apparatus, It must "be remarked here that deionized water, if
used directly, may contain some organic impurities.
iii) Greatest care must be taken against dust contamination or the intro-
duction of any foreign material from the outside which may cause disruption
of the film structure. It is thus necessary to cover the whole experimental
set-up with a suitable transparent shield box and to design it in a way that
all the operations needed during experimentation, e.g. the movement of the dipper,
be operative from outside.
iv) Any sort of mechanical vibrations or other disturbances during
deposition may result in the cracks in the deposited film giving rise to voids
and inhomogeneities in the film, For this, the experiment should, preferably,
be performed on a rigid foundation or,if a table is used, antivibi-ation
mountings must be fixed to the four legs.
v)The specific concentrations of the substances must he added in the
bath solution to hold the required pH value of the solution. Any change in
its pH may markedly affect the nature of the deposited film, Specifically,
barium ion concentration should be properly maintained to avoid the presence
of free stearic acid molecules in the deposited film.
vi)Since the hands of the workers are themselves the most likely source
of greasy contamination, it is necessary to clean them as thoroughly as
possible. Specific attention is also required to avoid any surface active
contamination.
-8-
vii) Th<? film transfer should "be accomplished under as high a lateral
compression aa possible. The higher the surface pressure, the more compact
will the film be. Nevertheless,the use of oleic acid as "piston oil" ensures
a satisfactory film deposition without any voids in it.
viii) The first mnolayer should be transferred as slowly as possible to obtain
a smooth surface, because it acts as the foundation stone for "building-up of
the multilayers. It is also to "be observed that the slide emerging from the
water is completely dry "before being re-immersn:.
Since the whole phenomenon is on a molecular level, even small con-
centrations of impurities or contamination in any form may affect the results
markedly. Therefore, conclusively the key-words in the preparation of Langmuir
films are the purity and cleanliness of all the components used, as is in the
case of some surgical operation. It would be "best for greater details to refer19)to the recent work of Kathoo who has described at -length the preparation of
Langmuir films for electrical studies. One novel feature which has been19)introduced by Uathoo is the use of a device to raise and lower the substrate
across the monolayer covered surface. The PTFE piston used for the device is
pulled upwards by suction of a pump and falls dovn under gravity when the
suction is reduced appropriately. This device could control the speed of the
dipper in the range 0.1 mm/a to 1 mm/s.
b) Selection of substrate and cleaning procedure
Obviously, one of the major factors governing the degree of uniformity
of the deposited film in the above process, is the smoothness of the substrate
used. Therefore, considerable attention must be paid to the selection of a
substrate with smooth surfaces and subsequent thorough cleaning to minimize
the appreciable variations in the electrical properties. Microglass slides
having no scratches are first selected and rinsed thoroughly with deionized
distilled water. The planeness and the smoothness of the selected slides is
then checked by the standard method of matching their surfaces vith a master21)optical flat . On proper illumination, the formation of reasonably straight,
equidistant, parallel and smooth fringes show that the slide surfaces do not22)
have much curvature and are almost plane , The smoothness of such selected
microglass slides can be of a much higher degree than that of the usual optical
flat . This is, for example, revealed by the absence of "wriggle" in the
fringes of equal chromatic order (FECO) in extensive film thickness measurements
on a k layer "barium stearate film using "Gold Seal" micro-slides. The author
himself has used "Gold Seal" microslides in the electrical measurements and
found them satisfactory for obtaining uniform Langmuir films.
-9-
The selected microglass slides are then treated vith chromic sulphuric
acid (cold saturated), rinsed vith doubly distilled water and allowed to remain
in a caustic soda solution for several hours. Finally» these are thoroughly
washed using a strong jet of doubly distilled water and dried by a current of
warm air. Ultrasonic cleaning with Ultrax cleaning fluid and then washing
with distilled water, also results in satisfactory cleaning. Both of these23)
methods are used "by Bucher et al. in their work for building up of the
monolayers of dye stuff - arachidic acid. The degree or cleanliness of a
slide may then be checked by immersing it through a cleaned water surface
covered with a small amount of talcum powder> lycopodium or such like.
Presence of any greasy contamination on the slide would be detected by the
grains of talcum being pushed away while the slide is immersed. This simple2*0test was devised by Langmuir and may also be applied to control the
cleanliness of all the glass components used for the experiment. Absence of
any "breath figures" on the slide after highly breathing on it is also a good
test of the,cleanliness.
2. Evaporation method
18} ,The recent attempt by Baker to deposit thin films (.of the order of
one monolayer thick) of organic solids by thermal evaporation in vacuum opens
up a new field in respect of organic films. • It was reported "by Luff and25)
White that thermal degradation of high molecular weight materials takes
place during vacuum evaporation and thermal hazard is much lower in the case
of lower molecular weight solids. Thus, it was realized that thermal de-
composition of stable compounds is less likely to occur. Making use of these
arguments, Baker was successful in depositing thin films of stearic acid
(mol. wt — 28*1.5) and melissic acid (mol. wt 1+66.8) by thermal evaporation
in vacuum. The unchanged composition of both the evaporated films and the
evaporant residue thus evidenced for the absence of thermal degradation in
this case. He could obtain very thin films of the order of one monolayer in
thickness by maintaining a slow controlled rate of evaporation and other
experimental requirements. It has also been pointed out "by him that the
purity of the remaining evaporant and evaporated films was slightly higher than
the originally used acid salts and has ascribed it to be due to the "outgassing"
of the acid prior to deposition. This is a remarkable achievement indicating
the absence of any foreign impurity in these organic films unlike in the case
of other materials evaporated thermally. Whereas, this technique is of
particular interest in the field of tribology because of the possibility to
produce a monolayer film of boundary lubricant, the author feels it to be of
-10-
equal importance for the physicists to study the electrical characteristics of
such evaporated films. Although, Baker has not mentioned explicitly the
defect nature of these evaporated organic films, the absence of thermal de-
composition and reported "better purity are, evidently, the two important
features. None of the other works on evaporated film systems seems to have
reported these features. After this successful attempt, undoubtedly, future
workers will introduce many more refinements to obtain evaporated films of
several other organic solids and studies of their electrical behaviour would
reveal many hidden facts.
III. PHYSICAL PROPERTIES OF LANGMUIR FILMS
Many interesting physical properties of Langmuir films have been in-
Vestigated and these have been reviewed well in the existing literature
In this section, only those properties which have relevance in their electrical
behaviour are discussed briefly. Nevertheless, it has been thought desirable
to introduce this section because some of them require specific attention of
the workers in future investigations.
1, Nature and orientation
The nature and orientation of the deposited film is mainly governed by
the angle of contact between the monolayer covered solution and the solid
surface and hence the condensed monolayers may be obtained with different
molecular orientation, at will. These monolayers are characterized as X,
Y or Z type and their molecular arrangements are shown in Fig.2. The
molecules in the case of X and Z type of films are oriented in the same
direction and thus the surface of the film will be composed of a carboxyl
and methyl groups, respectively. On the other hand, in a Y type film, the
molecules in adjacent layers are oriented oppositely and the film surface is
composed of methyl groups. Of these three types of films, the one which has
been studied is the Y type in which the monolayer transfer takes place "both
ways, i.e. on each dipping and withdrawal of the slide across the surface.
On the other hand, in X films, transfer takes place only when the slide is
being dipped and,for Z type, it takes place only when the slide is withdrawn.
The orientation is termed "exotropic" when the methyl groups touch the solid
surface and the carboxyl groups remain away from it in the first monolayer. If
the orientation of these two groups is reversed, the monolayer is termed
"endotropic". Thus, an X type film is made up of a aeries of exotropic
-11-
layers; Y type is made up of a series of alternating exotropic and endo-
tropic layers and the Z type is made up of a series of endotropic layers.
However, it must be pointed out that Z type films are rather uncommon.
2. Uniformity and thickness
Two crucial factors for making satisfactory electrical measurements on
films are their uniformity and accurately known thicknesses and both of these
requirements are fulfilled to greater extent if one -uses Langmuir films.
Metrical thicknesses of several fatty acid monolayers have "been measured using
the best known optical methods. For example, the significant contribution in
this context has been due to Srivastava and Verme , They have
employed Tolansky's standard multiple beam interferometric • technique with
fringes of equal chromatic order (FECO) which ensured absence of any error
in their measurements due to surface irregularities. They thus obtained
extremely sharp and smooth fringes and determined thicknesses with high
accuracy. Subsequently, "C" spacings of these films vere also measured by26)
the same workers using microfocus X-ray technique and an excellent agreement
was found between C spacing values and those obtained by interferometry.17)Recently, Scott and Sirhata have used the above technique with some
17)modifications. They employed a protective collodion layer in measuring
thicknesses which improved the quality of fringes and hence the precision of
measurements. Use of the protective layer thus avoided the damage of the
monolayer in vacuum which is expected to have a marked influence on the27) 28)
overcoating deposited. In fact, it was indicated * that over 60% of
stearic acid film is removed by exposure to vacuum for 30 minutes.
As far as the uniformity of these films is concerned, it has already
been discussed that the selection of substrate must be made carefully. How-29)
ever, it has teen shown by Holt that increasing the number of monolayers
increases the degree of uniformity of the film,and many subsequent workers
have also noticed this fact. To achieve the required uniformity of the film,
the deposition of the first monolayer is very important since any voids or
imperfections in it, may lead to major disruption of the subsequent layers.
Therefore, one must make sure that the slide emerging from the solution is
completely dry before being re-immersed.
Whereas the films could be built containing as many as 3000 layersin the case of barium copper stearate simply by repeated dipping and with-drawal processes , some problems vere encountered when thick multi-
acid 6)layers of other fatty^salts vere deposited. Blodgett and Langmuir noticed
"fogged" appearance of thicker barium stearate films (300 layers) and also
-12-
observed the "cracking" tendency of these films which increased with increasing
thickness. They then indicated that addition of copper ion concentration
(ft 2 x lcf M) in the solution checked both these difficulties and facilitated
film deposition. Other factors which seem to crop up sometimes during
deposition are the gradual crumpling or collapse of the stearate monolayers.
Henke avoided them simply by spreading a fresh monolayer over the solution
after every forty or fifty dippings,which has already been recommended in the
film deposition technique for obtaining a large number of monolayers. However,
in the opinion of the author, the collapse behaviour of stearate monolayers
is very unlikely to occur. Otherwise their use in the commercially available(c.f. •Ref,7j
step gauges, whose functioning is based on the known and controllable
thicknesses of the monolayer, would have been impossible.
3. Film structure
The obvious choice for investigating the detailed film structure is the31)well-known electron diffraction technique. Germer and Storks , particularly,
investigated the molecular arrangement of deposited mono- and multilayer
films of fatty acids and their soaps using this powerful technique. In their
studies in transmission as well as in reflection by depositing films on thin
supporting organic foils of Resoglaz and on a clean metal surface, the
molecules were found to stand almost perpendicularly to the solid surface in
the first monolayer. Barium stearate molecules had a more precise normal
alignment than the stearic acid molecules. The hydrocarbon chains of
stearate molecules in the first monolayer were close packed but irregularly
arranged on a clean metal surface whereas regular arrangement waE observed
in layers on top of the first one in case of films on Resoglaz foils. Thus,
the layered structure of these stearate films was shown to form simple
hexagonal crystals with the symmetry axis perpendicular to the film. The
stearic acid molecules were found to form monodinic crystals with their
chain axis nearly in the plane containing the surface normal and the dipping
direction. In earlier studies ' , it has been shown that these films
consist of superposed sheets of oriented molecules and form positive uniaxial
birefringent crystals with the optic axis perpendicular to the plane of the
film. However, some structural changes due to thermal disorientation of the
molecules or to evaporation were noticed at relatively high temperatures vhen
the temperature was raised in electron diffraction studies . This
might be possible because the soap multilayer films are relatively soft and29)
have,low melting points. On the other hand, in recent investigations of Holt
remarkable thermal stability of such films has been reported. In fact, these
-13-
films could be heated at 90 C in damp air over prolonged periods (the melting
point of stearic acid is 69°C) without any damage. Further, Holt observed
no ehraga in th« electrical prop«JrtlM «vari though tna film* v«r« r«p*»t*aiy
and rapidly cycled from liquid nitrogen temperature to 50 C.in air.
Based on the above structural information obtained by electron
diffraction technique, these films were known to be monocrystalline in nature
and thus regarded as a special case of a layer-by-layer mechanical growth
forming almost "two-dimensional" crystals. There is, however, evidence that
barium behenate multilayers do, in fact,show an absence of crystallizationg
which has been demonstrated by electron micrographic studies . It has37)
also been reported by Khott et. al. that the formation of crystallities
may require some time after multilayer formation and no crystallities were
observed until 30-60 minutes passed. Mare and Messier have attributed
the two absorption peaks,observed at -100°C and -30°C, to dipole moments in
the amorphous and crystalline regions of the behenate layers, respectively. In
fact, they had no clear cut evidence for amorphous zones in the monolayers
and the qualitative interpretation offered by them was "based merely on the
assumption of disorganized structures. Therefore, the author would like to
stress here that in future investigations one must be careful in the inter-
pretation of the electrical data in terms of the film structure. Renewed
structural investigations would also be desirable to make sure of the presence
of amorphous regions in these films.
k. Defect nature
Many of the earlier workers have reported that organic filma obtained
by the classical process of monolayer transfer always contained holes, cracks39)
or such like imperfections. For instance, Epstein demonstrated in the
electron microscopic studies that the deposited films are discontinuous and
consist of clumps of molecules, when the transfer was accomplished at low1*0)
surface pressures. Similarly, Ries and Kimball observed discontinuous
circular "islands" or aggregates in the stearic acid monolayer transferred
at 10 dynes/cm. These observations were not surprising because the higher
the surface pressure, the more compact will be the film formed. In fact,
the monolayers have been shown to be continuous and homogeneous at higherhi)
pressures . Nevertheless, it would be an oversimplification to regard
the film transferred at high surface pressure as perfectly uniform, coherent
and defect free. Artifacts are introduced if proper care has not been tafcen
either during transfer processor the subsequent thermal evaporation of metal
electrodes over the film may disturb the film structure. In some radio-
autographic studies, the presence of spots, striations and other irregularities
like "folded over" regions in the deposited "barium stearate monolayer was2T)demonstrated in the earlier work of Roberts and Gains . On tha other hand,
1 )Handy and Scala also carried out radioautographic studies and have shown
that these layers are generally uniform with no apparent gross defects. One
of the reasons for the presence of one or the other type of defect in the
films demonstrated by these early workers is that the original technique
of B-L could not be perfected until then. Many of the recent modifications
in this technique have made it possible to obtain films largely free from
gross defects or imperfections -
The one work giving quantitative proof of the absence of even a small
fraction of holes and conducting imperfections in such monolsyers is that of Mann
and Kuhn • They had employed a modified B-L technique and took other
precautions. In their studies on "breakdown phenomenon of Langmuir films
(described later) Agarwal and Srivastava have also indirectly shown the
absence of "weak spots" and thus they inferred that the films are free from
gross defects. The development of striations in the deposited film was
ascribed to the irregularity of dipping and withdrawal motion of the sub-
strate . Recently, Nathoo has introduced a novel technique for this
and succeeded to obtain acceptable and reproducible devices for electrical
measurements with the use of this technique and other precautions that were ,
discussed above.
In spite of the fact that structural perfection may be achieved by
using Langmuir mono- and multilayers unlike in the evaporated film systems,
much attention must s t i l l be paid to identify the defect
structure of the films. Reliability and interpretation of electrical data
very much depends on the nature of defects and therefore special care must
be taken. As has been pointed out earlier,-tie deposition technique requires
one's experimental skill and enginery to obtain a structurally stable and
uniform film.
5. Skeletonization process
It is an established fact that the films of metal salts of fatty
acids or such like may contain free acid molecules, apparently in the solid
solution if the metallic ion concentration is low; their proportion depending
upon the pH of the solution ' . These free acid molecules can be
dissolved out of the multilayers by soaking the film for a short time in
benzene to which 1% of 90% ethyl alcohol is added . This process is
termed "skeletonization" and the film thus obtained is said to be
-15-
"skeletonized °} containing voids or holes in the places previously occupied
by the acid molecules. Optical studies have shown that the films remain
optically clear even after this treatment if the amount of acid molecules
contained is not more than 6o^but a striking change is observed in the inter-
ference colour reflected by the film. Further, it haa been shown that this
change in colour is due to a change of refractive index and not to a change
of thickness. If half of the acid content is dissolved, there is found to
be less than 1% of change in the film thickness. The stability of such a
skeleton film is not affected as indicated by its extraordinary rigidity even
if it contains as much as ^0$ of voids. The proof that skeletonized film did
not collapse was checked by Race and Reynolds who could restore the
original optical thickness by refilling the free spaces with some non-polar
oil. In fact, these holes and voids in the skeleton films may be refilled
by substances like petrolatum or other hydrocarbons; subsequently, if re-
guired, these substances can be dissolved in benzene giving again the same
skeletonized films. This method has already been used for getting the films?n \ much
with various refractive indices between 1.18 and 1.51. However not; attention
is paid to the study of electrical behaviour of such skeleton films carrying
foreign impurities like hydrocarbons. The author feels that the
method is very attractive for obtaining doped-like organic dielectrics as
is well known in the case of silicon, etc., and further exploration of their
electrical properties would "be worthwhile. Long back, Race and Reynolds 'the
made an attempt to study/electrical behaviour of such skeletonized films and
found the expected lowering of dielectric strength by skeletonization. The
concept of the physical structure of skeleton films, i.e. the remaining soap
and air (or any hydrocarbon substituted) are electrically in parallel, was
found to hold true in all the electrical measurements.
As discussed above, the removal of free acid from the deposited films
may also take place during their treatment in vacuum for electrode evaporation.
This process is known as "vacuum skeltonization". In fact, Roberts and Gains2*^' '
had noticed such removal of stearic acid from monolayers deposited on
various substrates. Evidently, the presence of voids or holes as reported
by many workers may be an effect of vacuum skeletonization. Many workers
failed to obtain useful devices for electrical measurements and it is presumably
because the spaces previously occupied by the free acid molecules might have
been filled with the metal during thermal treatment in vacuum and thus
shorting the device. Because of this fact, it is often stressed in the
literature that Langmuir films must be treated carefully in vacuum, particularly
when • electrical measurements have to be made. However, great caution must be
taken to hold the required pH of the solution because the proportion of con-
version to soap very much depends on it. Since the soap or metal salts of long
chain/acids are ionic substances forming stable monolayers (cf. Ref.7, p.193)n
there is a possibility that some ion content is present in the deposited mono-
layer which might get transferred during deposition. It has been shown
that with some cations, the ion composition of the deposited film is very
different from that on the water surface but much is not known about the
nature of such ions. The»presence of such ions and their motion under
applied electrical field has recently been pointed out by Barraud and co-1*5)
workers in their studies of thermally stimulated currents on MIM structures
with Langmuir films. A detailed study of the nature of ions and their
influence on electrical properties has to "be carried out in future works.
IV. DIELECTRIC PROPERTIES
Interest in the dielectric studies of Langmuir films of fatty acids
and their metal salts grew rapidly in the period just after the advent of their .
deposition technique. After that» such studies were taken up after a gap of
more than two decades when it was realized that Langmuir films can be uniform
in thicknesses and act as a good insulator. In this section the experimental
measurements on resistance, capacitance and dielectric constant of these film
systems are described. These measurements have been divided into two parts;
the first describes the work carried out in the late forties and the second
includes recent experimental work. Further* the dielectric loss measurements
and the theory for calculating the static dielectric constant are discussed,
and the readers are reminded to attach special importance to both of these
aspects in future works,
1, Earliermeasurements on capacitancet resistance and dielectric constant
The earliest work of Porter and Wyman (PW) reports the measurements
on impedance of films and related phenomena which they carried out on films
of Cu-Ba stearate and Ca stearate using both X and Y type of films. Initially,
mercury droplets were used for small area probe measurements and an a.c. bridge
was used for impedance measurements. The resistance of the films was found to
"be very low ( <£ 1 ohm) with high signal voltages, whereas it was of the order
of megohms with signals of 1 or 2 V. Ehe results were not satisfactory and
they then replaced the a.c. bridge by a radiofrequency bridge. The measurements
at frequencies of 1 MHz and 0.21+1* M H Z and for films containing
-17-
7 to lUl monolayers determined dielectric constant values ranging between 1.9
and 3.5 with an average value of 2.5. Considerable variation in the results
obtained on the same film vith different drops, or with the same drop in
different places, was due apparently to differences in the droplet size or in
the exact area of contact between the drop and film at different places. Ho
significant difference was found in the capacitance or resistance values between
X and Y films. In both types of film the capacity decreased with thickness,
as was expected from the relation used by them for capacity determinationj i.e.
in which 6 denotes the dielectric constant* TJ the number of layers, d
the layer thickness, and A the area of contact between drop and film. On
the other hand,the values of the resistance per layer showed a definite in-
crease with the thickness of the film. The specific resistance of films,
thus determined from their values of the resistance per layer, was about 10
ohms, having regard to the size of the drops and the thickness of the layers.
Some d.c. resistance measurements were also made by them using a drop of tap
water in place of a Hg drop,and a striking difference between the results was
found. With mercury drops the film ruptured permanently when a breakdown
voltage of 10 V cm was exceeded and the resistance was low. Whereas, no
such rupture was observed with water drops. Thus it was concluded that the
different affinities of the two liquids for the surface may be responsible
for the observed difference and probably the water molecules were able to
penetrate the films. Another conclusion arrived at by PW was the evidence
of a phenomena of power absorption in the films, increasing with frequency
over the range studied. This, in fact, was thought because the resistance
of the films measured at both frequencies (l and 0.2HU MHz) was far
less than^they determined from their d.c. measurements.
1*3)Race and Reynolds (RR) made some preliminary attempts to measure
the dielectric properties of barium stearate films in 1935 and found some
experimental limitations. For instance, the "bridge equipment used was not
sensitive; the area of contact of mercury droplets was not accurately
measured and the dust contamination could not be avoided, nevertheless, these
results encouraged them to develop a sophisticated experimental set-up and
necessary bridge equipment . Multilayer films were deposited on a clean
polished chromium slide and mercury droplets were used as the counter-electrode
All the measurements were made over a frequency range from kQ to 10 Hz '
using comparatively thicker multilayera ("both of X and Y type) of various
stearates and arachidates. The results obtained "by them are summarised below:
-18-
A) The dielectric constant of various multilayers was * 2.55 with standard
•variations of 3$. The variation was attributed more to the pH of the. solution
and therefore the proportion of acid to soap. .
ii)Dielectric constant was found independent of thickness as well as of
frequency in the range studied.
h3)Further, these workers measured the dielectric parameters of films
which were soaked for one to ten seconds in a solvent such as benzene. The
benzene dissolved the free acid leaving the soap film as a skeleton with air
filling the space previously occupied "by the acid. For calculating the
dielectric constant of such a skeletonized film, the remaining soap and the •
air spaces were considered in parallel. The relation thus used vas
• \
where € Q and e^ are the respective dielectric constants of unakeletonized
and skeletonized film; e = 1 is that of air and d,/drt- is the ratio ofa 1 0
densities of skeletonized and unskeltohized films. Their results on un-
skeletonized (0) and skeletonized (l) films of cadmium arachidate are given
in Table I. In the table, t and t. refer to the apparent optical thick-
nesses after and before skeleton!zation determined by Blodgett . To check
the collapse behaviour of the skeletonised films, the air space was refilled
by allowing a drop of oil to run over the film and the dielectric constant
was measured again and calculated usingEq. (k.2)by substituting the dielectric
constant of oil for e . They used non-polar oil as well as a polar liquid
(tetrachlorodiphenyl) to refill the air space in the skeleton, the corresponding
values of the dielectric constant have baen represented by € and € ,
respectively, in Table I.
As is evident from Table I, the agreement "between measured and cal-
culated values of e. is sufficiently close. Obviously, the dielectric
constant of the film (€_) increased, as is expected, whenthe air spaces in
the skeletonized film were refilled with a non-polar oil (columns 8 and 9)
having dielectric constant of 2.15. This was interpreted to constitute
a check on the structural conception of a skeletonized film, i.e. the
assumption that the remaining soap and air are electrically in parallel
(columns 6 and 7). However, considerably higher calculated values of £
than the measured one (columns 10 and '11} are found with a polar liquid
{dielectric constant 5.22), which were interpreted as indicating a reduction
of polar orientation of the liquid within the matrix of cadmium arachidate
molecules.
-19-
Buckwald, Zahl and co-workers -7'*' also became interested in the
measurements of the resistance and capacity of alkali stearates consisting of
varying numbers of layers and deposited on metal surfaces. Preliminary
measurements were made on barium stearate films " and,later, calcium stearate
films were chosen for studies. A special cell was designed by them consisting
of a cylindrical glass vessel partially filled with calcium chloride solution
or such like. Two metal electrodes were immersed in this solution and films
to be studied were deposited directly at the electrode surfaces. Using a
slightly modified conventional a.e. Wheatstone bridge network, a number of
readings for films (l-Ul layers) indicated the decrease in capacitance with
increasing thickness forone set of their measurements. However, in another
set of data, an increase to a maximum of three layers was observed after which
it started decreasing. ' The latter data has been attributed to the presence
of serious imperfections in the films and the earlier one is believed to be
the true. An average value of dielectric constant C^-3.7) for kl layers
was thus calculated from the observed capacitance values ( 1CT F). They
found two-fold variation in the apparent dielectric constant and several orders
of magnitude variation in the resistance in the thickness range studied. Further,
both the capacitance and resistance were found dependent on the nature and
concentration of the electrolyte. From the studies using copper sulphate as
an electrolyte of varying concentrations, they inferred that the significant
variation of the electrical (and presumably the structural) properties of
the film may be due to the presence of the copper ion because these values
differed widely from those taken vith Ca Cl as electrolyte. Finally, they
have emphasised that the stearate films may be electrically modified by water
content of various metallic salts used as electrolyte solution and the fact
ie that the mercury-droplet method of RR presents many advantages over
that used in their work.
2. Recent measurements on capacitance, resistance and dielectric constant
h2)
Handy and Scala after more than two decades realized that mono-
molecular layers obtained by the B-L technique satisfy all the conditions
required for reliable electrical measurements. For instance, such organic
layers are thin, homogeneous, uniform in thickness and an excellent insulator.
In their detailed investigations, they tried to obtain a more compact film to
minimize the presence of voids or holes in the organic multilayer and used
evaporated metal electrodes. Fig.3 shows the results of capacity measurements
on some 75 samples in which reciprocal capacity (l/c) is plotted as a function
of predicted number of layers (H ). The capacitance measurements thus
-20-
performed on stearate films (1-10 layers) led to e values between 2,1 to k,2
with a bulk value of £ = 2.5. These results agreed well with those obtained
by PW using liquid contacts. Tig.k ahows a typical graph of the capacity
variation as a function of frequency (upper curve) over the frequency range
0.1 - 20 KHz . On the basis of this curve, Handy and Scala predicted a slight
dependence of € with frequency {a drop of about 5% over the range studied).
The lower curve of Fig.l+ shows the variation of the imaginary part of the
complex dielectric constant (proportional to C X T , D being the dissipation
factor) with frequency. Evidently, there is a slight maximum near 700 Rz
which has been taken to be the indication of the presence of a weak polar
adsorption mechanism with a characteristic relaxation time (i irfrn) of
0.23 m sec. The measurement uncertainty in the dissipation factor D becomes
comparable with the observed values towards the high frequency region (> 10
KHz) (see the lower curve of Fig,!+).
The contribution of the oxide layer formed between the lower metal
electrode and the organic film was neglected in presenting capacity measurements
in view of its small resistivity compared with that of the organic layers
measured. In the ohmic conduction region «< 50 mV), a wide range of variation
of resistivity for the same film thickness was observed which supported the1+2)
contention of a porous structure of these films. They thus concluded that the
resistivity obtained for smaller thicknesses is characteristic of the oxide
film and its upper limit for a given film thickness corresponds to the film
possessing the smallest fraction of voids. The important conclusions arrived
at were that the poor reproducibility of the characteristics is due to voids
and inhomogeneities in the film and a reactive and hence oxidized surface is
necessary for the formation and retention of low porosity layers. Handy and
Scala also found that the physicochemical nature of the substrate influenced
the adhesion of more than the first monolayer and that the porosity alone was
considered unlikely for such long-range effects. Positive ion adsorption,31)
as described by Goranson and Zisman has been suggested as a possible
alternative and was shown to be consistent with some aspects of the electrical52)
measurements. Beck has reported measurements on capacitance for fatty
acid monolayers of different chain lengths sandwiched between Al electrodes1
and found reproducible capacitance within 3% unlike that of Handy and Scala
who found a variation of 50% in capacitance.
Drexhage and Kuhn have measured the capacitance of cadmium
arachidate and cadmium stearate films sandwiched between aluminium electrodes.
Astonishingly, they found this method suitable for the determination of film
thickness even for a single monolayer, because the capacitance C is given by
-21-
Eq..{*i.l) . As is expected, a linear dependence of l/C with the number of
monolayers (M) vas found. Evidently this leads to an independent nature of e
with thiclcness. The slopes of the straight lines in a plot l/C vs U thus
yielded a value of d= 26.6 A for cadmium arachidate and d = 2k.6 A for
cadmium steaxate "by using € = £. 1+0 (obtained simply by the square of the value
of the refractive index of barium stearate determined earlier ). Holt
with a possibility of using Langmuir mono- and multilayer films as thin film
dielectrics carried out preliminary measurements on stearic acid and
stearate films sandwiched between metal electrodes. The standard deviation
of C/A on each device was as low as 3$ which he suggested to be of the order
of accuracy of producing and measuring the areas of the base and counter-
electrodes. No significantvariation of C and hence £ with frequency were
found. The interesting properties thus predicted by Holt, i.e. thermal
stability and uniformity of organic films, have already been discussed. Another
important suggestion made by him is that the dielectric constant can be varied
by altering the metal stearate proportions in the multi-monolayers.
In many of the measurements discussed above, the organic film was
sandwiched between evaporated aluminium electrodes. The fact that natural
oxide layer grows on the base aluminium electrode was realized but its effect1+2)
on the capacitance values of the device was neglected considering that the
resistivity of the oxide film is small compared, with the resistivity of the
organic layers. In the work of Mann and Kuhn , the presence of such thin
oxide layer, between metal electrodes and fatty acids is considered; the
relation used by them is
where d and e are the thickness and dielectric constant of the oxideox ox • the
layer, C is a dielectric constant of air and all^other symbols have the
usual meaning. For calculating € values, they plotted l/C as a function
of superimposed layers (N) and,from the slope of straight lines thus obtained,
the d/e values were determined. A plot of l/C vs Hi is given in Fig.5 for
stearate and behenate films which shows the same y intercept in both cases.
The capacitance and hence the e values were determined by MK for a series of
fatty acid monolayers of Cd salts. For instance, the values for stearic
acid £ « 2.71 ± 0.17 and for arachidic acid £ = 2,52 ±. 0.07 determined by
them were found to be comparable with respective values «" = 2.59 ± 0.08 and
e, a 2.1*9 ± 0.06 determined earlier by RR . However, their 1 5 calculations
show an increasing € with decreasing chain length which is qualitatively
-22-
explainable due to the influence of the highly polarizable cadmium carboxylate
groups which increases with decreasing fatty acid chain length.
Recent measurements of capacitance by Leger et al. " ' have also
shown a linear dependence if l/C with .respect to the number of transferred
monalayers . They made use of this simple method for knowing the thicknesses
of organic films and found it to yield satisfactory results. In an attempt1Q)
by Uathoo , a change (of less than 10?) in the capacitance of stearic acid
films, increasing slightly towards the low-frequency region, has 'been reported
in the frequency region 10~ to 100 Hz .
More systematic investigations have been made by Khanna and Srivastava
recently for measuring the capacitance and hence dielectric constant of a few
fatty acid salts. In .their measurements» sandwich structures of the type Al-
film-Al were employed and the capacity values were determined initially without
considering the effect of the oxide layer . For all the film systems,
measurements were made on 8 and 110 layers using a universal bridge. The
capacitance as well as the dielectric constant values were found to be
distinctly different in the two cases. However, all these results were in
good agreement with their theoretically calculated values of e for monolayer
and multilayer of all the substances." The values obtained by Khanna and
Srivastava, experimentally as well as theoretically are tabulated in Table II.
A slight variation in these values arises because their theoretical
formulation ?'' >? is based on an ideal structure of the film which is59) 'difficult to achieve in practice. Subsequently, Khanna calculated the s
values considering the effect of the oxide layer because this can be electri-
cally insulating and can contribute significantly to the capacitance of the
sandwich structure. For this, the expression already employed by Mann and
Kuhn (Eq,(^.3)) was used and the values for d = 30 A and £ = 8 wereox ox
taken. The e values thus obtained have also been given in Table II.
Evidently, the effect of the oxide layer is more significant in the case of
a monolayer as compared with that for multilayers.To justify the significant difference of e values for monolayers and
59)multilayers obtained theoretically as well as experimentally, Khanna carried
out a systematic and detailed study of thickness dependence of the dielectric
constant from 1 to 80 monolayers. It. vas observed that the dielectric
constant increased slowly with thickness initially but saturated at a
particular thickness ( 21 1000 A ) for all types of Langmuir films. A typical
graph is shown in Fig.6 in which the dielectric constant c is plotted as a
function of film thickness d for barium stearate films. The upper carve (B)
represents the variation of e when oxide layer effects were taken into
-23-
account and the lower curve (A) shows when no such effects were considered.
This unusual behaviour has not "been reported in the earlier works and the
constancy of the dielectric constant with respect to thickness has been
shown, * * * However, Khanna has explained her results qualitatively
as given below. Mainly, the observed dependence has been attributed to the
porosity of films in the lower thickness range which decreases with the in-
creasing thickness. In fact, an exactly similar thickness dependence of
dielectric constant has been observed in evaporated ZnS films , which has been
interpreted in terms of decreasing porosity or increasing continuity of films
with increasing thickness. The saturation of dielectric constant at higher
thickness range has also been observed for evaporated ZnS films. To explain
why the other workers failed to obtain such dependence, it has been suggested
that the films obtained by them were more compact. For example, in the present
wort, the film transfer was accomplished at a surface pressure ^ 30 dynes/cm^
whereas in the work of Handy and Scala , a high surface pressure of 39 dynes/cm
was used. Ries et al. * have already shown that voids in the monolayer
are inevitable if the transfer is accomplished at a surface pressure significantly
below the collapse pressure {1+2 dynes/cm in the case of stearic acid films).
Further, the thickness of about 1000 A at vhich the a value saturates has
been predicted as a rough estimate of the range of electrostatic interaction
in Langmuir fatty acid soaps.
The author feels, however, that the explanation offered above is in
no way sufficient for this unusual behaviour of dielectric constant. Presumably
the influence of highly polarizable metal carboxylate groups which increases
with increasing volume ratio would be a sound explanation for increasing €
with decreasing chain length for a given number of layers . For the
part that £ saturates after increasing with thickness, the qualitative
explanation may be that dielectric lossess are significantly less at higher
film thicknesses and therefore the dielectric constant does not change. The
magnitude of dielectric losses has been related to the compactness of theog \
monolayers in recent measurements of Marc and Messier and it has been shown
that it decreases with increasing compactness. nevertheless, rigorous
experimental work is required to establish such unusual behaviour and to in-
vestigate the role of structural defects like voids, holes or inhomogeneities
on the dielectric constant of Langmuir films. One more suggestion the author
would like to make is that the observed increase of £ may be understandable
in terms of the increasing polarization with increasing number of monolayers.
This, in fact, has already been shown by Tanguy and Hesto experimentally,it
but/requires some structural changes inside the film,vhich,in itself,is aproblem yet to be studied. This complex situation thus requires the specific
attention of the experimentalists.
-2k-
3. Theoretical work
Unfortunately, not much, attention has "been paid to developing the
theoretical formulation of dielectric parameters of Langmuir films. The only-
attempt in this direction has been due to Khanna and co-workers ->TJ»5o) ^ o
developed a theory for calculating the static dielectric constant of Langmuir
films. Their theory is "based on the layered structure of Langmuir films
constituting a simple hexagonal array with the molecules oriented normal to
the supporting surface and separated "by a distance 1+.85 A. ' Muller's
assumption for calculating the Lorentz field for simple hexagonal lattice
could not be applied here because the molecular size ( e 25 A) is large
compared wibh the lateral intermolecular distance {k.85 A ) . The local field
has been calculated by summing up the externally applied field and the field
due to all other dipoles within the specimen. A reasonable assumption has
been made for this purpose that each .molecule is divided into the small CH
groups separated by 1.27 A along the chain axis (androf course,the end carboxy-
late group). Further, all the dipoles are assumed to be point dipoles, parallel
to each other, with their axis taken to be the external field direction (Z axis).
a) Monomolecular films
Following Salem's assumption that D, the mutual distance between
two parallel linear chains, can be chosen as small as about k A if the basic
units (in the present calculations, GHp groups are the basic units) are bonds
of average length 1.5 A , an expression for the contribution due to inter-
molecular interaction between two molecular hydrocarbon chains was obtained
- p U-x)
where p is the moment of the basic CH9 unit (p - «E ) , oi is the electronic—2li 3 10c
polarizability of the CH group (1,81 x 10~ cm , cf.Ref.66), \ is the distance
between the two CH_ groups along the chain axis (1.27 A) and N is the number
of CH groups along the molecular chain. The electronic polarizability here
is assumed to be isotropic and the two chains of fatty acid soap molecule are
considered to be coincident. For calculating the effect of the end carboxylate
group (unit), the relation derived was
El(end) = pend
2 2 -i2X7 - D n2X2 -
-25-
J-"U
where p , is the moment of the end unit and X. is the distance of the i,end 1CH unit from the end unit along the Z direction.
For calculating the intramolecular contribution to the local field the
interaction between the units of the same molecule was calculated using the
expression^—' 2p,
4 = > -^ * (U.6)i Zl
where p. is the moment of the i unit and Z. is the distance "between the
reference and the different unit along the Z direction.
b) Multimolecular films
In case of multimalecular films, the molecules are considered to be
continuously distributed in all the layers except in the one in which the6k)
reference molecule is situated. Thus, following Muller's- treatment for
layer lattices, the contribution of the continuous distribution to the local
field is inrP (P is the polarization)and that of the reference layer is the one
calculated for a monolayer by the above method. The density of soap films is
taken here equal to that of the bulk which is not a very good approximation.
For calculating the dipole moment, the total polarizability has been
taken as the sum of electronic and atomic polarizabilities of all the groups
and the orientational polarizabilities of the polar end group. However, the
atomic polarizabilities, being small,are neglected. The contribution due to
metallic ion presence in the end group have also been neglected in the cal-
culation of orientational polarizabilities and the end group is assumed
equivalent to the - COOH polar group, its dipole moment being 1.5 debye
(cf. Ref.67). Finally, the static dielectric constant e. along the field
direction (the symmetry hexagonal axis) which is one of the principal axes of
the dielectric tensor, is calculated using
€ = 1 + UTP/E • (1.7)s
Using this theoretical approach, they first calculated e for barium5
stearate mono- and multilayers and found these values to be 1.6 and 3.0,
respectively. Later they J reported corrected values of € for bariums
stearate as 1,7 and 3.1, respectively, for a monolayer and multilayer. Here
another assumption was introduced; that a barium stearate molecule is equivalent
to the two "effective molecules" each consisting of one hydrocarbon chain and
the half of the end carboxylate group. These "effective molecules", were
regarded to form a hexagonal array with their axes normal to the supporting-26-
57)surface. Their previous assumption, that two hydrocarbon chains are
coincident* was hot consictent with the film structure reported by Germer31)
and Storks . Further calculations of s: for a few other fatty acid soaps
were carried out ° and these are given in Table II. All these calculations
are done only up to seven nearest neighbours,and short-range interactions being
small have been neglected. The surface effects have already been considered
in their calculations on monomolecular films and these are assumed to be in-
significant for multimolecular {sufficiently thick) films.
It is evident from the above review of dielectric properties that our
knowledge about them has advanced somewhat through experimental measurements
but theoretical understanding is particularly lacking- It is therefore
necessary to attach special attention to more rigorous theoretical work.
The only theoretical formulation of Khanna et a'l. ' described above has
been based on a simple classical approach^and a number of assumptions made
do not hold in true experimental conditions. For example, Miiller's treatment
for layer lattices,which has been followed for multimolecular films, no longer
holds if the macroscopic volume of the sphere is large enough to permit the
molecules to remain embedded in a continuous medium with the same dielectric
constant. The interaction of the sphere with surroundings thus requires
electrostatic calculation which would be worthwhile to account for in future
works. Metal ions such as Ca have been shown to reduce the mobility of the
polar groups by linking one molecule with another . The contribution of
such ions in the calculation of electronic polarizability of the end carboxylate
group [such as (C00)p Ba] has "been neglected in the above calculations. Many-
more assumptions in their approach require due attention. More interesting
would be the extension of their work to time-dependent fields which obviously
would involve the complex dielectric constant of the system and € would
become a special case of such a formulation. Both the real and imaginary parts
of complex dielectric constant» would be frequency-dependent and the imaginary
part would "be of more relevance to account for the dielectric losses. More
problems would crop up if one enteredinto the quantum-mechanical approach but
their description is beyond the scope of the present paper. For interested
workers, reference should be made to the monograph by Frohlich - which
is a comprehensive collection . on the theory of dielectrics.
k. Dielectric loss measurements
Some preliminary dielectric loss measurements were carried out in the
late Forties by RR . They found that the dielectric loss neither depends
upon the frequency, in the range from ko to 10 Hz, nor is it affected by the
-27-
film constitution. The order of magnitude determined vaa low, usually lessI40) 29)
than 0.001 times that determined by Buckwald et al. • Similarly, Holt
also obtained some information that the tangent loss angle (tan 6 ) drops from
0.02 at 225 A to 0.006 at 3700 A . The variation of tan5 with frequency was
predicted to follow normal thin film behaviour. However, these measurements
did not furnish any useful information except the order of magnitude of tsn<5 ,
since the standard deviation was 25%.
A systematic study of dielectric losses in Langmuir films of calcium
stearate and behenate has been reported recently by Marc and Messier
The device used is Al-film-Al and the measurements were performed in the
temperature range (-170° C to +50 C) and over the frequency region (100 Hz -
50 KHz). All measurements were made after a 2U hour degassing under vacuum,
since the dielectric losses decreased rapidly by a factor of 2 during the
first hours of degassing. The results thus obtained are analysed by means
of the equivalent circuit shown in Fig.7, where C and S refer to themm
capacitance/area and resistance of the organic monolayer and C» and RQ to
that of the aluminium oxide layer. R is the series resistance of several
ohms due to the metal electrodes,which introduces series losses of the form
2irRf; f denoting the frequency. Since the loss angle of the device (tan<$ )
varied as f at high frequencies, they have subtracted the series losses
arR C co systematically and considered the following quantity:
s
0 m
Evidently this
tan 6 - RC
•
quantity
C u
P
where ( n )u m
is proportional to the losses per unit
area in the alumina and the monolayers. Further» it was assumed that all the
N monolayers are equivalent and therefore can be represented by H impedances
in series. Calling d the thickness, e the relative dielectric constantm m
and p the resistivity of each of the monolayers, P may be given by^n s '
P
It thus shows P to depend linearly on H . - Fig.8 shows the variation ofs
loss angle of the behenate film (tan S ) measured at 1 KHz as a function ofm
temperature. The curves have been plotted for films consisting of 7> 13
and 25 monolayers which vere transferred at a surface pressure of UU dynes/cm.
-28-
Evidently, two absorption peaks are observed around -100 C and -30° C . The
peak, around -100° C is present in all • . three films- studied,whereas the other
appears only in thick structures (M = 13 and 25) which are known to crystallize
more easily. In a separate measurement, they found that the latter peak
disappeared when the structure of the film is disorganized (see curve b of
Fig.9). Curves(a)and(b)ln Fig.9 represent dielectric loss against temperature
measured in two places of the same film. Curve(a)corresponds to a point in
the middle of the film where the molecules are well organized and
curve(b)to a point near the edge of the substrate where the layer might get
disturbed due to dipping or withdrawal of the latter.
All the results thus obtained in their work are summarized below:
i) Tho-losses per unit area increase linearly with the increasing
film thickness, thus suggesting the movement of dipoles linked to the organic
molecules. This is in accordance with the expression showing the relation of
P and H (Eq. (1+.9)).5
ii) The losses due to the oxide layer are much higher than due to
the organic films and particularly for monolayer, the losses due to oxide are
dominant.
iii)The tan &m do not change much with frequency. The rise in the
measurement frequency merely causes a slight displacement of the tan 6"m
curve towards high temperature,
iv) The1 dielectric losses are much greater when the structure of the
monolayer is disorganized (see Fig.9). However, the order of magnitude of-3
tan 5 is 10 for a compact monolayer which they believe to be relatively
smaller than expected. For its explanation, their suggestion is that the Ca
ions considerably reduce the mobility of the polar groups by linking the
molecules,and the rotation of a dipole around the molecular axis introduces
fev losses.
v) The two absorption peaks observed around -100° C and -30° C have
been related to the degree of organization of the monolayers. Finally, based
on some experimental evidence of getting similar peaks in several other types
of organic substances {in polyethylene a peak similar to that at -100° C ;
in stearon and nonapentacontanone a peak similar to that at -30° C) these
peaks have been interpreted in terms of the dipole movements. The respective
peaks are thus explained to arise due to the movement of the dipoles in the
amorphous region of the monolayers and due to the molecular rotation around
the long axes in the crystalline parts of the layers.
-29-
The interpretation in terms of the movement of dipoles in these layers
does not seem to be convincing because the measurements are carried out under
homogeneous applied fields- In fact, motion of dipoles is possible only when
non-homogeneous fields are applied. Since the soaps or metal salts of long
chain fatty acids are known to be ionic substances forming stable monolayers
(cf. Ref.T, p.193) there may be some ion content present in the deposited mono-
layers. Therefore, it appears to the author that the observed peaks may be
attributed to the motion of these ions rather than the dipoles. The existence
of amorphous region in such organic monolayers, which these workers have thought
of, has not yet been verified in any other experimental work. Nevertheless,
this is the only systematic work on loss angle measurements which distinguishes
between the various kinds of losses and thus led to estimate, in particular,
those arising due to Langmuir monolayers themselves. In presenting their
measurements, they were able to exclude the dielectric losses due to the natural
19)
oxide layer whose capacity was determined to be 920 nF/cm
It is interesting to point out the recent studies on tan5 by Nathoo
for a 23 layer film of stearic acid at room temperature over the frequency-U
region 10 to 100 Hz . Under these experimental conditions, no peaks were
observed and the dielectric losses were found to decrease by an order of
magnitude with increasing frequency in the range studied.
V. ELECTRICAL CONDUCTION PROPERTIES
Greater emphasis was laid on studying electrical conduction through thin
dielectric films, particularly in the period following the second World War.
It was realized that the conductive properties of films should play an important
role in the development of devices. Most of the earlier work on conduction
behaviour has, however, "been done using either vacuum deposited films or
thermally grown oxide films on base metal electrodes. The reason which com-
plicated the interpretation of experimental data, has been the inherent defect
nature of such films. No doubt, highly sophisticated techniques have become
available now for depositing thin films, nevertheless, it is hard to accomplish
a judicious study of electrical conduction in thin dielectric films because of
their "structure sensitive*1 properties. The electrical conduction in films
has already been widely treated in the literature in recent years. The readers
are advised to refer to the recent 1
comprehensive picture of the subject.
69)are advised to refer to the recent book by O'Dwyer who has given a
-30-
Because of the defect nature of the films, it is not possible to
provide interpretation of experimental data using only one of the earlier
concepts of field emiiBion from a freo-electron metal by Fowler and Wordheim,71)
or metal-vacuum-metal junctions "by Frenkel , or metal-insulator-metal72)
junctions by Sommerfeld and Bethe , etc. In fact, the conduction process
has been recognized as an involved process and many co-operative phenomena
such as space charge modulation (due to trapped carriers) or,say, the Poole-
Frenkel effect or Schottky effect have been found to occur. In the oontext of
this paper, basic theories of conduction phenomena in thin insulating films
are discussed briefly just to remind the reader/the underlying basic concept.
The simplest expressions used for the interpretation of experimental data are
also given.
1, DC conduction theories in thin filmR
All the theories discussed here will apply to metal-insulator-metal
systems and the insulator in question may contain a significant proportion of
impurities. In general the electron transport at a metal-insulator interface
depends upon the nature of contacts and the conduction processes, in turn, are
distinguished as follows:
i) If the contacts are ohmic (if <• . ) , the conduction in the systemm 1
is limited by the rate of electrons flowing through the bulk of the insulator
rather than the rate at which the electrons are supplied by the electrode.
The process is then termed "bulk limited".
ii) In the case of blocking contacts (ip > ty±) , the rate of electrons
flowing through the system is limited by the rate of electrons flowing over the
interfacial barrier and the process is referred to as "electrode limited".
iii) Tf the contacts are neutral, i.e. ^ = ^ , there is no necessity
of charge transfer between the electrode and insulator. This situation
corresponds to a transition between an ohmic and blocking contact. Under
special circumstances, neutral contacts also occur if 1J1 < ij and the trap
level is positioned more than 1 eV above the Fermi level. Traps in insulating
films arise because of the grain boundary defects or because of the large
stresses, particularly, in vacuum deposited films. These traps are an
"extrinsic" property of the amorphous insulator and their concentration may beig _^
as high as 10 y cm J .
However, if the system is subjected to sufficiently high voltages,
transition from bulic limited to electrode limited processes,or vice
versa,may occur again depending upon the nature of contacts. We shall be con-
cerned here with several distinct conduction mechanisms only and the discussion
-31-
of theories will be limited to three voltage ranges: low, intermediate and
high.The process is electrode limited for low and intermediate voltages and is
bulk limited in the higher voltage region. In practice, however, many more
co-operative phenomena might occur, because no clear cut criteria may "be assigned
to distinguish these three voltage regions. For greater details, reference73)should be made to a recent monograph on the subject.
a) Fowler-Nordheim type tunnelling
This process requires that the system be subjected to high voltage
biases so as to lower the barrier thickness at the negatively biased Fermi
level to less than 50 A . The well-known expression of Fovler-Hordheim type70)
of tunnelling is then given by
T 3.33 x 10 1 0 F 2 f 0.69 <t>3/2
where J is the current density, F is the applied field and $ is the mean
potential barrier height above the Fermi level of the cathode. This expression
is true only if the ratio of the insulator effective mass to the free-electron
mass (m ) is unity. Otherwise the exponent of equation would become
y \m f) ^ This process is electrode limited.
b) Bicliardson-Sihottky effect
The current flow through the insulator in this process is governed by
the rate at which the electrons are thermally excited over the interfacial
potential barrier into the insulator conduction band. Thus, the current
flowing in the system is determined by Richardson's thermionic equation,
= AT2 axpf- j|j , (5.2)
where A is the well-known Richardson constant ( — m | — 120 if J is
expressed in A cm~ ) and <j>n is the interfacial barrier height. However,
when the applied field interacts with the attendant electrode image potential,
it causes attenutation of the barrier height from <& and permits more current
to flow over the barrier with increasing applied voltage. This effect is7k)
known as the Schottky effect and the attenuation thus caused is given by
iiji * 8 F* . The current is then determined bys s
-32-
J =
vhich is the Richardson-S eh ottky equation.
The additional exponential term accounts for the attenuation of the
barrier height caused due to the interaction of the applied field and the image
f e3 )1/2force. 8 is a constant = y—-5 , e* being the high frequency
Q
dielectric constant of the material, For sufficiently ligii fields across
the interface, tunnelling may be the dominant conduction mechanism and it is
then governed by a suitably modified form of Eq.{5.l) in which the image force
has been included,
c) Space charge limited currents
In both of the processes discussed above, space charge effects were not
considered. However, the traps are bound to exist in most of the film systems
used and thus it is necessary to account for their contribution. First, the
situation will be discussed for a trap-free insulator. In the case of ohmic
contacts, a negative space charge region exists inside the surface of the
insulator and an equal and opposite (positive) surface charge on the
electrodes. When the field is applied, the space charge in the insulator
increases and its redistribution takes place depending upon the nature of
voltage bias. After this redistribution, if the positive charge exists on
the cathode, i.e. the cathode region acts as a charge reservoir to supply the
required charge to the anode, the process is called*space charge limited"(SCL).
For trap-free insulators, the expression derived by Mott and Gurney to
calculate the current density is given by
where d is the insulator thickness. However, in practice, much lower
currents have been observed and are found to be temperature sensitive,vhich
could not be explained with the above equation. Rose then pointed out
that when the insulator contains traps, a large fraction of the injected space
charge condenses into traps. Thus, it was suggested to multiply the right-hand
fclBide of Eq..(5.M'by a quantity 6 = N p exp[-Et/kT),where Nt is the trap
density. E is the depth of the trap "below the bottom of the conduction
bandrand Nc is the effective density of states in the conduction band. Thus
Eq..(5.J+) may be expressed as:-33-
L 8 <T I2
Since 6 is independent of V , the proportionality J ^ Y , as in
the case of trap-free insulator^still holds.
d) Poole-Frenkel effect
When the applied field is high,the coulambic potential "barrier of a77)donor centre or trap is lowered by an amount
which is twice the attenuation of the barrier height due to the attendant
image forces in the case of the Schottky effect. This occurs because the
potential energy of an electron in a coulombic field is four times that
due to the image force effects. fJpp is thus
• " • £ € <(5.7)
This, in turn, amounts to the lowering of the ionization potential, Eg, ofthen
the donors in a solid. The bulk current density is,( determined by the well-
known Foole-Frenkel equation
where J = ey N F exp(-Eg/2XT) and is the low field current density. How-
ever, in actual practice with thin film insulators* the traps having a
coulombic type of barrier would experience the PF effect at high fields and
thus result in the increase of escape probability of immobile electrons therein
The expression for insulators with shallow traps is then given by:
T . To o+[fljLfJ-] (5.9)
The reader is reminded that the coulombic "barrier exists not only
between a donor centre and its electron but also between a charged trap and
an electron. The charged trap means a trap which is charged when empty and73)neutral when full. Furthermore, Simmons has discussed another situation
-3k-
when the insulator contains "both donor centres (.lying "below the Fermi level)
and shallov neutral traps. The expression for the bulk I-V characteristic
of the film ia then given by
-J:•where
(5,11)
The suffixes d and ta here refer to the donor and shallow traps,
respectively. This equation holds even if there exists a range of trap and
donor levels and one of the trap levels is more dominant than the others.
Final remarks should be made that although the functional dependence
of the current density is the same for the Schottky effect (Eq.(5.3)) and the
Poole-Frenkel mechanism, a clear-cut distinction may be made by a plot of1/2
log J versus F /kT which is a straight line of slopes gQ or 8 _ ,b irF
respectively. Therefore,close scrutiny must be made in the interpretation
of experimental data to avoid any mistake arising due to this similar functional
form. Since the scope of this paper does not allow one to mention many other
co-operative conduction processes occurring in thin films, it would be advisable
to refer to the earlier cited Refs.69 and 73 again.
2. Conduction mechanism in Lanfimuir films
Of all the conduction mechanisms that are discussed above, the dominant
one in Langmuir films seems to be tunnelling. Obviously -t for measuring
tunnelling currents, one needs to separate two metal electrodes by a barrier
of 20 A - 100 A in thickness which must be uniform, homogeneous and an excellent
insulator. Thin oxide films grown on "base metal electrodes were employed for
such studies by Giaver and Nicol et al. because of their compactness
and uniformity. The fact that Langmuir films of several organic materials
may fulfill all these requirements for studying tunnelling was recognized byfli \ ftT \
Miles and McMahon , These workers employed a monolayer of bariuma
Btearate ( - 25 A) as an insulating barrier between tin-lead and tin-indium
evaporated film pairs. The experiments, however, did not prove useful
because ah irreproducible and strongly time-dependent resistance of the device
was observed in these studies. This is not surprising,because the monolayer
transfer technique vas in its infancy and only a handful of electrical
• . - 3 5 -
measurements were carried out "before that. It was also not known whether voids,
defects or inhomogeheities in these films could be reduced to such a limit
as to permit the construction of tunnel sandwiches. Hw«ver» these studiesfin)
of Miles and McMahon provoked interest Of many others in the conduction
behaviour of Langmuir films.
Handy and Scala subsequently tried to obtain new information on
the conduction behaviour of Langmuir multilayers and considerable efforts were
made by them to deposit compact films for their studies. In their extensive
studies on Ba or Ca ste&rate films {1-10 monolayers) with counter-evaporated
electrodes of Sn, Cu, Ag and Pb (and also using Hg as the counter-electrode in
a few cases), the conduction characteristics were found to "be highly non-linear
and strongly temperature dependent. Pig.10 shows the typical behaviour of a
symmetric Sn-3 layer calcium stearate-Sn sandwich ( ^ 1 mm area) at 0.1 Hz
and at room and liquid nitrogen temperatures. They further studied thethick
behaviour of a 3-layer and a 5-layer^film at constant voltages for two junctions
which is shown in Fig.11. Detailed analysis of temperature dependence of
conduction for a 3-layer sandwich led to believe that the electron tunnelling
is the operative conduction mechanism at low temperatures (< -100° C) and
small thickness, and thermionic emission at high temperatures {y hQ° C) and
larger thicknesses. A superposition of both of these mechanisms has, however,
been assumed by them at an intermediate temperature range. Handy and Scala
also determined the zero voltage thermal activation energy 6, as 0.25 < $> < 0.30 eVrb b
when the thermionic contribution to the current was plotted on a Schottky plotf I-I' 1of log • • •• — vs . 1/T as shown in Fig,12. In an attempt t o analyse t h e i rI T J
Qp") fi^}
results using theoretical expressions derived by Stratton and Simmons ,
they failed to obtain a satisfactory fit of the observed data. Consequently
no satisfactory explanation could be given for the observed temperature;
dependence. In fact, they have shown that the fatty acid films are promising
for electron tunnelling and low values of $. could be due to the defects and
impurities within the film. One of the important conclusions arrived at is
the necessity of a reactive and hence oxidized surface for the fornatioh and
retention of low porosity layers. In an attempt with the use of noble metals
(e.g. Ag or Au) as the first electrode, the Junction resistance vas found to
be less than 1 ohm and none of the gold based samples were successful for
electrical studies.8MIt appeared also to Horiuchi et al. that Langmuir films should
prove suitable as an insulating barrier between1metal electrodes,because of
their known/thicknesses. The inspiration came largely through the pioneering
and extensive studies made by Handy and Scala . who had shown promising
-36-
behaviour of such films for studying electron tunnelling phenomena. Calcium
stearate films sandwiched bet-ween asymmetric electrodes of tin and aluminium
were employed "by them to study the voltage and temperature dependence of the
current. As has "been shown ^n Fig.13a the observed current I, was found to
be strongly temperature dependent above 200° K and,nearly independent below
this temperature. This was the situation when the lover tin electrode, was
negatively biased. Conversely, when the upper aluminium electrode was
negatively biased, the current Ip was observed to be temperature independent in
the temperature range studied. This type of behaviour of MIM junctions
suggested that more than one mechanism of electron transfer is operative
{see Fig.13b). Then they analysed their V-I characteristics using the trape-
zoidal energy barrier model of Simmons a 3^» 8 5' and Pollack et al. '' T' and
thus determined barrier heights <£_ at the two interfaces; 6, =1.3 eV (for• D ' b
stearate-aluminium) and <fi m 0.8 eV (stearate-tin). This indicated that the
work function of tin is lower by 0,5 eV than aluminium, which is in contra-
diction with the known difference of 1.5 eV, This discrepancy, however, could
not be resolved even when they studied symmetric junctions of tin electrodes
and took the natural oxide layer into consideration.
From their detailed analysis it was shown that the temperature and
voltage dependence of the current satisfies the Schottky emission mechanism
whereas tunnelling has been predicted from the voltage dependence of the low
temperature current. Alternatively, the two mechanisms have been found to
be operative for negatively biased lower Sn electrode and upper Al electrode,
respectively. To resolve the difference of barrier height determined from
the two operative mechanisms* they have assumed the presence of some conduction
patches ir. the films which would let the tunnelling current pass through them
readily. This seems reasonable,as tunnelling emission was found to be more
strongly dependent on the barrier height than the Schottky emission. From
their studies of I-V characteristics,which were symmetrical with respect to the
polarity of applied voltage, the dominating conduction mechanism has been
suggeeted to be barrier limited because the characteristic was dependent on
electrode material. To avoid the formation of oxide layers, efforts were made
to study gold based devices,but measurements could not be carried out because
the samples prepared showed ohmic characteristics with very small resistance.
1+2) 81) 8MUnlike in the measurements of previous workers * * , , Mann and
IS) 881 •
co-workers succeeded in giving a quantitative proof of the tunnelling
theory in their measurements of d.c. conductance through Langmuir films of a few
fatty acid salts. The success of their measurements lies in the fact that33)film deposited by using a modified B-L technique did not possess even a
-37-
small fraction of holes and conducting imperfections. Analysis of the
experimental data wus made in accordance with the concept of quantum-mechanical
tunnelling proposed by Sommerfeld and Bethe
72)To determine the conductivity, Bethe's relation for small values
of applied voltages was used, which is expressed as:
) 1CTt
where ^ is the differenc "between the work function of the metal and the
electron affinity of the insulator. All other symbols have their usual
significance. This equation refers to equal metals and?if the metal electrodes
are different, ^ should be replaced by
respectivewhere ^ and jL are the <p values for the (metal electrodes. Since each
insulator is expected to have some impurity-dependent conductivity o*. in-
dependent of d , besides the tunnelling conductivity a , they have
analysed the results "by separating out these two contributions from the
experimental total conductivity a(- a. + a ) .
Mann and Kuhn investigated many samples in the thickness range
containing 5 to 21 layers with lower Al electrode "biased both positively and
negatively. The measurements of d.c. conductance at 20 C and -35° C , with
different upper electrodes, indicated that the logarithm of the current
density j is proportional to V and inversely proportional to d, as is
expected. Fig.lU is a typical plot of Nj versus V of cadmium arachidate
film,where H refers to the number of layers. The dotted line represents
the current due to impurities j. and the other curve for H = i gives the
total current j which is a sum of y and j. (tunnelling current). In
Fig.15, log Cf. Co* * a - a. ) is plotted against thickness d for four types
of cadmium salts of the fatty acids CH (ClO „ COOH with n = 18 - 21,
Evidently, the experimental points for both 20 and -35 C are on the same
straight line in accordance with Bethe's equation. This exponential decrease
of conductivity with increasing1thickness of the insulating barrier d
verifies the tunnelling theory. The vacuum work functions of Al and Hg
determined in their work theoretically were found in good agreement vith the
known values. '•
-38-
Further experiments were performed for measuring the photocurrent in
the sandwich assemblies between similar and dissimilar electrodes. From these
measurements, the vacuum work functions of Al and Hg were determined, in-
dependently .and an excellent agreement "between theoretical and experimental
values led to the "belief of the lack of larger imperfections in the monolayer
structures. Absence of even small fractions of holes and conducting im-
perfections in the devices which strongly influence the resistance have been
verified quantitatively. If is thus remarked that the earlier attempts on
similar measurements failed to yield satisfactory, results because of the high88)
sensitivity to imperfections in.;the organic films. These workers ' had
also verified the tunnelling theory in a previous work on photographic
sensitization. Fatty acid layers were used as "distance keepers" in the
sensitizers to distinguish between energy transfer mechanism and electron
injection mechanism. The dye was still found to be sensitized which led to
the conclusion that the sensitization is due to the energy transfer mechanism
since the electron tunnelling through fatty acid monolayers was extremely89) :
improbable .
A strong evidence for tunnelling at low voltage has been presented by
•Leger et al . • who studied electronic transport through organic films of
cyanine with two grafted stearic chains (St2 Cy). A typical I-V characteristic
at 1.5° K is shown in Fig.l6 for a sample Al-cyanine-Pb for different voltage
ranges. The curve(B)is quadratic for voltages less than 0.1 V and then exponential.
Such behaviour is clearly non-compatible vith a conductance dominated by pin-
holes in the,organic film and tunnelling through the natural oxide layer. In
another curve (A), the energy gap of superconducting lead i^Q) i s clearly
visible. Another striking feature of the characteristic is i ts non-linear
behaviour for voltages >50 mV which,together with the above observations,
led to conclude that tunnelling is the dominant conduction mechanism at
low voltages. However, contradictory results have been obtained in a similar19)work on Langmuir films of stearic acid containing 5 layers . Fig.IT shows
a plot of log I versus V1'2 at room temperature which fits veil to an ex-
pression of the form I °c exp 0 V1^2 with. 8. <• 1.8 x.105 (Vm)1'2 .. Predicted
theoretical valueB of B are ^ = M * 1° (Vm) for Poole-Frenkel con-
duction and .&_ = 2.3 x 10~7 ;(Vm) ' for Schottky conduction. From such a
S !comparison, i t was concluded that the latter is the dominant mechanism instearic acid.films. Dominance of Schottky conduction mechanism in Langmuir
•• ' 90)filmB in the; range of 1-10 monolayers has also been proposed by Snvastava
based on experimental studies of thickness dependence of breakdown field
(described later in detail)*
: . .. -39- ' :
All the conduction studies, described above, have been carried out in
the low thickness range of Langmuir films tl-10 inonolayers) and different
conduction procesaea have been found operative under different conditions.
However, it appears from careful examination that the existing data predict
tunnelling mechanism more decisively even though there are indications of .
Schottky conduction. In fact, tunnelling has been found to be the dominant
conduction mechanism in many other types of film systems in the comparable
thickness range and tunnel diodes have been developed. On the other hand,73) ;
it has been discussed that Schottky "barriers in thin films are much more
easily investigated using the a.c. technique. Nevertheless, to make sure that
tunnelling is the operative conduction mechanism in Langmuir films, more
experiments are required.
Since, tunnelling currents are known to have a-slight temperature
dependence and the Schottky currents depend strongly on temperature,•such
temperature-dependent studies may prove useful for the choice of the conduction
mechanism in these films. Unfortunately, most of the data have "been inter-
preted without considering the nature of contacts vhich have a lot of influence
on the underlying conduction process. it is essential to examine the
nature of contacts in the future studies on these films. The presence of a
space charge region in KIM structures has also not been encountered in any
of the measurements up to now, and this is perhaps the reason that some workers
have indicated for more than one conduction process to be operative but failed
to determine what exactly it is. Due to the lack of any judicious study which
accounts veil for all the factors influencing the conduction process and in
the light of the existing contradictory experimental evidence, it is premature
to conclude anything about the conduction process in Langmuir films. In
general, renewed investigations on conduction mechanisms under different
conditions are desirable to resolve the discrepancy and to solve the complexity
arising because of the earlier results.
3. High field conduction characteristics '
Multimoleciilar films of stearic acid were chosen for high field con-
duction studies because of their ordered structure and because such films had
been used by many workers in the study of tunnelling phenomena. The interest
of Nathoo and Jonscher was to study the behaviour of these films under
high applied (a.c. and d.c.) fields and to get more information about their
dielectric properties. The sandwich structure of the type Al-film-Al was
used and the measurements were made by placing the device in vacuum as the
exposure to air-caused reversible increase of d.c. conductivity "by an order
of magnitude. The thickness range 6-UO monolayers was chosen which they
thought beyond the range in which tunnelling would be the dominant conduction
mechanism.
The response of the metal film device to a step voltage WAS characterized
"by an initial surge followed by gradual decay which lasted for hours and some-
times for several days in thicker films. At room temperature the current (i)0 7was time dependent following the relation I oC t . On removal of a steady
bias and shorting of the sample, a corresponding discharge current was observed.
Fig.18 shows a typical recorder tracing of the decay current after switching
on a steady field for 8 layer thick film vhich follows I u t n "behaviour with
t = -0.7 obtained by the slope of log I vs.log t . The d.c. characteristic
shown here was obtained by subtracting the discharge current from the charging
current. Further, the current was slightly non-linear at fields < 10^ V cm*"1
varying according to I &- E . At higher fields in excess of 10 V cm~ ,
the characteristic, however, was found to obey Foole's law of the form
93)where 2S is the distance between donor-lilte Poole centres , i . e . neutral
when occupied by the charge in question. To determine the value of 2S, Nathoo
and Jonscher Obtained I-V characteristics for a film containing 11 layers at
four different temperatures and found that each of the characteristics obeys
Poole's law. At temperatures between 198OK - 295°K, the values of 2S varied
between 33 A andUOA . It is expected that the donor-like centres should be
separated by an integral multiple of the molecular length of the fatty acid91)
film under investigation. In the present case, they expected that the donor-
like centres are very likely to fall at the more reactive COOH end group of
the molecules which separate them by twice the molecular length (^50 K) .
Since the observed values of 2S were found to lie between one:and two
Massumingan effective carrier9k)temperature T in excess of the ambient lattice temperature . Subsequently,
these authors measured "low field" a.c. conductance c(w) at peak amplitudes
< 10 V cm and observed a relatively weak temperature dependence of the
absolute magnitude of cr{w) . ; Fig.19 is a plot of o(w) as a function of
the frequency for a stearic acid film (ll layers thick) at different temperatures.
Evidently* all the characteristics may be represented by the power law
dependence on frequency df the, form
"a(u) oc w n , . '.' (5.15) .
-in-
where n is temperature dependent and yaries "between 0,5<n<l at high and
lay temperatures, respectively. Similar behaviour of a(w) was observed in
many other hopping systems, but these were mostly disordered or amorphous.
From the fact that organic films, which are known to be largely ordered, have
also shown a similar "behaviour, they concluded that the frequency dependence
of a is a general feature of hopping systems and not restricted, to disordered
solids. Regarding the nature of the hopping charges, their suggestion is
that theEe could be electrons'or protons». much less likely heavier ions. Later,95)Jonscher has justified the observed behaviour of <r ir these ordered
systems. He explained that even a perfectly ordered system in which all
"hops" had equal probability distribution, the actual delay times would vary
over a certain rangefwhieh could be considerable,to give rise to the observed
dependence.
Further measurements vere made "by these workers in the "high
field" regime extending up to 2 x 10 V cm in a wide range of frequencies.
In particular, their interest was to get information on whether or not the time-
and frequency-dependence of the response was a function of the amplitude of
the electric field. All the measurements were made on a 5-layer thick stearic
acid film sandwiched between aluminium electrodes. Fig.20 shows, the frequency
domain measurements on two areas of the same film for a wide range of field
amplitudes. Curve (a) shows that the results are different only in the lower
frequency region (<0.1 Ho) which is due to the increasing level of d.c.
conduction and is strongly non-linear for fields over 10 V cm , Curve (b)
represents a similar characteristic taken at 1 x 10 V cm when the d.c.
level was subtracted from the total current. On extrapolation of the high
frequency line, the exponent of Eq.(5-15) n ^ 0.8 was found. Obviously,
the entire characteristic at a field of 2 x 10 V cm shifts upwards by a
factor of about 2.5 in the frequency region studied but its slope does not
change much. Thus it has been suggested that the onset of the frequency
dependence of the characteristic does not change significantly because of the
onset of a.c. non-linearity.
Time domain measurements were made on a different area of the same
film and the plots of charge currents tafter subtracting the steady (d.c.)
current) and discharge currents versus time are shown in Fig.21. The data
fits well to give a dependence of the type I ^ t n which,in the linear
regime,transforms to give the conductivity dependence of the type given by
Eq.(5.15). The slope of discharge currents increases from 0.71* at the "low
fields" (2.25 x 105 V cm"1) to 0.87 at the "high fields" (1.8 x 106 V cm"1).
The discrepancy between the field-dependent slopes in the time domain and in
-1*2-
the frequency domain is attributed more to the inapplicability of the Fouriertheoretical QT)
transformation in the non-linear regime. Even the^suggestion that the
effect of a high field on the step function response be represented to a first
approximation, by a change of the amplitude of response vithout a change of
time dependence could not "be strictly applicable in the present measurements.
However * it has been taken to explain the results as a reasonable first approx-
imation which then permits Fourier transformation in the non-linear regime.
Validity of such an approximation is further supported "by the fact that a small
but definite increase of capacitance with increasing field has "been observed
in both the time and frequency domain measurements. Such behaviour was also98) 97)
found in the measurements on silicon oxide and is also expected theoretically
The whole range of data has been considered as evidence for a small
but definite effect of high fields on thetime dependence of step: function1 response.which is more pronounced in the time domain than in the frequency
domain. Such a definite trend could not be determined in the frequency981
domain measurements on silicon oxide . Although the results have been9l) 96)
interpreted well qualitatively by the authors . , no commitment has
been made . regarding... the transient behaviour of such films under high
applied fields (d.c.) or about their a.c. conductivity. Nevertheless, they
have opened up a new series of measurements and further measurements on these
guide lines would yield many more facts about the high field conductionpoint
characteristics of Langmuir films. The most interestingj^would be to know
the nature of the hopping charges by studying the field assisted hopping
mechanism between closely spaced donor-like centres in several other organic
compounds*
VI. ELECTRICAL BREAKDOWN BEHAVIOUR
The problem of dielectric breakdown in solids has been the subject of
numerous investigations for a long time. Nevertheless, the basic mechanism
of dielectric breakdown has yet to be understood in all details. One of
the crucial points in the breakdown process is to establish whether the "break-
down results from a strictly local eventor,rather,is a consequence of
phenomena propagating with an increase of intensity. Thickness dependent
studies have, however, implied that a mechanism of the latter type, presumably
avalanche multiplication of charge carriers, is the prime agent of "breakdown
initiation. But the concept could not gain universal acceptance because
-U3-
some experimenters * attributed breakdown to thermal effects, i.e. the
current density, even without an avalanche multiplication, generated more heat
into the dielectric than the dielectric itself had dissipated. Thus» the
whole breakdown phenomenon seems to be based on two principal theories, via.,
electron avalanche mechanism and thermal effects only. The former,which
interpret the breakdown behaviour of Langmuir films,will only be discussed
briefly here. For details of these theories (both electronic and thermal)
as well as for wider knowledge about the experimental data on other types of
film systems reference should be made to reviews,, books and other papers
cited in the recent bibliographical survey on the subject .
On the basis of the nature of the initiating step and also of the
continuation process of breakdown in thin insulating films, Klein has
characterized the breakdown events into three categories:a) thermal; b) electronic;
and c) electronic, modified thermal. In the case of thermal event, the Joule
heat causes a significant temperature rise resulting in an exponential rise
in the electrical conductivity of the specimen. Instability arises in the
field range above 10 V cm when the temperature rise may be of the order of
a few tens to a few hundred deg. C. Such events have been observed in the3 7 - 1
range 10 - 10' V cm and might occur beyond this range. Electronic eventsh -1
occur at low fields (1-10 V cm ) and the temperature rise is unnoticeable,
in contrast to thermal events. The event is initiated by an electronic process
such as impact ionization, double injection or such like. Such events have
been found to occur in semi-insulating films and not in insulators. The last,i.e.
electronic modified thermal event, is that which has been found to occur
in a wide variety of insulating films. The event is necessarily initiated by
an electronic process such as avalanching at fields typically larger than
10 V cm (range extending from 10 V cm to 10 V cm ). In contrast to the
pure electronic event, the temperature rise may be sufficient to cause local
instability by a thermal process and current runaway occurs due either to the
temperature rise or to field distortion. The electrostatic energy dis-
charges at the instability point, leading to irreversible changes by melting
and evaporation. The electronic breakdown involves only a small volume of
the specimen,in contrast to thermal instability involving the whole uniform
specimen.
To make sure which of the breakdown events is taking place in thin
films, one must know the basic underlying mechanism. The breakdown in films
is particularly sensitive to the conditions, such as, specimen thickness,
nature of electrodes, ambient medium, temperature and the type of voltages,
etc, Nevertheless, thicltness dependent studies of breakdown phenomena carried
-kk-
out in the past few decades, led to a "better understanding of the complicated
"breakdown mechanism. Most of the experimental data could be satisfactorily
interpreted in terms of the breakdown theories assuming the electron avalanche
as the basic mechanism.
1. Electronic breakdown theories
In fact, the earliest concept of electron avalanches produced by impact
ionization was predicted by von Hippel J and FrShlieh ^ almost at the
same time in the thirties to explain the intrinsic nature of breakdown in
insulators. However, the whole concept could not be acceptable as such because
their theories failed to explain the observed data in thin films. In sub-
sequent development of theories, many refinements and new suggestions were
incorporated by different workers but none of them could be widely acceptable.
It was so until 196^, when Forlani and Minnaja made a successful break-
through and proposed a theory of electrical breakdown in thin dielectric films.
This theory could explain a wide range of existing data on thickness dependent
breakdown studies and also explains the data on Langmuir films. For the reason
that earlier theories have already been reviewed well in many of the publications
(see Ref.1) we shall limit our description to the recent developments in
the breakdown theories made after
a) Torlani-Minna.1a's (F-M) theory
The fact not considered in the earlier proposed models was the origin
of the charge carriers in an insulator; F-M in their theory assumed that these
are injected from the cathode into the insulator by field emission and are then
multiplied by impact ionization. Thus,considering the consequences of avalancheTO)
multiplication of Fowler-Hordheim emission , the injected current density
J j-i- c a n ^ e writtencatn
= To «+.[-' (6.1)
where
is the effective height of the potential barrier at the cathode-dielectric
interface. Other notations have the usual meaning. The validity of Eq.(6.l)
has been discussed at least in the range of applied fields for which break-
down is expected. Neglecting the effect of lattice vibrations in the dielectrics
in slowing down the electronE in a strong field, the collision-ionization per
unit length (d,) is approximately (cf.Ref.69, p.22*0
a - eP/I , (6.2)
vhere I is the difference between the mean energy of an electron when it is
able to ionize and mean energy of electrons emerging from an ionization event.
Further, considering the film thickness much lower than the recombination length,
the current arriving at the anode as a result of the collision-ionisation multi-
plication will "be given by
(6.3)
If it is assumed that the threshold current density above which melting or
evaporation of the dielectric material occurs is of the same order as Jn ,
the following expression for the critical field strength is obtained;
(6.1;)
theThis relation evidently shows the dependence of|breakdown field on
specimen thicltness and has "been shown to explain the experimental data for
various inorganic dielectrics by Budenstein and co-vorkers , and for
Langrauir films by Agarval and Srivastava , It has "been discussed
further by F-M that if the electron injection takes place at the negative
electrode and the effective height of the potential barrier is very low, the-1/1*thickness dependence of F is expressed as F_ 0^ d . They have thus
concluded that the real thickness dependence of F^ lies "between these two
idealized situations.
At a later stage, F-M have discussed another situation,in which the
electron injection is governed by Schottky mechanism rather than by the
tunnel effect. This is expected when the applied field is very large because
of the predominant role of the electron image force on the shape of the potential
barrier. Therefore, Eg., {6.1} should be replaced by an equation similar to
(5.3 in Sec.V )» giving the injected current density of the cathode. Following
the same arguments adopted in their earlier approach and by means of proper
evaluation of order of magnitudes, F-M obtained the following expression for F_:
- 0*tf- /JL.-L
Till a equation practically means the independent nature of the breakdown voltage
with respect to dielectric thickness. It is valid even if the voltage drops
in a non-linear way through the dielectric layer. This equation has "been
shown to interpret thickness-dependent data on Langmuir films in the lower
thickness range .
It must be pointed out that Eq.(6,1*) does not contain any temperature
term,because the basic conduction mechanism is taken to be the tunnel effect.
However, the dependence of the type F^ cc d" ' , based on the electron-
phonon interaction, can be shown to depend upon temperature. An increase
of dielectric strength with increasing temperature would be expected, consistently
with other theories based on electron-phonon scattering a02)>103),112} ^ A
reverse dependence is obtained if electron-electron scattering is properly taken
into account, '
b) O'Dwyer's theory
In spite of the fact that the F-M theory ' is based on many simplifying
assumptions, the important omission in it is that of space charges which are bound
to arise because the positive charges left behind by the ionizing electronsllM
were assumed to be immobile relative to the latter. O'Dwyer then proposed
a simple theory assuming that the mobility of conduction electrons is much
greater than the mobility of the holes. For considering the space charge
effects, Poisson's equation is added to the current continuity, and to the
heat transport equation (when necessary). To make life easy* it was further
assumed that the collision ionizations are achieved only by those electrons
which are accelerated from a low energy to the ionizing energy without suffering
any collision with the lattice during that period. With these assumptions,
the functional dependence of field strength with the dielectric thickness could
be expressed in terms of certain dimehsionless parameters, although no explicit
form was given for the function itself.
The fact, however, is that the data on thickness dependence of break-
down strength could not be explained by fitting it in terms of the dimensionless
parameters, i.e. dimensionless mean field strength and dimensionless dielectric
thickness. When F-M r L 1 ' pointed out the above difficulty* O'Dwyer 69),115)
made further progress and proposed a more realistic model. Now he assumed that
highly mobile electrons injected into the insulator at the cathode by field
or thermionic emission produce avalanching and that the current is totally
electronic. Again he obtained the set of equations and solved for the
steady state. The highest field, below vhich the instability and current
runaway might arise* was taken to be the d.c. breakdown field.
-1*7-
Whereas O'Dwyer's modified theory could explain data of various
experimental investigations on alkali halides, it did not prove to be sothe
successful to explain^ data quantitatively on thin films. For example,
it failed to interpret the destruction of the film at higher current densities
and lover fields because the model is based on the current instability criterion
rather than the condition for destruction. It has been pointed out by
Klein that the earlier theories as veil as the above theory assume uniform
charge injection and avalanching in the vhole specimen, whereas, in practice,
the situation is different. He has shown quantitatively that the continuum
theories, as that by O'Dwyer, do not explain the breakdown event adequately
if it is a consequence of localized-avalanches which are spaced widely from
each other.
c) Klein's localized electronic breakdown theory
The concept of localized breakdown was given in earlier theories of
Frohlich and Seitz but their theories based on single electron
avalanche do not include the effect of space charge and the occurrence of117)instability prior to destruction. Following the suggestion by Watson et al.
that a rapid succession of avalanches may produce the breakdown, Klein
has developed a model to show how a succession of avalanches can produce break-
down in thin films. This model, in fact, explains the dependence of break-
down phenomena on almost all the parameters and is supported by a wide
variety of experimental data. However, only essential points will be
discussed here, which are relevant in the context of the present paper to
explain the data on Langmuir films.
According to the proposed statistical model assuming localized
electronic breakdown, the breakdown is initiated by electron injection into
the conduction band of the insulator and develops in a sequence of stages.
The injected electron (by field emission or thermionic emission) at the cathode
produces an avalanche of free electrons by impact ionization and the positive
charges are left behind in the insulator. The latter, being nearly immobile,
drift slowly to the cathode and form a positive charge cluster which results
in the enhancement of the cathode field. The local electron injection rate
is thus increased and a finite probability reached for an electron to be
injected during the transit of the charge cluster through the insulator.
Consequently, there is further avalanching for the continuation of the break-
down process. Since the local cathode field during the transit of the charge
cluster increases with avalanche length, the formation of large avalanches
also depends upon the film thickness, in addition to many other parameters
involved.
T"-
In the present model, the ionization coefficient Ct has been assumed118)
to obey the relation a = F/V. which was proposed by Shockley for "high
fields" applicable to semiconductors; F is the applied field and V. is
the ionization voltage. This was found suitable in view of the fact that the
observed breakdown fields extended over a wide range in most insulators. How-
ever, it has been shown that the proposed theory also inteprets the breakdown
events at the "low fields" obeying the relation-proposed in O'Dwyer's model
for Ci . The mean avalanche length a has been shown to depend upon c£
by the following relation;
U v - " - p- > {6,6)
where K > 1 is a factor taking account of the average increase of the mean
free paths for ionization towards the anode,and i is the number of collisions
of the injected electron while travelling through the insulator. It has been
shown that the effective field acting on the electrons in the film is so low
beyond a that no more impact ionization occurs when a is not much
smaller than the insulator thickness d
How, if the film is too thin (d •£ a }, the full-size avalanches cannot
develop and the probability of "breakdown becomes very small. Therefore, for
the incidence of breakdown,even at relatively low rates, it is necessary to
raise the applied field to an extent that a £ d or from Eq. (6.6)
F >
Evidently, this shows that the breakdown field {at which breakdown
begins to occur) in very thin insulators is inversely proportional to the
thickness d . Further, it has been shown for d > a^ , that the rate of
decrease in the breakdown field with increasing d becomes smaller for all
values of mean time to instability. The observed experimental .data on Langmuir
films (discussed later In this section) support the predictions,of this theory.
For the discussion of temperature effects on the breakdown field, It
has been shown that it varies with the nature of the injection current process,
the hopping mobility and the ionization coefficient values. To predict the
temperature effects on electronic breakdown events, one must determine how
the above parameters change. In consequence, some of the temperature effects
promote and others oppose the development of "breakdown. Also experimentally
both types of behaviour of the breakdown field (increasing and decreasing with
temperature) have been observed in various thin film insulators.
This theory of localized electronic breakdown in insulating films
has been shown to• interpret the observed data of various types of film systems
over a wide rango including almost all the parameters, such as, temperature,
thickness, electrode materials space charge, etc., which are Itnown to affect
the breakdown phenomena. In addition, one of the most successful parts of
this model is that it also explains the completion of the "breakdown event
including the destruction of the insulating film. It has been discussed that
the breakdown follows in a sequence of stages, i.e. in the initiating staget
harmless avalanches occur in the whole specimen and the temperature rises
during the avalanching of electrons- Consequently, a significant enhancement
of the electrical conductivity takes place which causes voltage collapse
through a current runaway. Impact ionization may stop during the latter stage
but the "breakdown events continue until the destruction occurs because of the
thermally unstable state, due to temperature rise at the site. Thus, a complete
breakdown event mainly comprises the initiation of breakdown > instability due
to heating and,finally,the destruction of the capacitor with voltage collapse.
In the author's opinion, Klein!s proposed model is the most successful approach
among all the models discussed by various workers. Nevertheless, it does not
take into account explicitly the presence of defects in the insulating films
and future works might improve the theory further,
2. Experimental studies on Langmuir films
Langmuir films of fatty acid salts are known to have high breakdown
strength s 10 V cm for a long time. The earliest studies of the break-
down of the films under high d.c. voltages were carried out by Porter and
Wyman . The experimental arrangement employed for such measurements was
very simple,consisting of a galvanometer to read the current. Using a Hg
drop as the upper electrode, no current was detected until the voltage
applied across the specimen reached a certain critical value in the case of
30 layer thick filjns vhereas,with thinner films, appreciable currents were
present almost from the start. The breakdown per layer was found to rise
sharply at a thickness of about 20 layers and its order of magnitude was
£ 10 V cm" . Very high specific resistance 10 ohms, below the
breakdown voltages^was found. Later, Hg droplets were replaced by a drop
of tap water and the device behaved almost the same way as with Hg, However,
there was one striking difference,that the critical voltage at which there
vas a sudden rise of current was independent (nearly) of film thickness. It
was higher for thin films and lover for thick films than the corresponding ones
when mercury was used. No striking difference was observed in the breakdown
-50-
voltage of X and Y films and some difference in I-V characteristics is attributed
simply to electrical asymmetry in the film,probably associated with the initial
layer. Further, the film was disrupted permanently when the "breakdown voltage
vas exceeded and mercury drops vere used. On the other hand,with water drops
there was no change in the nature of I-V curves even after the film had been
subjected to a voltage beyond the breakdown point. The difference has been
ascribed to the different affinities of the two liquids for the surface and
presumably the water moleciiLes were able to penetrate the film.
Similar attempts were made by Race and Reynolds in the early period
and data was collected on films of cadmium arachidate obtained from solutions of
varying pH values in the range 5.1 to 6.62. As is known, the proportion of
conversion to soap depends upon the pH of the solution, the highest dielectric
strength was found only at the higher pHs, which correspond to a higher
proportion of converted soap. Experiments on skeletonized films showed
lowering of dielectric strength, as expected. Further measurements on di-
electric strength as a function of frequency showed a wide scatter in the
observed data and thus no conclusion could be drawn. However, at 1000 Hz,
the highest valueE vere obtained. They had admitted that,in spite of their best
efforts, the inclusion of dust particles in the multilayer films could not be
checked and,therefore, the data, in general^were disappointing. Nevertheless,
they succeeded to deposit 201 layer thick films that could withstand from 150
to 190 volts corresponding to the dielectric strength of 2 x 10 V cm" ,
Because of the difficulty to obtain large areas of the films free from cracks,
holes or dust particles, the possibility of using such films as practical
capacitor dielectrics was ruled out in their work.
Evidently, these earlier measurements do not predict anything about
the breakdown mechanism and no efforts vere made by these workers to interpret
their data in terms of any known theory at that time. Some more attempts also
yield only the orders of magnitude of dielectric strength of Langmuir films.
For instance, Thiessen et al. vere able to apply electric fields up to
5 x 10 V cm on barium and calcium stearate film capacitors in the thicknesso e 29)
range 100 A - 1000 A . Similarly, Holt , while considering the use of
Langmuir films as thin film dielectrics, reported the dielectric strength
— 10 V cm for varying thickness range of barium stearate films. It was
not until the last few years that Agarwal and Srivastava ' *
have made detailed and systematic investigations on thickness-dependent behaviour
of breakdown strength (d.c. and a.c.)011 various fatty acid soap films.
Such studies were undertaken to illustrate the underlying breakdown mechanism
as well as to obtain information about the breakdown strength over a vide
-51-
range of film, thicknesses, which is of fundamental importance for the develop-
ment of devices. Similar measurements on temperature dependence behaviour123)
were also undertaken in the same school and some preliminary measurements on
a.c. breakdown strength of barium stearate films were carried out. In the
following section, all experimental investigations have been reviewed. '
a) Thickness-dependent studies
In studies of Agarwal and Srivastava 1 0 9 «110'» 121),122) t h g c a p a c i t o r
geometry of Al-film-Al has been Used and the thickness range covered is 1-80
layer, corresponding to 25 A - 2000 1 . In their earliest investigations only
on barium stearate films Hg drops were used as the upper electrode.
Three principal breakdown voltage thresholds ,vi2. onset breakdown, destructive
breakdown and maximum breakdown voltages have been encountered. These have
been distinguished by the initial abrupt rise in the current, by the commencement
of "visitle" destruction in the film and "by the destruction of the film capacitor
over large areas, respectively, when the applied d.c. electrical field was in-
creased linearly across the capacitor. For the sake of simplicity, we shall
describe here the data in two parts, one consisting of onset breakdown
voltages when there is no destruction of the film and the other consisting of both
the events of destructive type.
i) Onset breakdown voltage
Measurements of d.c. onset breakdown voltage are carried out * * '
on mono- and multilayer Langmuir films of barium paLaitate, margarate, stearate
and behenate. Fig.22 shows a log-log plot of breakdown field F, versus
film thickness d on barium stearate films. The dielectric strength is alwaysCL
found to be a power-dependent function of the film thickness d , i.e. Fn oc a
with ct varying in the different thickness ranges. The curve (A) in Fig.22
is a plot corresponding to 1-10 layers and the curve (B) corresponds to the
higher thickness range of 10-80 layers. Evidently a - 0.01 in the lower thickness
range (25-250 A) and & = 0.59 (approximately) in the higher range (250 A -
2000 A) . Similar values of a = 1.08, 1.08 and 1.02 (for 1-10 layers) and
<L = 0.52, 0.52 and 0.53 (for 10-80 layers) were obtained for barium palmitate,
margarate and behenate films, respectively. Agarwal and Srivastava y '
interpreted their results in the higher thickness range in terms of F-M theory
of breakdown for small thicknesses which predicted 0, to be 0.5 for the high
energy gap dielectrics and for not very large electron affinity (Eq,. {6.h)),However, for the results in the lower thickness range, it was proposed that these
may be explained in terms of the increased boundary scattering.
-52-
90)In a recent communication , Srivastava has explained the above results
in the lower thickness range in terms of the F-M theory which considers a
transition from tunnelling to the Sehottky effect "because of the predominant
role of the electron image force. Evidently, Eq..(6.5) thus explained these
results showing the independent nature of "breakdown voltage with respect to90)
film thickness. Based on this interpretation, Srivastava has proposed
that the breakdown data in the lower thickness range of Langmuir films constitute
evidence of Schottky dominated dielectric "breakdown in these films. It is also
pointed out that the breakdown field for 10 monolayers (thickness ^ 250 A)
may be regarded to be near the "transition field" on these fllmSjWhich is not
possible to calculate theoretically mainly because of the unknown band structural
parameters of the insulating films. The author , however, has discussed
recently that it is premature to agree fully with the evidence of Schottky90)
dominating mechanism in Langmuir films . It is because most of the data on
conduction mechanism in these films have shown that the electron tunnelling is
the operative conduction mechanism in the lower thickness range- Further,
Eq.{6.5)» also shows an explicit dependence of the breakdown field upon the
temperature. Unfortunately, no experimental work seems to have "been reported
showing the temperature effects in the breakdown field in the lower thickness
range and thus supporting the contention of Schottky-dominated breakdown in12U)
these films. Even, the work by Agarwal and Srivastava on barium stearate
films (thickness ^ 515 A) in the temperature range -^O C - Uo C neither
corresponds to the thickness range under consideration nor conforms to the F-M
theory based on electron-phonon interactions.
On the other hand, it has been shown that- Klein's statistical model
is much liiore realistic for the interpretation of the present thickness dependentthickness
data in the lower range. Obviously, Eq.{6.7) explains the observed^independent
nature of the breakdown field at which breakdown begins to occur. The
relevance of this theory in the present context is supported by the fact that
the whole breakdown phenomena from its inception till completion has been in-127)
terpreted by Klein's theory in a recent communication . One of the
limitations of the F-M theory is that it does not interpret the occurrence of
destruction of the film due to thermal instabilities developed at high applied
fields.
ii) Destructive breakdown
Destructive breakdown in Langmuir films consists of two events, i.e,
the destructive breakdown voltage and the maximum breakdown voltage. These
events have been determined in various fatty acid soaps in the thickness range
-53-
o o
temperature in the range -1+0 C to +40 C . The capacitor geometry used
was Al-film-Al and the two breakdown events were distinguished on the same
lines as discussed in the thickness-dependent studies. Curve A in Fig.26
shows the plot of temperature versus breakdown field F for a barium stearate
film (thickness 515A). Evidently, F decreases slowly with increasing
temperature which is similar to that observed by Budenstein and co-workers
in some evaporated thin film systems. Theoretically such behaviour Is
expected provided that the electron-electron scattering is properly taken into
account. However, these workers failed to interpret the results in terms102) 103) 112)
of any electronic breakdown theory based on electron-phonon scattering * *
or in terms of recent F-M theory, which showtan increase of dielectric strength
by increasing the temperature (see Eq.(6.5)).125)
Further, these workers obtained data on the temperature dependence of
the destructive breakdown field F__ jn barium stearate film capacitors andb max
found that the decrease of F, with increasing temperature is more rapidb max o r
than that of F . Curve B in Fig.26 is a typical plot of F with respect
to the temperature of a barium stearate film (515 A thick). These results
were also not explained by them because no theory is yet available to account
for the temperature effects on destructive breakdown field. In fact, the
temperature behaviour of electronic breakdown events is not yet well understood
both theoretically and experimentally.' The recent Klein model explains
both the possibilities that breakdown strength can either increase or decrease
with increasing temperature, depending upon the nature of electron injection
which effects the mean time to instability and consequently the breakdown1210,125)
strength. J-V characteristics plotted in the work of Agarwal and Srivastavathe
and^transmission photomicrograph of the capacitor at destructive breakdown
voltages do not exhibit any more information than that obtained in thickness-
dependent studies. The breakdown voltage has, however, been found to decrease
with increasing capacitor area in agreement with the observations of107}
Budenstein et al.
c) a.c. breakdown studies^ ' 123)
In these preliminary measurements of a.c. breakdown , the onset
breakdown strength has been studied as a function of thickness and frequency
in the thickness range 100 A - 1500 A and in the frequency region 10-200 KHz .
Barium stearate films were always sandwiched between two aluminium electrodes
and the occurrence of breakdown was inferred from an abrupt rise in the
current. Fig.27 shows the behaviour of a.c. onset breakdown voltage V^
with film thickness d at a fixed frequency of 30 KHz . Evidently, V ^
-56-
increases with increasing film thickness. When the corresponding breakdown
strength 7 . was plotted as. a function of thickness d on a log-log scale>
it was founa that F c*r a with & = 0.68. Incidentally, this is the same
behavioiir as these workers * had already obtained in their corresponding d.c
studies. Further, a graphical representation of V versus frequency t for
a film of thickness 500 A (Fig.28) shews that V increases with increasing129)frequency. Similar studies were carried out by Klein and Levanon on
SiQ films and three types of a*c. breakdown processes were distinguished as
were found in their d.c. investigations . A theory was also proposed by129)
Klein and Levanon which was based on the thermal nature of breakdown and
predicted a decrease of breakdown strength with increasing frequency for the
high-frequency range studied. In view of the observed contradictory123)
experimental data in the Langmuir films, Agarwal and Srivastava deduced
that the breakdown is non-thermal in nature. Fortunately, the electronic
nature of the first breakdown event has already been established by these
workers through their d.c. investigations,
The experimental data could not be interpreted because of the lack of
an a.c. electronic breakdown theory. It has, however, been pointed out that
in spite of the high structural perfection of these films, the breakdown is,
presumably, initiated as "single hole" breakdown. The scatter in the
experimental data has "been attributed to the inevitable voids and inhomogeneities
in the deposited film as well as to the effects of electrode thickness whicht
of course, the authors tried to control in their measurements. The onset a-c.thickness,
breakdown strength of 'barium stearate films, in order of increasing film / -jraafound to lie between 1.33 x 10 and 0.22 x 10 V cm" , which is smaller than
6 6 - 1the corresponding d.c. breakdown strength (1.75 x 10 - 0.27 x 10 V cm" ).
13This observation gets support from the observations of Race and Reynolds
who obtained higher values of d.c. breakdown strength than the a.c. values.
The situation regarding breakdown phenomena in Langmuir films can best
be summed up by saying that it is not yet well understood. Although thickness
dependent studies could be interpreted adequately in the light of existing
theories, these are not enough to make sure what exactly is the underlying
conduction mechanism. On the other hand, temperature dependent data is
just in its primitive stage and could not be interpreted. Many factors
may be pointed out in these studies which,if taken into consideration,
would lead to a better understanding of the whole phenomenon. To list a few;
the breakdown rate has not been investigated and the I-V characteristics are
recorded only up to the stage of destruction. Even the I-V characteristics
do not give sufficient information about the pre-breakdovn conduction. The
-57-
influence of oxide layer formed between the base aluminium electrode and the
film has also not been discussed.
Whereas these factors must be considered in future works, more attention
must also "be paid to the determination of breakdown strength as a function of
electrode material, polarity of applied voltage, ambient conditions and nature
of applied voltage wave form, etc. A few groups, e.g. Klein and co-workers,
Budenstein et al. and Qsburn's group have been actively engaged in the study
of breakdown behaviour of varied types of dielectric films,and each of their
papers "brings new information. The reader is therefore advised to refer to
their original papers and follow all the guidelines in the studies on Langmuir
films. The recent bibliographical survey on breakdown conduction in thin
films is a comprehensive and up-to-date collection of such literature.
One of the most significant contributions of Ag&rwal and Srivastava,
who have carried out experiments on Langmuir films, is that the breakdown
strength haa been determined over a range of three decades of thickness which
ia very useful from the device applications point of view. The reliability
of these data lies with the fact that the onset breakdown strength, which is
of prime importance in the device applications, has been found to fit the
theories very well over the whole thickness range for all types of
Langmuir films studied. The a.c. breakdown and temperature dependent
behaviour have only been done on barium stearate films and hence more
experimental work must be done before one can place too much reliance on
these data. Presumably, the recently proposed theories must also be modified
to explain the observations of the latter two types of experiments on Langmuir
films. necessarily, further development in the understanding of electrical
breakdown depends on the nature of experiments and one must identify the
physical mechanism rather than simply study the dependence of breakdown strength
on various parameters.
VII. FORMING PROCESS AHD DIFFERENTIAL NEGATIVE RESISTANCE
The .application of voltage greater than a minimum Vp across a metal-
thin film insulator-metal device nay cause a radical and essential change-in
i ts electrical properties, such as a large permanent increase in the con-
ductivity. This process is designated "forming" process and V? is known as
the "forming voltage". This process has been found to occur most readily in
many thin film,,insulators with reactive anions, such as oxides (SiCL, Ta2<> ,
AlrtO_) and fluorides(caF > MgFp. MnF ) and was thought to be a characteristic
1 -58-
feature of non-stoichiometric insulators. After the sample was "formed", the
device also showed a pronounced differential negative resistance thenceforth
DNB) in its I-V characteristics. Until a few years ago, DNR behaviour was
established only for amorphous or polycrystalline insulators. Recent invest-
igations on Langmuir films * have shown that LKR also occurs when
organic monolayers are sandwiched between metal electrodes and formed. There-
fore the phenomena under consideration are to a large extent independent of
specific properties of the insulating material. However, some distinct
differences vere observed in the case of organic monolayers, which will be1 described in the course of the discussion. Before we proceed to dicuss the DUE
in Langnruir films, some of the characteristic features of the "forming" process
will "be outlined and. the models to explain the process will be discussed
"briefly. For greater details on the subject the readers are referred, to a132)recent review paper
1, Factors influencing forming
The major factors which influence the degree of forming are the applied
voltage and its nature, temperature, atmospheric conditions in vacuum, in-
sulator thickness and the electrode material. Application of a voltage pulse
( V ) of a few seconds duration or a sinusoidal voltage of amplitudes greater
than V at room temperature "forms" the sample to a degree reflected "by the
maximum current flowing through the sample. The new I-V characteristics of
a formed sample exhibit currents larger by a factor a 10 than the unformed
If, however, the voltage exceeds VF, the degree of forming may increase further,
providing V does not exceed a certain final forming voltage to cause catastrophic
breakdown of the insulator . The rate of the forming process is very
much sensitive to the temperature. In general, if the temperature is high
enough, the sample will form to completion at a faster rate than it does at
lower temperature. At a particular value of Vp , the device may continue
to form if the temperature is increased without any change in V^ . However,
the degree of forming is constant at given Tp (forming temperature) and Vp
(final) providing the latter is not increased any more. It has been shown that132)
the forming voltage is independent of the insulator thickness . Neverthe-
less, the degree of forming does depend on the thickness d and the peak current133)
of a formed device is approximately given by :
Imax
-59-
The degree of forming being sensitive to the peak current , should alsoincreasing
depend on thickness. In fact, the degree of forming decreases with/thickness, the
thickness range of the insulators suitable for the forming process is regarded
to lie between 100 k - 3000 A . As already indicated, for a thin insulator,-
V p may exceed the breakdown field and also the probability of electrode-electrode
tunnelling becomes significant. The decrease in peak current is observed
with increasing insulator thickness ^^^.
The forming process is found to occur under varying vacuum conditions,
for instance, pressure of 1 torr or lower was found to be adequate by Hickmottj jl ^
10 to 10 torr was found necessary in the investigations by Barriac et al. 1 -
who also reported that no forming occurs at 10 or above. The widely accepted
pressure, however, is 10 torr or better. This process is not affected too
much if gases like He, Ke, H , Ar are used instead. However, an oxygen
atmosphere completely inhibits the forming and may have permanent effects. The
forming process is investigated using a wide variety of electrode materials such
as Al, An, Ta, Zr, Ag, etc.. with different combinations of the counter-
electrodes. Different behaviour of the device has been exhibited with different
polarity of the electrodes. The dependence on electrode material is different
for the anode to which forming is sensitive and the cathode, vhich does not1^2)
appear to affect forming. Dearaley et al. ' have tabulated different
electrode materials based on the experimental observations and have shown which
of these exhibit forming or not in different conditions.
2. Models of the forming process
Since, Hickmott pronounced the occurrence of "forming" in M M
devices and proposed a model to explain the observed behaviour of the device,
many more models have been developed. Each of these explained the "forming"
and consequent changes in the device characteristics in s somewhat different
manner. However, none of these models accounts fully for the numerous
observations demonstrated in different devices. To remind the reader of the
very basic concepts of these various models, we reproduce a concise summary of1S2 i
all five models in a tabular form (Table III), after Deamley et al.
More discussion will follow when the experimental observations on Langmuir
films are described later in this section.
Once the device is "formed" many electrical phenomena have been observed,
which an unformed device does not exhibit. The most commonly observed is the
differential negative resistance together with switching and memory phenomena.
In some structures, however, electroluminescence and electron emission might
also occur. Since Langmuir films exhibit only DNR in both the experiments on
diode structure (MIM) and triode structure (MIMIM) we shall discuss
-60-
briefly how the DNR can be explained In terms of various models. The other132)
observed phenomena* described in an earlier review , vill not be included
in the present paper. Further, ve shall limit our discussion to the voltage-
controlled (H-type) type DNR which has been observed in Langmuir films.
The earliest model proposed by Hickmott 135J»136J states that the
ini t ia l part of the I-V characteristic is due to space-charge limited con-
duction in the impurity band of the insulator. When a high field is applied,
some process reduces the number of impurity centres which in turn reduces the con-
ductivity and an N-type characteristic results. For instance, neutralization
of the impurity centres by field induced inter-impurity tunnelling may be one13U)possible process to cause the reduction of conductivity. In another model ,
the electron transport mechanism has been assumed to be hopping between traps
of very similar energies when the applied voltage exceeds a particular, limit ty t
below which electron tunnelling from one electrode to another is probable.
At high voltages (> C ), the insulator can be crossed only by those electrons which'
are entering the insulator with energies close to the electrode Fermi energy and
onlyif they are making transition to traps of lower energy. Since the top of
the impurity band in the centre of the insulator has been assumed to fall below
the anode Fermi level, charge trapping would occur in the depletion layer
near the anode. The density of states of electrons within the electrode
decreases resulting in a rapid fall of the number of electrons contributing to
conduction and thus giving rise to the negative resistance in the characteristic.
In a model by Barriac and co-workers non-ohmic behaviour at low-
voltage has been attributed to space-charge limited ionic currents. They
have thus introduced an additional feature of ionic motion unlike in the
previous models of gickmott ' and Simmons-Verderber . At higher
voltages, the dominant transport mechanism has been suggested to be electron
tunnelling,which neutralizes the positive ionic space charge by being trapped.
Consequently, the two current-carrying components are largely cancelled and an
N-type characteristic results.
A model based on an altogether different approach involving some
conducting filament formation in the device is due to Dearnley
Basically, the model assumes ohmic conduction along the filaments which are
physically different from the host matrix. The appearance of negative
resistance has been attributed to the rupture of these filaments because of
the .Joule heating to a point exceeding the melting point of the matrix. On
fracture, the filament becomes non-conducting,resulting in the fall of
conductivity- An earlier model having features in common with
Dearnley's model postulates only one filament and explains the occurrence of
-61-
switching in the device but no negative resistance. One of the recent models1^2) 132)
by Sutherland is similar to that of Dearaley et al. and this also
assumes that the DNR arises from local breaking of thin conducting filaments
by Joule heating. Thus, al l these models discussed above propose diverse
mechanisms for DIIH and i t will be shown later in this section which of these
offer interpretation to the observations in Langmuir films.
3. Experimental observations on Langmuir films
Experiments were performed both on diode structures ,(metal-organic
film-metal) and triode structures (metal-film-metal-film-metal) using Al
electrodes and organic films of cadmium arachidate. The unformed samples~3were placed in a sealed brass box evacuated to 10 torr whose temperature
could be varied between TT°K and room temperature. Then a unipolar sawtooth
voltage, which increased at 0.3 V sec and decreased to zero within 0.02 sec,
was applied across the devices. The J-U characteristics do not manifest DNR132) \3h)in the return trace if the voltage is removed so quickly * . The
measurements on diodes have used 11-layer thick . films and on triodes 9-layer
thick films. In the case of diodes, characteristics are also obtained with
Au as the upper electrode to explore the influence of electrode material on
the characteristics. Some devices with Au as the base electrode were preparedU2)
which invariably showed short circuit as was reported by Handy and Scala
a) DNR on diode structures
Once the sample is "formed" at TT°K with top Al electrode positive,
typical J-U characteristics are obtained, similar to that shown in Pig.29
for a sample consisting of 11 organic monolayers. As a rule, DNB is not ob-
served in the first voltage cycle but appears in the second one. However, it
disappears in the successive two to four further cycles and finally the
characteristic becomes smooth. Fig.29 at T7°K corresponds to this smooth
characteristic. On warming up of the device at about 5 deg/min, the J-U
characteristic changes at about 150°K and behaves at different temperatures
as shown in Fig.29, Evidently, the most pronounced DNR (peak to valley
ratio of the current is maximum) is observed at about 190 K and it goes on
diminishing beyond this voltage, disappearing above about 210°K . Almost
similar results were obtained with Au top electrode but the device always
exhibited short circuit above about 170°K and in this case appreciable DNR
occurred at about 150°K.
To investigate the temperature effects, the unformed sample was
always cooled to 100°K and sawtooth voltage applied until the device exhibited
smooth characteristics. In Fig.30, curves 1, 2 and 3 represent the respective
-62-
characteristics of an unformed sample in the first voltage cycle; formed
sample and second voltage cycle exhibiting DNB; smooth, characteristic after
a few voltage cycles. The device vas now treated at a particular temperature
in the range(ll0OK - 300°K) and every time the voltage was removed while
raising the temperature. The DHR was observed only for the first few voltage
cycles in the systematic studies of different temperature regions. The
typical curves so obtained are shown by the curves h to 7 in Fig,30. The
observations made in different temperature ranges are:
i) 110°K - 150°K exhibited DNR but not well pronounced
ii) 150°K - 190°K better DNR than above
iii) 190 K - 225 K pronounced DNR (see curve k Og Fig.30 )
iv) 225°K - 300°K increase in conductivity; little or no DNE.
It was also observed that the current maxima shifted towards lower
voltages with increasing number of voltage cycles until the DUE disappeared.
However, it could be re-obtained by increasing the temperature in the range
100 to 225°K . Thus, it was found that the larger the temperature increase*
the better the DNR is developed. Curve 8 in Fig.30 illustrates a character-
istic at 190°K>vhich the device exhibited in the first voltage cycle after
interrupting the voltage bias for about 3 hours. The DNE is more pronounced
for longer resting times. Curve 9 shows a smooth characteristic obtained
after a few more voltage cycles at 190°K after interruption as in curve 8. •
Further experimental observations concerning DNR in diode structures by
Gundlach and Kadlec may be summarized as below,
i) The device could be formed by applying 5 to 6 V or more when
the top electrode (Al or Au) was biased positivv.-ly. The temperature at which
the device "formed" was T7°K unlike in the case of many oxides and insulators
(T = 200°K for Al-SiO-Au 133' and T a 250°K for Al-Al-oxide-Au *). This
has been explained in terms of the very weak binding forces between continuous
hydrocarbon chains oriented perpendicular to the film plane of the insulator.
In contrast to oxides which do not have such an anisotropy in binding forces,
the mobility of impurity ions (here metal ions) along the long chains of the
molecules is expected to be higher than in oxides. For the fact that the
device cannot be formed (at 77°K) when the base electrode is positive, a
qualitative suggestion has been made that the thin Al-oxide film (formed
between the lower Al electrode and the Langmuir film) inhibits the metal ion
injection from the base electrode into the organic layers.
-63-
: ; ) Once the device is formed, the DNR is observed for both polarities
of voltage bias. The threshold voltage for the appearance of DNR was found"
to l ie between 2 and 3 volts.
i i i ) The investigations of DNR with respect to the multilayer thickness
showed that the former becomes mote pronounced with the increase of the la t te r .
The average values determined in Ref. 130 for the peak-to-valley ratios from the
observations of 35 samples, were U.2, 2.1 and l.li5 for 11, 9 and 7 monolayers,
respectively. No DNR was observed for the devices having less than 5 mono-
layers, Gundlach and Kadlec were not able to explain this behaviour
of the characteristics.
To explain the existence of DNB, no single theory was found adequate.
However, the. appearance of DNR below 225°K (see Fig.29) has been related to
the so-called "dead time" which is a measure of the time between filament132) 132)
fracture and reaching a state from which i t may reform . In this model
the transition between two states is interpreted as relaxation of space charge
polarization. On the other hand, the dead time is related to electron13M ll|3)diffusion through the insulator " . However, both of these models
predict strong increase of dead time with decreasing temperature. It has been
thus suggested that these devices, presumably, show a dead time of the order
of hours because DNR appears below 225°K, But life was not so simple because
the dead time has been reported to increase irreversibly from seconds to hourso lUU)
by warming the sample up to 500 K for about an hour . This observation
was made on Al-Al oxide-Au samples. Finally, the results have been left with-
out an adequate explanation by the existing theories.• - ' within experimental error,
The peak voltage U was found to agree,/with an empirical relation1?61 •P ' r • ^ '
given by Hickmott , U = k - O.36 v£y. , when the device was warmedsteadi ly (see Fig.29) . For cadmium arachidate, the d i e l ec t r i c constant
CO \€.= 2.1+ was considered to obtain U a 3.1+5. However, the values ofr • p
U measured at constant temperature (Fig.30) are much .higher. ThereforeGundlach and Kadlec 130^ suggested that U cannot be related simply to band
Vmodel parameters which assumes the constancy of U for the devices of thesame material. Confirmation of this suggestion appears from the measurements
on Au-ZnS-Au devices , It has been shown that U can be increasedP
drastically by applying voltage in the region of current minimum of the
characteristic for a period extending to few minutes. In the model proposed
U is related to the distribution of filament resistance which in turn dependsPon spacing of positive centres in the conducting filament. To some extent.
the observations of the present workers seem to be supported by the filamentary
models which imply that the dead time and the voltage for current peak (U^)
-6k-
can be changed for a given sample, hut no satisfactory explanation could "be
possible. The disappearance of DWR above a certain temper attire has also been
left unexplained. Nevertheless* these observations hava shown that low
frequency DNS does occur in Langmuir films and the phenomena under consideration
are to a large extent independent of specific properties of the insulator
material.
b) DHR on triode devices
Hickmott made the first attempt at studying this phenomenon in
triode structures Al-(cathode)~SiO-Al(grid)-SiO~Au(plate). Similar studies
were also performed on aluminium oxide triodes . In both the formed
devices, the potential distribution was highly non-linear between the plate-
grid ( V ) and grid-cathode (V ) regions, V was proportional to V and
was rather small whereas V was nearly equal to V (pc refers to plate-cathode
region). This was the situation when the DNH became fully pronounced. When
the polarity of V was reversed, the voltage distribution did not change,pcHowever, the main voltage drop shifted to plate-grid and then V vas nearly
egproportional to J . [in the discussion V , J , V and V refer to
cp pc pc pg gc
plate positive and V , J , V and V to plate negative.] Hickmott
could not satisfactorily explain these phenomena in triodes.
Gund]ech and Kadlec after a successful attempt to study the DNR
on diode structures (discussed above), extended their studies on triodes by
malting use of organic monolayers and following the arrangement and circuitry,136)
etc., of Hickmott . However, all three electrodes used were of aluminiumo
to avoid unnecessary asymmetry. The grid,of thickness * 50 A, contained a
significant number of pinholes and the other two electrodes (plate and cathode)
were several thousand angstroms thick. The device was formed at T7°K as
in the case of diodes; DNR was found to occur in the second voltage cycle
and to disappear in the next few cycles. At about 1TO°K , the DNR was more
pronounced and V and V changed little as compared with that at 77°K (VpC
was equally divided between V and V at 7T°K). In Fig.31, the J -Vpg gc pc pc
characteristic at 170°K and the change of V and V with respect to Vare shown. Evidently, both V and V increase nearly proportionally to
pg 6CV but these are independent of the J - V characteristics. Thepc PC pc
explanation offered for the latter observation is in terms of the filamentary
model of Dearnley et al. according to which DNR occurs by filament
breaking because the fractured filament does not contribute to the current.
After the so-called dead time, the filament may reform , which presumably
occurs at about 2,8 V in the present case (see Fig.31). It has been assumed
-65-
that the current J flows along the conducting filaments which cross thepc
grid at pin:holes.of
Wanning up/the trio&es further results in a drastic change in the
J - V characteristic as veil as in the voltage distributions V and Vpc pc pg gc
At about 230 K, the DNE disappears (see Fig.32) which is typical for organic
multilayer triodes. A slight non-linearity in the characteristic is observed
in the plate-grid region; the spike 'a1, however, originates in the grid-
cathode region. The dotted curve in Fig.32 is a J - V c characteristic at
205°K,vhich has been considered to represent the sum of two types of currents
below 5.5 V. The first is slightly non-linear without shoving JMR and the
other shoving DNR vhich ceases at 230°K. Non-occurrence of DNR above a certain
temperature and hence the disappearance of the current associated with DHR
has been interpreted by assuming that either the filament does not reform or
i t needs rather a long time. The presence of spike 'a1 and the slightly
higher current above 5-5 V haY.e been attributed merely to some new events
occurring at 230°K. Nothing in particular has been stated about these new
events.
After a sufficient number of voltage cycles, the voltage V or VrO g
was found to be generally proportional to J ; unlike that shown by the dotted
curve in Fig.32. Reversing the junction bias at 270°K (plate negative) led
to an increase in V from 7.5 V to 9-5 "V, which resulted in higher currentcp
J (than J ) and , the re fo re . a changed J - V c h a r a c t e r i s t i c . On the o thercp pc ' cp cp
hand, V decreased considerably but i t was s t i l l p ropor t iona l t o J . OneC£ 131) c p
remarkable fea ture i s t h a t Gundlach and Kadlec a l so succeeded t o developthe main vol tage drop in the grid-cathode region when V was p ropor t iona l t oJ . This was not repor ted in t he e a r l i e r experiments on oxide t r i o d e s
pc
Thus, i t has been concluded t h a t t he re i s no region in t he monolayer t r i o d e s
which e x h i b i t s p r e f e r e n t i a l DNR and i t might be due t o t he e lec t rode symmetry
of these t r i o d e s . Resistances R and R of a formed device (measured a tpg gc
0.3 V) were found to be 1 to 2 orders of magnitude higher than R .which
resembled the observations on SiO triodes
In both the experiments on diodes and triodes, Gundlach and Kadlec
have , explained the observations qualitatively in terms of the filamentary
models; nevertheless, they have remarked that the model itself is still
speculative. With the use of Langmuir films as the insulator, some typical
observations, e.g. the device formations at 77°K, the dependence of DNR on the
film thickness and disappearance of DNR above a certain temperature, are yet
to be interpreted. Therefore, the immediate requirement is the development
of a new model essentially based on filamentary models which could partly-66-
explain the DKR phenomena in these films. One suggestion of Gundlach
and Kadlec, that the thin oxide film (between the "base electrode and the
organic multilayer) inhibits the injection of metal ions from the "base electrode-
into the organic layers does not appear to "be convincing.
In the recent work on thermally stimulated currents in sueh organic
layers» the device used was Al-film-Al ana the ion displacement from one
electrode to the other has "been shown to form ionic space charge vhich is
responsible for the polarization. Particularly in future works, the
possibility of the occurrence of several other phenomena,like switching and
memory, which have already been observed in a wide variety of oxides, must
be ascertained in Langmuir films. This might seem to be a speculation at the .
first instance,but the reader must remember that even the WB. was never thought
to occur in these monolayers before Gundlach and Kadlec made a successful
attempt. At least, the studies of DNB have shown that such a phenomenon is
not restricted to insulators of a particular chemical composition and crystal
structure. More investigations on DKR occurrence must also be carried out
using different types of organic films sandwiched between varying metal
electrodes to ascertain the Influence of the electrode material on "forming"
process and to strengthen the proof of DNE in these insulating materials.
-67-
VIII. IONIC TRANSPORT PHENOMENA IN LANGMJIR FILMS
The study of ionic polarization and transport in -many insulating
materials has been pursued *by workers engaged in this film investigations.
Such studies were reported by Curie as early as 1886 on quartz crystals,
and subsequently Warburg and Togetmeier demonstrated the steady-state
transport of ions like Na , Li and K through quartz at temperatures below
25O°C. Since then, ionic motion has been shown to occur in a vide range of
insulating or semiconducting materials and there has been a many-fold
development of the subject.
In recent years, some problems of commercial importance, for instance,
the surface stability of MQS structures for reliable operation, have been
associated with the motion of ions within the oxide layer on the semiconductor
surface and,therefore,the necessity of such studies was realized. The change
in the capacitance-voltage characteristics of MIS structures has been
suggested to provide a powerful tool for the observation of ionic motion in thinIkf)
insulating films. Similarly, the method of measuring ionic thermocurrents
is known to be a more sensitive method to investigate polarization effects due to
ionic motion in insulators. It has been shown that this method reveals finer
details of the phenomena than the earlier two different methods in use,i.e. by
considering the change in time of the electrical current in a device subjected
to d.c. external field * and by measuring the dielectric losses
Fortunately, in recent years some workers have made successful attempts
to study Langmuir films in the form of MIM and MIS structures to determine
the polarization effects and C-V characteristics, with respective structures.
Here we shall mainly be concerned with the discussion of such studies on Langmuir
films. Preliminary data obtained on change of capacitance with applied voltage
on these films will also be given in this section.
1. Properties of MIS structures
The study of capacitance-voltage characteristics in MOS structures has
proved to be of potential importance in developing a certain type of devices. To152}
investigate the usefulness of Langmuir films for such devices, Tanguy
carried out studies on MIS structures using the organic films of Ca-behenate
and orthophenanthroline with three stearate chains. The p-type silicon
surfaces, after etching in CPlt, cleaning and drying;were used to deposit organic
multilayers and were then covered "by aluminium in the form of two circles
-68-
connected lay a channel wnich served as the top contact. The back contacts
were of evaporated gold or of an Au-Si eutectic. To know the thickness for onemonolayer of ortliophenanthrollne, measurements of the device capacitance C were
themade with respect to/number of monolayers IT deposited. The linear relation
between l/C and H thus gave a monolayer thickness of about 15 A, All the
capacitance measurements were performed on a General Radio 1620 A bridge under low
pressure for good reproducibility of the results- Fig. 33 shows the C-V
curves obtained with 3» 5 and 9 monolayers of orthophenanthroline in which both
accumulation and inversion regions are visible. Evidently, even a small
variation in bias voltage results in a large inodulation of C, which has
been thought to arise from the small thickness of the insulator and from the high
resistivity (- 600 ft cm) of the p type silicon. The curves shown in Fig, 33
represent the situation a little time after etching. In the case when silicon
was left for a few hours in the ambient conditions before roonolayer deposition,
the flat band voltages shifted towards the positve charge and the slope of the
curve was found to become less steep than shown in Fig. 33. When the aluminium
top electrode was replaced by gold, the flat band voltage shifted to zero and
the distance between the flat band voltages of both curves corresponded to the
difference between the respective work functions.
In several studiesJ-53j-15oJ o n M o s structures, the two types of
hysteresis, normal and anormal, are attributed to the ion displacements in the
insulator and to the trapping at interface states. Tanguy ^ ' also obtained
both typesof hysteresis (shown in Fig. 3i+}- Fig. 31+ shows C-V curves of
5-layer thick orthophenanthroline (OP,) device at various temperatures. The
normal hysteresis is related to V (flat band voltage) shift opposite to the
voltage bias.whereas the anormal one is characterized by V__ displacement towards
the bias voltage. In accordance with the earlier workers, Tanguy has also tried
to correlate the existence of these two hystereses with ionic motion in the
insulator and trapping,respectively. Presence of ions in the fatty acid soaps is
a well known feature as these are ionic substances forming stable monolayers
{•Ref. 7.p,i93) and there is a possibility that some ion content might get
transferred during monolayer deposition. Many times, ions can be adsorbed later
during electrode evaporation. Tanguy has, however, made use of the novel
technique of studying thermally stimulated currents (TSC) in these organic films
to make sure of such ionic motion. TSC measurements were performed on MIM
structures with one layer of 0P_. These experiments gave a clear-cut indication
that the short circuit currents arising on warming up of the structure are a
consequence of ion displacements because the amplitude of the current peaks varied
greatly with the bias temperature. At lower temperatures, such peaks were not
' ; -69-
observed. Further, it was estimated from TSC peaks at 20°C that about
2 x 10 charges are displaced or trapped, which corresponds to a C-V curve
shift of the order of 100 mV for 3 OP- monolayers. To make sure of this
statement, C-V curves were obtained for 3 OP. monolayers with different
temperature M a s conditions which yielded the V,™ shifts agreeing quite well
with the above estimation. This estimation assumes that the mobile ions can
move from one electrode to another. In a separate measurement of TSC in
calcium behenate monolayers, it was found that the number of free ions is less
important, which has been attributed to the fact that stearic acid chains are
more compact in behenate than in OP .
Anormal hysteresis, which appears below 20°C in the case of 0P_
(see Fig. 3*0, has been found to become very important for calcium behenate.
In the latter case, a continuous Vp shift extending to about I V in an hour
has been observed. To explain the existence of anormal hysteresis, Tanguy
assumed that the traps occur in the natural oxide layer. However, it could not
be proved in these preliminary investigations that no trapping exists in '
molecular layers. In one of the models155) proposed to explain the anormal
hysteresis in MOS structures, it has been assumed that the holes emitted from
the valence band of the silicon are trapped in the oxide or at the interface and
these positive charges cause V_ shift towards negative voltage. Some workersI'D
found a correlation between the number of charges in the oxide layer and
the number of surface states in the silicon. This has been regarded as an
explanation of the observed instabilities with the calcium salt in the presentinvestigations.
1*5)In an earlier communication, Tanguy and co-workers reported I-V
characteristics on MIM structures using OP as an insulator. When similar |
experiments were repeated on MIS structures by Tanguy, the currents observed :
were in good agreement with those obtained from MIM structures. It has .
been found however, that the currents are weak compared with those obtained from
MOS structures j5°) with the oxide of comparable thickness. Fig# 3c *
shows the I-V curves for MIS structures obtained on 1, 5 and 9 monolayers of I?
OP . Evidently, there is a shift of I-V curves near zero volts which reveals ;
the existence of an electromotive force in the monolayers when there is no f
external bias. It has been suggested "by Tanguy that the e.m.f.'s arise, ;
presumably, from chemical reactions in the layer at the interface. In an earlier \
work "by Leger et al. , the existence of e.m.f. in MIM structures was •
reported. It was found that the Junctions act as batteries because a voltage I;
of about 200 mV was obtained when a high impedance voltmeter was connected !•
across the junctions. The existence of e.m.f. has been suggested to arise !
due to the motion of charged ions across the insulator to attain the equilibrium \
- 7 0 - • .• I
state. These workers could not cement "what type of ions actually cross themolecules "but some kinetics- of these tons nave "been etudied in the range
20°-T0°C. For instanca, th« Internal reflistunoe R i n t * ftxp(E0/KT} wheraEQC- O.36 eV)is the activation energy, vhich is comparaTale with the activation
energy {0.6 eV) required for water molecules to cross a fatty acid monolayer1 en 1
spread on water surface to prevent its evaporation Jyj, It has also been-
noticed that the initial e.m.f, of the battery depends on the nature of the
sample under investigation but EQ remains the same. To realize the practical2
importance of observed e.m.f. in these junctions, 226 Junctions of 0.25 ram
each were prepared and connected in series. The values obtained for total
voltage was 50V, R. , - 10 fi and the total volume of MIM structureslilt*2
- 0.1 imn . . It has been thus suggested that such a high impedance battery with
so small a volume might prove of special significance in a few practical applic-
ations. The capacity of these cells {- 1 uc) has "been found to correspond to
the oxidation of about one atomic layer.152)
One of the most significant contributions of Tanguy has been in
obtaining the reproducible MIS structures which behaved almost as MOS
structures. This has opened up a new field of investigations for studying semi-,
conductor surfaces using organic molecules which might be exposed to high electric
fields. In these "basic investigations, occurrence of anormal hysteresis in
MIS structures could not be explained very well and,therefore,further studies
with other types of organic molecules are needed for a better understanding of
the phenomenon. Technological applications of MIS devices, like those of MOS
devices, must also be worked out after successful studies on MIS structures.
2. Thermally stimulated currents
1VT).16O) a ^
The ionic thermoconductivity method , which involves the study of
the thermal activated release of dielectric polarization has already been used in
a wide variety of dielectrics. The essential steps of the method consist of;
(i) Polarization of the sample in a static field E at the temperature T for
a time t - The temperature should not be too high to develop heavy space-charge,
(ii) The sample is then cooled down to a temperature TQ•« T so that the
relaxation time X'fT ) is of the order of several hours or longer, (iii) The0 , ,
electric field E is now turned off (at T = Tn) and the dielectric is warmedp u .
up with a constant rate g = dT/dt -and'the "discharge current is- registered as a
function of temperature. The polarization current i(T) first increases
-71-
exponentially, reaches a maximum and then drops to aero. Thus the analysis of
the peak obtained allows- the determination of the activation energy of the
relaxation process and reveals the finer details of polarization.ICO \
Tanguy J ' initially used this method on Al-orthophenan-
throline-Al structures- to observe ion displacements in -monomolecular films. Such
studies were carried out to explain the normal hysteresis in MIS structures and
it was stated that the mobile ions in the mono-layers can move from one electrode
to another. In the structures studied, ion displacement was held responsible for
the large variation in the amplitude of the current peaks at higher temperatures.
Later on, Tanguy and Hesto have -made a systematic study of the polarization
phenomena in MIM structures by thermally stimulated currents (henceforth TSC)
method. We shall first discuss the theory used by these workers to explain TSC
in organic films and then the experimental observations will be described.
a) Theoretical model
Of all the models, the one found suitable to explain the TSC peaks
assumes that the electrical charges can move between two potential wells
separated by only a fev Angstroms. In this model, the charge contained ~by the
two potential wells changes when the device is subjected to an applied field and
gives rise to a volume polarization called orientational polarization in the
specimen. The position and shape of the TSC peaks thus obtained depend only
on the potential barrier height and not on the initial experimental conditions
(temperature,electric field, polarizing time) as in the other theories
Assuming N pairs of potential wells per unit volume which are randomly
distributed and if 2a is the separation between two centres, the total polar-162)
ization Pj is given by :A
P = eaN Z tanh (uZ)dZ , (8.1)
00
eaETwhere u a ^m
• P
If the electric field EL is relatively small, then u « 1 and Eq, (8.1)
reduces to „/ ,2 „B(ea) E
Pi = 3kTP
-E (8.2)
which shows that polarization is a linear function of the applied field K .
Further, the relaxation time T which is the average time needed by a charge to hop
from one position to another over a potential barrier £ » has been defined as
% ' (8.3)
where rR is a constant.
-72-
The initial polarization is a function of the polarization time t
during which the sample is held at the initial temperature T (room temperature)
according to
If a MIM device is then cooled rapidly to 77°K, it releases the stored
charge on reheating so that the discharge short current may be expressed as
"dP _ P
dt"~~r > (8.5)where P is the deviation of polarization from equilibrium.
If the reheating is a linear function of time, T = T- + gt, where 3 is0 160)
a constant, the discharge current as a function of temperature can be writtenas (using expressions (8.3) and (8.5)1
P. . . . _
(8.6)
This polarization current gives a maximum at T = T and its' derivativem
at T = T presents a relation between T , €. and tn expressed as;m m u
Clearly, from Eqs. (8.6) and (8.7), the shape of TSC depends only
on the parameter £ ancL the maximum peak amplitude depends only on P . In
their wort, these authors have fitted the experimental peaks on the theoretical
curves of Eq. (8.6) and determined the polarization parameters £ and Y o .
Separate values of N and a have also been calculated by fitting the experimental
curves of P as a function of the electric field in the theoretical curves
obtained from Eq. (8.1).
b) Experimental measurements
Tanguy and Hesto tried to obtain data under various experimental
conditions,i.e. with varying temperature (T )» voltage (V ), time (t ) and. P P P
specimen thickness (a). in the measurements with respect to T« on different
numbeisof monolayers of OP^, the TSC peaks were found independent of T .
Fig. 36 gives a graphical representation of TSC peaks on 23 monolayers thick
0P3 at various temperatures ranging between 28l°K to 315°K when all other
parameters V . t and 3 (rate of heating) were kept constant. Evidently, a
-73-
polarization maximum is observed at T = 301°K and there is a decrease beyond
this temperature due to thermal agitation. This is clearly compatible vith the
theoretical expression (8.1*) at high temperatures vhen *y "becomes small and
hence P. reaches a maximum. However, for T > 301°K, the peak tail has been
assumed to appear because of space charge formation. Since TSC peaks are veil
defined in position at room temperature, absence of space charge formation has
been concluded.
When MIM structures were biased at different voltages V for a fixed
time at fixed temperature, the TSC peaks obtained vere perfectly stable. According
to expression (8.1), PQ must show a saturation at increasing electric field for
u > 1, which these workers failed to obtain for 3-monolayer thick 0P_ structures6 -1
because the applied field exceeded the dielectric breakdown (~ 5 x 10 V cm" ).
However, on thick structures, such saturation was found to begin at E = 10 V cm"
for 7 layers and E = h x io v cm" for 21 layers. It thus appears that the
applied electric field sufficient to obtain maximum polarization depends upon the
thickness of the film which in turn increases a in the expression for u (see
Eq,.(8.l}). While studying P as a function of polarization time t for 3
monolayers at 8°C, keeping the bias under one volt, two distinct curves corres-
ponding to t, 60 seconds and f > 2000 seconds were obtained. The former has* 2
been related to the charging time of lower energy peaks and the latter to thecharging time of higher energy peaks. The time needed to form 80$ of the total
polarization was found to be ten minutes at 8°C and to reduce to one minute when
. the temperature was increased to 20°C.
Thickness-dependent studies of total polarization showed that the total
polarization charge increases almost linearly vith the number of monolayers Nb
when T , t and E are held constant. Clearly, increasing Kb increases the
value of a in Eq. (8,2) and thus the polarization charge increases because of
the Ma term. According to this expression.(8,2^ the total polarization must
not vary with increasing lib for a given electric field E because the numberP
of potential veils N in the dielectric does not vary vith che dielectric thickness.
Since, the basic structure of the monolayers changes with increasing dielectric
thickness, the respective distance between two potential wells 2a and the
activation energy c are modified resulting in the higher energy peaks of group d
with increasing Mb Fig. 37 shows the TSC peaks obtained at different dielectric
thicknesses vhen all other parameters were identical.
One of the important measurements carried out by these workers is the
study of influence of water molecules which, presumably, are adsorbed near the
electrodes (possibly in the alumina). Initially, when a MIM structure was placed
under low pressure and heated, a large space charge accumulation appeared and a
broad peak vith a maximum was obtainedt This "was the situation for temperatures
-Ik-
loeyond 6O°C and vlnen the dielectric thickness was relatively small, For
higher thicknesses CK 23 i&onqlayers-I, the' space charge disappeared under low
pressure at 3O°C during approximately one hour. On the contrary, th«
elimination of space charge needed about US hours for 3 monolayers and a few
days for 1 monolayer. Thia has teen qualitatively explained in terms of the
influence of the electrodes on the electrical properties of the MIM device
which decreased with increasing dielectric thickness. It has "been found that
the polarization due to residual water molecules became negligible after a short
stabilizing time of less than an hour at 30°C.
Tanguy and Hesto (63) further attempted to calculate the polarization
parameters by fitting the experimental peaks to the theoretical curve according
to Eq.. (S.6). In their experimental investigations, several groups of peaks
were identified, each characterized by a polarization temperature. In particular,
they concentrated on d peaks which are more pronounced in the case of multi-
layers and are composed of several mono-energetic peaks d. » d_ , d^ , d, and d
as shown in Fig. 37 • Since the polarization was found to depend greatly on
the temperature and polarization time, these workers fitted two peaks obtained
with 21 monolayers (see Fig, 38) at temperatures of 290°K and 297°K and
obtained the characterizing polarization parameters of all the d peaks. In
another attempt, a fit of theoretical values of Po(Ep) (according to Eq.(8.l))
with the experimental curve for 21 monolayers vas also obtained and the
parameters N and a were obtained for 3, T, 21 and 170 monolayer-thick film
structures. It has been reported that for depositing 170 monolayer thick film,
OP,. molecules dissolved in "benzene were directly laid on the substrate and the
solvent vas allowed to evaporate, From these theoretical values, it was found
that the polarization increases with increasing Kb , presumably due to an
increase in the average distance 2a between tvo ion equilibrium positions. In
fact, 2a has been found to increase from 1.2 A for 3 monolayers to 10 X for
27 monolayer thick structures. This phenomenon has thus been regarded aa
evidence for some basic structural changes in the multilayer with increasing
thickness. From the calculated high values of activation energies ranging
between 1.1 eV to 3.1* eV and because PQ varied rapidly with the
polarization temperature, the hopping charges have been thought to be ions
which move between two equilibrium positions with 2a distance apart.
It has been discussed earlier that Marc and Messier ' " attributed the
two absorption peaks to the dipolar movement in the amorphous and crystalline
regions in the behenate monolayers (Sec.IV) and thus suggested some structural
modification in these monolayers. Similarly, these workers have assumed
that as the monolayers grow thicker and thicker, the molecules tend to form
-75-
crystallities whose number and volume increase with increasing film thickness.
This assumption has "been further Justified "by some experimental evidence. Forthey cpuId *liu
instance,^ observe the structural modification optically "by a colour change
due to light reflection on the lower metallic electrode when the multilayers were
heated to 35°C. At this temperature, presumably the CH2 chain fuses and OP
molecules are reordered forming crystallities. These crystalline regions in
the multilayers have been assumed to he the trapping centres for the ions vhich
are thermally released and can move in disordered regions giving rise to the
observed polarization phenomenon. It has been remarked,further,that calcium salts
of stearic and behenic acid are relatively little polarizable especially when the
chains are long enough and compact. In these eases, the polarization has been
attributed more to an ionic space charge formation.
Whereas Tanguy and Hesto ; have carried out a systematic and detailed
study of polarization phenomenon in OF- molecules, a few points are not yet
made very clear in their analysis. To the knowledge of the author, there is no
other reported work in which the thick multilayers could have been obtained by
simply laying the molecules,dissolved in solvent,on the substrate. It is rather
surprising how these workers could form a uniform film in this simple way.
Nevertheless, this simple method seems to be fascinating and,in future works»
specific attention must be paid to the suitability and applicability of
this method to many other organic substances. Another remark must be made
concerning the degree of structural modifications in behenate or stearate molecules
and OF,, molecules. According to Tanguy and Hesto , the structural changes
in the more compact behenate or stearate molecules are less evident,whereas Marc
and Messier J \ have based all their experimental results on structural changes
in these molecules. Therefore, structural studies have to be renewed and
scruplous attention must be devoted to this point. The role of metal electrodes
and of the oxide layer formed, must also be considered in future investigations.
3. Capacitance studies with applied voltages
All the experimental investigations discussed in Sec.IV relate to the
capacitance studies of MIM structures as a function of monolayer numbers which
have also been employed by various workers as a simple tool to obtain monolayer
thicknesses in the case of several organic compounds. Similarly, Leger e.t al.a
obtained the • thickness of a stearate monolayer, about 22.A. These workers
further studied the variation of capacitance of MIM structures as a function
of applied voltage V and obtained some useful information. Fig.-3°A
represents plots of AC/C (where AC is the change in the capacitance)versus
\ -76-
applied voltage V on Al^iarium stear&te-Al samples with, ths capacitor area
- 0,25 ma. , for a varying number of layers. Clearly, the nature of the curves
varies with the number of deposited monolayers and all the data can "be fittedAC 2
to obey the universal linear dependence of the type ~rr - AE (shown in theArt £> p
corresponding Fig.39B ' vith A = (1.6 ± 0.15) * 10 V m . In thia
figure the applied electric field E has been plotted in terms of V/H, H
being the number of deposited monolayers. Obviously, the curves for AC/C
versus V are not linear and this has "been taken to be the evidence for an
internal field inside the organic layers. These curves, which have a minimum
for V = 0 t lOOmV, have been regarded to. set an upper limit for an internal2
field. The assumption that the linear dependence of capacitance on E is due to
the compression associated with the electrostatic pressure on the capacitance
plates,leads to a compressibility coefficient of the stearate molecules which
would be one order of magnitude larger than for bulk paraffins. Although, such
a value could be plausible, other phenomena,e.g. electrostriction action on the
dielectric constant of the insulator, should possibly be taken into account.
In the opinion of the author, it is hard to draw any conclusion from these
preliminary measurements. The physical mechanism suggested by these workers
does not appear to be conceivable until further measurements could be made.
However, the presence of an internal field inside the monolayers reported by
these workers have been verified further in the ionic transport studies of these15 2)
structures and some similar indications also come from the work of Tanguy ' .
As has been suggested "by Leger e^.al. themselves, the best course would
be to repeat similar measurements on various organic substances and to find out
whether the order of magnitude of the compressibility coefficient of organic
molecules is plausible.
IX. FUTURE TRENDS ABD PROPOSED APPLICATIONS
In fact, much has already been said regarding the future trends of
research on Langmuir films. Nevertheless, to sum up, a few more remarks are to
be made which concern us in any type of electrical studies of these films.
Essentially, one of the challenging problems is to solve the riddle about the
structure of Langmuir films. It has been regarded for a long time that the films
obtained by B-L technique form ''two-dimensional" crystals,whereas some recent
investigations completely rule out this possibility and indicate some structural
changes inside the organic layers. Renewed structure studies are particularly
important because all the electrical properties* by and large, are "structure
-77-
sensitive". Another equally important aspect ia to concentrate on many other
organic substances, e.g. proteins-, chlorophylls-, dyes, esters,etc., which are
eligible for monolayer formation and subsequent deposition "by using the
B-L technique. It is rather surprising that even in the recent investigations,
the workers have not paid -much attention to organic films other than those of
fatty acids or their salts. Presumably> only French workers have made a break-
through in this respect by studying orthophenanthroline molecules which were
deposited by E-L technique. It is interesting to point out specifically the
organic compotind "cholesterol" (C.-Hi -OK) which has been studied by some Russian
workers ' for "forming" and subsequent phenomena. In their studies the
monolayers or bilayers of cholesterol were obtained by a different technique
involving adsorption of polar molecules on a molecularly smooth surface of
liquid-mercury. However, these molecules, in principle, must also be obtainable
by B-L technique and efforts in this respect would be highly desirable. It has been
reported by the Russian workers (Ref.l61* and references therein) that electrical
studies on generally structured films of.surface active organic molecules are
of special significance to give information on the nature of phenomena occurring
in "biological membranes.
It has been noticed "by several workers that Langmuir films deposited on
gold base electrodes are shorting in nature and the electrical measurements are
not possible on such structures. On the other hand, with aluminium base
electrodes, the oxide layer formation does complicate the situation In the
interpretation of electrical data. There are some controversies in the existing
literature about the influence of oxide layer on the electrical properties of
Langmuir films. For instance, some workers have taken into account the oxide
layer whereas others have neglected it(considering its low resistance as compared
to that of the organic films. There are also few instances In which the
electrical data could be Interpreted in terms of the oxide layer presence in the
device. The reason for obtaining shorting devices with gold base electrodes
has been the poor adherence of molecules and hence the increasing porosity in
the organic films. Despite the fact that some alternative metals have to be
tried for the base electrodes to avoid oxide layer formation, the author
proposes to take advantage of the recently developed thermal evaporation
technique to obtain organic films on gold base electrodes. Comparison of
studies performed on Langmuir films and on evaporated organic films of the same
substances on Al and Au electrodes,respectively, might prove useful to
separate out the Influence of the oxide layer. The presence of water molecules
adsorbed during monolayer transfer in the B-L technique can also be avoided if
the renewed studies are performed on evaporated organic films. Fortunately*
-78-
Baker has pointed out that organic film could "be obtained "by thermal
evaporation down to one monolayer in thickness which, possessed more purity
than the original evaporant. Scrupulous attention however, must "be paid to
develop the thermal evaporation technique in future.
Despite the fact that Langmuir films have been reported to possess
structural perfection and could "be obtained free from even a small fraction of
holes and conducting impurities, it seems to "be an optimistic view to call
these films ideally suitable for electrical studies and device applications.
The problems which have hindered their use in the electronic device application
until now are the same as with many other films systems. For instance, problems
with respect to stability of the fatty acid films, which are soft, possessing
low melting points, and tend to collapse slowly with time. Although considerable .
efforts have already "been expended to improve the disposition technique and to
obtain stable films, in'actual practice, however, every new application
requires a lot of painstaking work before Langmuir films with reproducible and
stable electrical properties can be obtained. Nevertheless, some of the
possible applications are discussed here to throw light on the status and role
of Langmuir films mainly in the electronics industry.
One of the most attractive applications of Langmuir films may be in the
fabrication of "low loss capacitors" which find their use in active filters.
For the choice of suitable material to fabricate a capacitor, one needs to
consider many dielectric parameters,e.g. dielectric constant, dielectric strength,
dielectric loss, stability, temperature and voltage coefficients of capacitance,
surface smoothness and insulation resistance. In the case of fatty acid
films, several of these parameters are already known. For instance, the
dielectric constant of stearate is 2.1+ and films with capacitance densities
0.008 to 0.07 pF/cm have been obtained. These films could also have "been
deposited containing as many as 300 layers ( at 7500 A) and the capacitance
has been found to vary only slightly with the frequency. The dielectric
strength is a function of thickness and it is of the order of 10 Vcm ;
dielectric losses 2?0.02 to 0,006 have been reported. All these experimental -
values fulfill the optimum conditions to a greater extent for "low loss
capacitors" as has been reported by Harrop and Campbell . Fortunately,
many organic polymers and other organic materials have already been used for
this purpose.
Another possible application may be to develop tunnelling spacers between
superconductors, as was attempted by Miles and McMahon ; a long time ago.' These
workers failed to obtain useful tunnelling spacers simply because the device
obtained showed highly non-linear characteristics. In fact, the deposition
technique was not well developed in those days and one may hopefully proceed in the
light of recent advances. For the controversy regarding conduction behaviour
whether it is dominated by tunnelling or Schottky emission, the best test would
be to study inelastic electron tunnelling between two metal films separated
by organic insulating material. If it exists. This technique of studying
inelastic electron tunnelling was- introduced "By- Jaklevic and Lambe *
who distinguished it from the well-known tunnelling mechanism in which an electron
goes through the barrier without energy loss. Inelastic tunnelling results in
the energy transfer from electrons to the vibrational states of molecular species
contained within the barrier. It has been pointed out that for such studies
the insulating film between two metal films must be about 50 A thick which can
support a field of U x 10 vcm~ ., This may, however, introduce some difficult-o organic
ies because the dielectric strength of 50 A thick]film has been reported to be109), 121) h
little lower but these are likely to be overcome in future investigations
as the dielectric strength measurements vere made under atmospheric conditions
rather than under vacuum.
Since Langmuir films have been shown to undergo "forming" and to exhibit
voltage controlled negative resistance, some practical applications may be
discussed which make use of such characteristics. The best use may be in the
development of a thin film resistive memory arrays based upon voltage controlled
negative resistance in Langmuir films. Such a device was proposed by Hielson
and Eashara and vas described later by Simmons and Verderbei °* for
SiOx materials. Before making an attempt In this direction, one has to study
more carefully the memory switching of the diode and triode structures and
whether the information is stored in the form of permanent changes in the interior
of the film. Similarly, one can also think of some "hot"electron devices* In.
these devices, the basic requirement Is that of a metal-insulating film-metalso
structure in which the anode is sufficiently thin/that hot electrons traversing
the insulating film are able to escape into the vacuum above. It has been shown
by Gundlach and Kadi;c 13°J>131J tnat the phenomenon of DNB may be attributed
largely to the exists ice of filamentary conduction paths in these films. If the
proposed filamentary conduction does exist In these devices, this application
would be worthwhile to attempt. As has already been pointed out, if electro-
luminescence is also exhibited in these films many more applications might be
possible.
The application of Langmuir films in the form of metal-insulator-semi-152)
conductor devices is also possible,as Tanguy could successfully deposit
these films on silicon surfaces. In fact MOS capacitors and transistors are •
already in the development stage and therefore analogous MIS structures might
also be considered for such devices. Presumably, MIS capacitors and
transistors vould be more successful than the corresponding MOS ones because
the organic films may be obtained largely free from defects,which, in the case of •
oxides, IB rather difficult. The use of organic films in the form of MIS-80-
structures also opem up another practical application for poasiyated dielectric
films on silicon. Pt \yiner films hare already 136611 shown to possess favourable
properties for similar purposes °i. The only difficulty may "be due to the
hysteresis phenomenon, "but it must "be surmounta"ble "by appropriate treatment of the
silicon surface. It would also be advisable to explore the possibility of
achieving controlled doping of an organic dielectric in the way that was attained
with crystalline germanium and silicon. The'fact^that alternating monomolecular
layers of different organic materials may be deposited m e on top of the other,
offers an exciting possibility to fabricate some novel devices. One similar use
of fatty acid monolayers has "been in the form of distance keepers between dyeop \
molecules, while studying photographic sensitization mechanismsome uf the possible
Having discussed / applications of Langmuir films, it is essential to pointou
once again that although these films possessmany attractive features unlike those
of the evaporated film systems,nevertheless one must not be too optimistic and
expect all that has been said. In addition to these electronic device applications
one of the most successful uses of Langmuir films has been in the fabrication of a
fairly stable optical device called "step gauge" or thickness gauge which is
commercially available from General Electric Co., USA. Other optical applications
of these films are worth mentioning,e.g. antireflection gratings and interference
filters. From the purely scientific point of view, the resemblance of Langmuir
films to biological membranes suggests that a careful study of these films would
lead to the understanding of many biological or life processes. Some more159)industrial applications of these films are to achieve evaporation control and
to reduce friction , etc.
Since only a handful of electrical data is available on Langmuir films,
it would be unfair to make any committment regarding their practical applications.
A few proposals made above in this respect are in a way speculative until these
possibilities are explored further. Therefore, it seems to the author that
the potentiality of Langmuir films 3till lies beneath the hands of experimentalist!
and many-fold development of thin film technology would "be achieved if successive
efforts are continued. The application potential of these films is more likely
come up from a systematic study of many diversified electrical phenomena. Of
all the proposed applications, the development of capacitors is expected to
lead in the near future to electronic hardware for thin film microcircuitry.
-81-
ACKNOWLEDGMENTS
The author wishes to express his sincere gratitude to Professor V. Celli
for critical reading of the manuscript and for many helpful discussions. Thanks
are due to Professors A. Many, C,A. Neugebauer, N.H. March and G, Dearnley,
who were at the International Centre for Theoretical Physics during the Winter
College on Surface Science, for many "useful suggestions to improve the text
of the manuscript. The author is grateful to Dr. V.K. Srivastava, in partioular,
and his other colleagues, who (while the author was in India) initially inspired
him to write up the present manuscript. The preparation of the manuscript
vas initiated during the'author's participation in the Winter College on
Surface Sciences at the ICTP and completed during the following Workshop.
It is with great pleasure that the author expresses his thanks to the
ICTP staff for all the co-operation and help without which work on this
manuscript would have not progressed so smoothly. The author also wishes to
thank Professor Abdus Salam, the International Atomic Energy Agency and UNESCO
for hospitality at the International Centre for Theoretical Physics, Trieste.
The author would be failing in his duties without acknowledging the permission
granted by the respective authors and publishers of different research papers
{and journals) for the reproduction of the figures and tables in the present
review. The Swedish International Development Authority is gratefully
acknowledged for making possible his Associateship at the ICTP.
-82-
APPENDIX
NOTATIONS IN THE PAPER UNLESS OTHERWISE DEFBffiD IN THE TEXT
6 = dielectric constant
C a capacitance
d = monolayer thickness
M = no. of monolayers transferred
A = area of the film capacitor
n = number of carbon atoms in the molecular chain
I = current -
V » voltage
k « Boltzmann's constant -
T = absolute temperature
r a work functionf = frequency
J = current density
e = electronic charge
•R = Planck's constant
m = electronic mass
e Q B permittivity of free space
y - electron mobility
^ = potential barrier height
-83-
REFERENCES
1) Agarwal, V.K., Thin Solid Films (to be published)-
2) Lord Rayleigh, Phil Mag., 8, 321 (1899).
3) Adam, N.K. , The Physics and Chemistry of Surfaces (Oxford Univ. Press,
(Clarendon), London and New York 19^1) 3rd ed., Chap. 2.
h) Langmuir, IMTrans. Faraday Soc. 15_, 62 (1920).
5) Blodgett, K.B., J. Am, Chem. Soc. £7, 100T (1935).
6) Blodgett, K.B. and Langmuir, I., Phys. Rev. 51^ 9&k (1937).
7) Gains, G.L.,Jr., Insoluble Monolayers at Liquid-Gas Interfaces (Wiley
(interscience), New York 1966),
8) Srivastava, V.K., Physics of Thin Films (Academic Press* 197*0, Vol.?.
9) Harkins, W.D., Physical Chemistry of Surface Films (Van Hostrand-Reinhold,
Princeton, New Jersey,195k).
10) Adamson, A.W., The Physical Chemistry of Surfaces, (Wiley (interscience),
New York i960).
11) Alexander, A.E., Rep. Progr. Phys. 9_, 158 (19^2).
12) Popiel, W.J., Sci. Progr. (London) 58_, 237 (1970).
13) Ries, H.E., Jr., Encyclopaedia of Science and Technolopy, (McGraw-Hill,
New York 1971) 3rd ed.
lit ) Davies, T.J. and Rideal, E.K., Interfacial Phenomena [Academic Press,
New York 196l).
15 ) Mann, B. and Kuhn, H., J. Appl. Phys. !+2_, 398 (1971).
16) Srivastava, V.K. and Verma, A.R., Proc. Phys. Soc. (London) 8p_, 222(1962).
17) Sirhatta, R. and Scott, G.D. , Appl. Optics, 10, 2192 (197D-
18) Baker, M.A., Thin Solid Films 8_, R13 (1971).
19) Nathoo, M.H., Thin Solid Films l£, 215 (1973).
20) Langmuir, I., Proc. Roy. Soc. A 170, 1 (1939).
21) Tolansky, S., An Introduction to Interferometry, (Longmans,Green, London
195?) p. 65.22) Strong, J., Modern Physical Laboratory Practice (Blackie, London I9IJO).
23) Biicher, H., Eisner, O.V., M6bius, D., Tillmann, P. and Wiegand, J.,
Z. Phys. Chem. Neue Folge 65_, 152 (196P).
2k y Langmuir, I., J. Am. Chem. Soc. 3 ., i860 (1917).
25) Luff, P.P. and White, M., Vacuum l8_, 1*37 (1968).
26) Srivastava, V.K. and Verma, A.R., Solid State Comm., ii»367 (19^6).
27) Roberts, R.W. and Gains, G.L.,Jr., Trans. Ninth Natl. Vacuum Symp., Am.
Vacuum Soc, 19^2, pp. 515-
28) Gains, G.L.,Jr. and Roberts, R.W.,Nature 12J_, 787 (1963).
-Bk-
29) Holt, L., Hature 21^, 1105 (1967).
30) Henke, E.L., Advan. X-ray Anal. 2, h60 (196U).
31) Germer, L.H. and Storks, K.H., J. Chem. Fhys. 6_, 280 (1938).
32) Tanaka, K., Mem. Coll. Bel. Univ. Kyoto A22_, 377 (1939).
33) Brummage, K.G., Proc. Hoy. Soc. A191» 2^3 C19U7}.
3^) Covley, J.M., Trans. Faraday Soc. UU_, 60 (19I+8).
35) Menter, J.W. and Tabor, D., Proc. Roy. Soc. A20U, 51U (1950).
36) Trumit, H.J. and Schi&lovsky, G., Proc. Eur. Regional Conf.
Electron Microscopy i960, 2_, 721 (1961).
37) Knott, G., Schulman, J.H. and Wella, A.F., Proc, Roy. Soc. All§,
53U (19UO).
38) Marc, G. and Messier, J., J. Appl.Phys. (To be published).
39) Epstein, H.T., J, Phys. Colloid, Chenu ^ t 1053 (1950).
U0 ) Ries, H.E., Jr. and Kimfcall, W.A., J. Phya. Chem. 59_, 91* (1955).
lH) Ries, H.E., Jr., Sci. Amer. 20^, 152 (196I).
k2) Handy, R,M, and Scala, L.C., J. Electrochem. Soc. 113, 109 (1966).
1*3 ) Race, H.H, and Reynolds, S.I., J. Am. diem. Soc. 6l_, 1U25 (1939).
kh) Sototka, H., Demeny, M. and Chauley, J.D., J. Colloid Sci. 13_,565 (1958).
U5 ) Barraud, A., Messier, J., Rosillo, A. and Tanguy, J., Colloque AVISM,
Versailles (Sept. 1971).
k6) Porter, E.F. and Vyman,J., Jr., J. Am. Chem. Soc 6p_, 1083 (1938).
U7 ) Reynolds, S.I. and Race, H.H., G.E. Rev. Ul_, 529 (1938).
U8 ) Blodgett, K.B,, Use of Interference to Extinguish Reflection of Light
from Glass (unpublished).
U9 ) Buckwald, C.E., Gallagher, D.M,, Haakins, C.P., Thatcher, E.M. and
Zahl, F.Ai, Proc. Nat. Acad, Sci. 2U_, 20U (1938).
50) Zahl, P.A., Haskins, C.P., Gallagher, D.M. and Buckwald, C.E.,
Trans. Faraday, Soc. 35_, 308 (1939).
51) Goranson, R.W. and Zisman, W.A., J. Chem. Phys. J_, U92 (1939).
52 ) Beck, K.H., Ph. D. Thesis, University of Marburg,W. Germany, 1966
(unpublished).
53) Drexhage, K.H. and Kuhn, H.f Basic Problems in Thin Film Physics,
(Gottingen, Vandenhoeck and Ruprecht, 1966) pp. 339*
5U ) Havkes, P.L., Thin Film Integ.rated Circuits,(AEI Engineering) 5., 30U
(1965).
55 ) Leger, A.,KLein, J., Belin, M. and Defourneau, D., Thin Solid Films
8_, R51 (1971).
56 ) Khanna, U. and Srivastava, V.K., Thin Solid Films 12, s25 (1972).
57) Khanna, U. Srivastava, V.K. and Agarwal, V.K. Thin Films 2_, 83 (1971).
-85-
58) Khanna, U. and Srivastaya, V.K., Thin Films 2_, l6T (1972).
59) Khanna, V., Ph.. D. Thesis, University of Roorkee, India, 1973
(unpublished).
60) Chopra, K.L., J. Appl. Phys. 36, 655 (1963).
61) Ries.H.E,, Jr. and Walker, D.C., J. Colloid Sci. l6_, 36l (1961).
62) Eies,H.E., Jr. and Kimball, W,A. , Nattire l8l_, 901 (1958)-
63) Tanguy, J. and Hesto, P., Thin Solid Films 21_, 129 (197*0
6k) Miiller H., Phys. Rev. 5p_, S Y (1936).
65) Salem, L. , J. Chem. Phys. 3J_, 2100 (1962).
66) Denbigh, K.G., Trans. Faraday Soc. 36., 936 (l9kC).
67) Smyth, C.P,, Dielectric Constant and Molecular Structure.(Chem.
Catalog., Nev York 1931).
68) Frohlich,H.. Theory of Dielectrics (Oxford Univ., Press, 1958). 2nd Ed.
69) 0'Dvyer, J.J., The Theory of Electrical Conduction and Breakdown in
Solid Dielectrics, (Clarendon Press, Oxford 1973).
70) Fowler, R.H, and Nordheim, L., Proc. Roy. Soc. Allp, 173 (1928).
71) Frenkel, J., Phys. Rev. 36., 160^ (1930).
72) Soramerfeld, A. and Bethe, H., Hanbuch der Physik.XXIV, Pt. 2, Editor
S. Flugge (Julius Springer, Berlin 1933) p. 133.
73) Simmons, J.G., D.C. Conduction in Thin Films, (Mills and Boon Limited,
London 1971).
7h) Schottky, W., Physik Z. 1£, 872 (191*0.
75) Mott, N. and Gurney, R.W., Electronic.Proceases in Solids (Oxford, I9U8),
76) Rose, A., Phys. Rev. 21, 1538 (1955).
77) Frenkel, J., Tech. Phys. USSR 5., 685 (1938); Phys. Rev. £lj_, 6U7 (1938).
7?) Mead, C.A., Phys. Rev. 128, 2088 (1962).
79) Giaver, J., Phys. Rev. Letters 5_« (I960); 5_, h6k (i960).
8C) Nieol, J., Shapiro, S, and Smith, P., Phys. Rev. Letters 5_, i*6l (i960).
81) Miles, J.L. and McMahon, H.O., J. Appl. Phys. 3g_, 1176 (l96l).
82) Stratton, R., J. Phys. Chem, Solid 23_, 1177 (1962).
63) Simmons, J.G. , J. Appl. Phys. 3U.. 1793 (19^3); 3JJ., 258l (1963).
Bk ) Horiuchi, S., Yamaguchi, J. and Naito, K., J. ELectrochem. Soc. llg,
63^ (1968).
85) Simmons, J.G., J. Appl. Phys. 35_, 2^72 (196^).
86) Pollack, S.R. and Morris, C.E., J. Appl. Phys. 35_, 1503 {196k).
87) Flannery, V.E. and Pollack, S.R., J. Appl. Phys. 37, 17 (1966).
88) Mann, B., Kuhn, H. and Szentpely, L.v., Chem. Phys. Letters 8_, 82 (1971).
89) Szentpaly, L.v., Mtfbius, D. and Kuhn, H., J. Cherc. Phys. 52, W18 (1970).
90) Srivastava, V.K., Phys. Rev. Letters 30, 10U6 (1973).
-86-
91 ) Nathoo, M.E. and Jonscher, A.K., J. Phys. C ^ L301 (1971).
92 ) Hill, R.M., Phil. Mag; ;23_, 59 (1971).
93 ) Jonscher, A.K., J. Vac. Sei. Teehnol. 8_, 135 (1971).
91* ) Jonscher, A.K., Phil. Mag. g3_, 205 (1971).
95 ) Jonacher, A.K., J. Noncryst. Solids 8-10, 293 (1972).
96 ) Jonscher, A.K., and Hathoo, M.H., Thin Solid Films 1£, Sl5 (1972).
97) Jonscher, A.K., J. Electrochem. Soc, (to be published).
98 ) Le Sueur, E.J. and Jonscher, A.K., Thin Solid Films 12., 59 (1972).
99 ) Fock, Archiv f. EL. 19_, 71 (1927).
ICO) KLein, N. and "Qafni, H., IEEE Trans. Electron Devices, ED1.3, 28l (I966).
101) Klein, N., Thin Solid Films, X, ll»9 (l97l).
102) von Hippel, A., Naturwiss, 1^, 79 (1935).
103) Fr'ohlich, H.» Proc. Roy. Soc. Al6o, 230 (1937); A172, 9^ (1938).
IOU ) Forlani, F. and Minnaja, N., Phys. Stat. Sol. U, 311 (196U).
105) Forlani, F. and Minnaja» H,, Uuovo Cimento 31> 12h6 (196U).
106). Budenstein, P.P., Hayes, P.J., J. Appl.-Phys. 36., 2837 (1967),
107) Budenstein, P.P., Hayes, P.J., Staith, J.L. and Smith, W.B., J. Vac. Sci,
Technol.6_, 28? (1969).
108) Smith, J.L. and Budenstein, P.P., J. Appl. Phys. U0,3fc91 (196?).109) Agarval, V.K. and Srivastava, V.K., Thin Solid Films 8_, 377 (1971).
110) Agarwal, V.K. and Srivastava, V.K., Thin Solid Films, 13., S 23 (1972).
111) Forlani, 7. and Minnaja, H., J. Vac. Sci. Technol. £, 518 (1969).
112) Seitz, F., Phys. Rev. 7^, 1378 (19^9).
113) Frohlich, H, and Paranjape, B.V., Proc. Phys. Soc. (London) E69_, 866 (1956),
lilt) O'Dwyer, J.J., J. Phys. Chem. Solids 28_, 1137 (1967).
U5) O'Dwyer, J.J., J. Appl, Phys. p_, 3887 (1969).
116) KLein, N., Adv. Phys. 21., 605 (1972).
117) Watson, D.B., Hayes, W., Eao, K.C. and Calderwood, J.H., IEEE Trans
Electl. In&ulation 5., 58 (1970).
118) Shockley, W., Solid-St.Electron. 2_, 35 (l96l).
'119) O'Dwyer, J.J., The Theory of Dielectric Breakdown in Stilids (Oxford Univ.
Press, London 196U}.
120) Thiessen, P.A,, Beischer, D. and Gillhausen, H. Frhr .v. Nat-urwiss, 28_,
265 (19^0).
121) Agarwal, V.K. and Srivastava, V.K., Solid State Comm. 12_, 829 (1973).
122) Agarval, V.K. and Srivastava, V.K., Electrocomp., Sci. and Tech. (to "be
published).
123) Agarwal, V.K. and Srivastava, V.K., J. Appl. Phys. J^ 2900 (1973).
1210 Agarwal, D.K. and Srivastava, V.K., Thin Solid Films 1^, 367 (1972).
-87-
125) Agarwal, D.K. and Srivastaya, V.K., Solid State Camm. 11^ lk6l (1972).
126) Agarwal, V.K., Thin Solid Films 2g (1974).
127) Agarwal, V.K., Thin Solid Films 22 (1974).
128) Klein, N., Adv. Electron. Electron Fhys. 26_, 309 (1969).
129) Klein, N. and Levanon, K., J. Appl. Phys. 38_, 3721 (1967).
130) Gundlach, K.H. and Kadlec, J"., Phys. Status Solidi(a), ip_, 371 (1972).
131) Gundlach, K.H. and Kadlec, J., Thin Solid Films 13., 225 (1972).
132) Dearnley, G., Stonehame, A.M. and Morgan, D.V., Rep. Progr. Phya. 33_,
1129 (1970).
133) Verderber, R.R. Simmons, J.G. and Eales, B., Phil. Mag, l6_, 10^9 (1967).
13^) Verderber, R.R. and Simmons, J.G,, Radio Electron. Engr. 33_, 3^7 (1967).
135) Hickmott, T.W., J. Appl. Phya. £3, 2669 (1962).
136) Hictanott, T.W., J.'Appl. Phys. 35_, 2118 and 2679 U9Sk).
137) Barriac, C , Giraud-Hiraud, F. , Pinard, P. and Davoine, F., C.R. Acad.
Sel. Paris 262, 900 (1966).
138) Barriac, C. , Pinnard, P, and Davoine, F., ibid, 266_, 23 (1968).
139) Barriac, C , Pinnard, P. and Davoine, F., Phys. Status Solidi 3Ji,
621 (1969).
Dearnley, G. , Phys. Letters J5A, 7^0 (1967).
) Gitbons, J.F. and Beadle, W.E., Solid state Electron- J_, 785 (196M.
Sutherland, R.R. , J. Phys. D {Appl. Phys) U_, U68 (1971).
Simmona, J.G. and Verderber, S.B. Proc. Roy. Soc. A301, 77 (1967).
ikk ) Bernard, J., Hang, R., Mentalecheta, Y, and Tregouet, Y., Fhys. Statis Solidi
(a) 6, K127 (1971).
11*5) Curie, J., Compt. Rend. 132, 93° d a86).
Ih6) Warburg, E, and Tegetmeier, F., Ann, Physik 35_, 55 (1888).
1U7 ) Bucci, C. and Fieschi, R., Phys. Rev. Letters 12,,l6 (1963).
lU8 ) Joffe, A,, The Physics of CrystalsT (McGrall-Hill Book Co., Inc., New
York 1928). *
lU9) Dreyfas, R.WMPbvs. Rev. 121^ 1675 (1961).
150) Haven, Y. , J. Chem. Phys. 21 , 171 (1953).
151) Watkins, G.D., Phys. Rev. 113, 91 (1959).
152) Tanguy, J., Thin Solid Films 13., 33 (1972).
153) Snow, E.H., Grove, A.S., Deal, B.E, and Sah, G.T., J. Appl. Phys. 36.,
166k (1965).
15^) Chu, T.L., Szedon, J.R. and Lee, C.H., Solid State Electron. 10_,
897 (1967).
155) Hofstein, S.R., Solid State Electron, 10, 657 (1967).
156) Bui, A., Carchano, H. and Sanchez, D., Thin Solid Films 13, 207 (1972).
-88-
157) Goetzberger, A., Heine, V. and Nieollian, E.H., Appl. Fhys. Letters, 12.
95 (1968).
158) Dahlke, ¥.E. and Sze, S.M., Solid State Electron. 10, 865 (1967).
159) LaMer, V.K.. Retardation6f Evaporation by Monolayers,, (Academic Press,Hew York 1962).
160) Bucci, C , Fieschi, R. and Guidi, G., Phys. Rev. lMJ, 8l6 (1966).
161) Bucci, C.A. and Riva, S.C., J. Phys. Chem. Solids 26_, 363 (1965),
162) Servini, A. and Jonscher, A.K., Thin Solid Films 3_, 3^1 (1969).
163) Jonscher, A.K., Thin Solid Films 1, 213 (1967).
l6k) Babaev, A.S., Kuznetsova, V.V. and Stafeev, V.I., Sov. Phys. Semicond.
1, 692 (1973).
165) Harrop, P.J. and Campbell, D.S., Thin Solid Films 2_, 2T3 (1968).
166) Jaklevic, R.C. and Lambe, J., Phys. Rev. Letters 1T_» 1139 (1966).
167) Lambe, J. and Jaklevic, R.C., Phys. Rev, l6£, 821 (1968).
l6S) Nielsen, P.H. and Bashara, N.M., IEEE Trans. Electron. Bev.» ED-11, 2U3
169) Simmons, J.G. and Verderber, R.R., Radio Elect. Eng., 35_, 265 (1968).
170) Tabor, D, Surface Colloid Science (John Wiley, Inc., 1972), Vol.7.
-89-
7-;-
£>£:.,\*:**-*•*'.J * ' ' ' ' '
ta) tc) Id)
mi• COOH
CH3
COOH
CH3
,COOH
CH,x-Type film
(a)
1111CH,
CH,
»COOH
•COOH
CH-y-Type film
CH
• • • ecooH
•COOH
CH,
iCOOHz -Type film
CO
-90-
WE
HO
uUJQ.ra
UTo8
1.4
1.3
1.2
1.1
1.0
O.S
O.S
0.7
0.6
0.5
0.4
0.3
0.2
0.1
1 1
-
-
/*' 1 1
1 1
,6'1 ]
'I
1 i
/ r' /
/
i
sf
*
—
—
2 3 4 5 6 7 8 9 10 11Predicted Number of Layers, N p
-f,t. 3
100 1000 10400Frequency, cps
0.0100.000
- 9 1 -
tUJ
3.0
2.5
2.0
1.5
10
r~— A ! 1 (
j
t 1
T
{I
1
{
*—
_J
3.0
2.S
2.0
-11-9UJ
200 400 600 800 1000 1200 1400 1600 1800 2000
d( Angstroms)-
-92-
R
. 11
tan <5m.
[CH31n
—
103
CHgJgo'
-44dyne
- 2
soq ca++
s/cm
I
1kHz
A
I f
N«25
^ ^ 7
° N=13
e°c.100 -50 0 +50
-93-
10
7,5
4fa^Ai+ta^mj ^
1kHzH-44 dynes/cmN-25
2.5
-0.6
/ /Liquid Nitrogen^(Temperature
0.6
-9k-
10
10-7
10"
10*
l I
3 laytrs at 200 mV
5 layeis at320mV '
Slayers at220mV -
5layeri atlOOmV
10
10
10
10"7
40 40 - 8 0 -120 -160 -200 -240Temperature °C
f.t. 11
0.8
07
0.6
as
as
0.2
270
"\
-50 -75 -1001 t 1
W— Logfl^-j
V
L
\ \ /
-150 -175i 1
Degrees Centigrade _
ws 1 X
/ IF)"! -
1 1 ( 1 16 7 8 9 10
Temperature 1000/°K
1.61.51-41-31.21.11-0 e0.9;. L
0.70.60.50.4
ato
11 12
. H
V(
I, • Al NEGATIVE '
o Sn NEGATIVE /
AREA 4X10"Zcmz
10- 8
CO
10"
10.-11
0.5 1-0V(VOLTS)
(a)
1.5
10.-a
(0
10
10 *
^ J2(1.4v)
-o-l
It • Al NEGATIVE
1, O Sn NEGATIVE
I I I I I I » » • I2-0 3.0 3.5 4.0
(b)
I • i /V4.5 5.0
"ft* 13
-97-
o
(sdiuej/
-98-
ltf*
» Vv •a
\
10 10*1st co
8t (*)
12
10"*
10
• 105 KA 295 K• 338KX362 KX373 K
10 10
• -99-
HO
—9
Iog( i /E)
-10
AVW1
• 2X10° V/cm
-10
An X
(A) ,
-11
(
c
- 1 2 <
-13
- 1 4
>
I
\\ V!v*A o! O *O ^
Discharge
•1.8 106v/cmX1.35• 0.9*0-45+ 02 25
Chargefatl fields)
0
k N" X X> O H o \ \V N . B
XX %
+ X "v x^
% ^3 4
log t (s)
CO
O
V
-101-
-1.3 -1.2 -1.1
log. d microns
c.
-0.9
-102-
13
E1Z
10 -
cQ „
M41E
S 6
t i l l ) ! 1 * 1
7 -
1.00.90.5
0-7. E
0-5 *O
- 0.2 X
-0 -1 E0.0
220 230 240 250 260 270 280 290 300 310 320 330
T*mparatura(°K)
!•7
•5
j i
3.0
2.0
1.0
0,0
I 1 i 1 1 'Frequancy CH 3 0 K H 2
Y T A-J*'T**^ 0
- 1 1
, , , , ,
10 20 30 40 SO
d/25.75 ( *•- n**—rt} »
f[KH,)300
3*
>*
\\
1 •
o
A
N7 \
i-}•4 4 •*
J .E^.
\ :
oCO
\^\
vO
- IOU-
100
10 P
1k
C nf
~N
_ 3monoUy
-V
| |
op3 «n siliconP typ* eOOfilim.cm
IslkHi
Smonobyefs
15 monoiaytrt
k*Vwh
i i >
16 1 03 0 05 1 IS
f 16.33
-105-
100
nf 5 OPimonolayart on P Si SO ohm.cm
-V 3 1 05 0i i I i +v
»V
K)-fk
io-« h
10" -
-
-
4 me
- V1
\\ 1 monaloysrr
(V)
OP 1 on *!
/ '
/
1 1 I
15 1 05 0 0.5 1 1.8
35
-106-
•wd*
*p*tO mo,
2:Tp=287#K
2W 240 270 300 330.
KA)
Id'
1
Tp.293°KVp= IVolt
. tp = 10 mn0 < 0.16°K/ite
I N b s a
4210 240 270 300 330
-107-
• +
Oo
- i nfl-
Pig.
Fie.
1
2
3
FIGUKE CAPTIONS
Diagrammatic representation of the apparatus for •building up the
Langnruir films [modified after Langmuir I-
The molecular orientation of built-up Langmuir films, (a.) X-type;
(b) Y-type; and (c) Z-type.
Reciprocal capacity l/C vs. predicted number of layers W . The
heavy dashed line corresponds to the bulk value (£ = 2.5) and the
dotted lines represent the experimental values of £ = 2.1 and
U.2. The cluster of points shown in the figure fall on top of
each other and are slightly misplotted to the left or right to show
the density distribution of data. [After Handy and Scala. ']
Fig, h Capacity C vs. frequency f (upper curve) and capacity x
dissipation factor (CXD) vs. frequency f {lower curve). [After Handy
and Scala. ^]
Fig.. 5 Reciprocal capacity l/C vs. number of layers N for cadmium
stearate (n = 20) and cadmium behenate {n = 22). [After Mann and
Fig. 6. Dielectric constant & vs. film thickness d of barium stearate
film. (A) without oxide layer effect, (B) with oxide layer effect.
[After Khanna. ]
Fig. 7 Equivalent electrical circuit of a metal-Langmuir film-metal
structures. [After Marc and Messier. ]
Fig. 8. Loss angle of hehenate layers tan <Sm vs. temperature & measured
at f ' 1 KHz and at a surface pressure T of kk dynes/cm.
[After Marc and Messier. ]
Fig. 9 Comparison of dielectric losses with respect to temperature (a)
corresponding to a point in the middle of the behenate film, veil
organized structure; and (b) at a point near the edgei disorganized
structure. Data reported for N *• 25» "^ - kk dynes/cm and
f - 1 KHz. The lowest curve represents losses due to oxide layer,r 38) 1[After Marc and Messier. J
110 Current-voltage charactersitics at 0.1 Hz of Sn-3 layer calcium2
stearate-Sn sandwich of «lmm area at room and liquid nitrogen
temperatures. Polarity is that of organic coated tin "base layer.
[After Hanay and Scalal*2^]
-109-
Fig. 11 Plot of current vs. temperature at constant voltage for ** 3 layer
thick film (uppermost curve) and for ~ 5 layer thick film tthree
lower curves). Polarity of test voltage is organic coated "base
electrode positve. [After Handy and Scala.
,t \ / _ *__ \ _ J \ I -* XFig. 12 Plot of log (i/T ) vs. (l/T) and log
T2'vs. (l/T) for 3 layer
sandwich, where I - total current, I = low temperature current in
amperes and T = absolute temperature. [After Handy and Scala, 3
Fig. 13 (a) Plot of I and I vs. V at fixed temperature for Sn - 3 layer Ca-
stearate-Al junctions;and (b) plot of I and I vs. — at fixed '
voltage for Sn-Ca stearate junctions. Black dots and white circles
represent I2 and I , respectively. [After Horiuchi et al. ']
Fig, lit' jN vs. V for Al-Cfi stearate-Hg devices when Al electrode positively
biased. The dashed curve N = 5 to 21; J = ,}. (impurity dependent
current), proportional to V and l/N. Full curve N = 1;
i ~ <L + J • (J4. tunnelling current). j. surplus over j. is given
by dashed curve* [After Mann and Kuhn * J,
fjtd 1Fig. 15 Logarithm of a = \—r- V •+ 0 (o tunnell ing conductivity) vs. d1 * I, v j z •
for Al-Cd salt of CH {CH ) COOHH d i t 20 d 35°C3 2 n
n = 18 - 21. [After Mann and
I, v jfor Al-Cd sa l t of CH {CH ) COOH-Hg devices at 20 and -35°C with
3 2 n-<>
2Fig. 16 Plot of current vs. voltage for Al-cyanine-Fb device of area "*lmm
at a temperature of 1+.2°K (curve A).
Both the curves (A) and (B) are plotted to show the behaviour in
different voltage ranges. The appearance of the superconducting
lead gap Ap^ is marked in the curve (A).
Fig. IT Current-voltage characteristic for a 5 layer thick stearic acid film at
room temperature plotted as log I vs. V . [After Nathoo. ]
Fifi. 18 Recorder tracing of the current I after switching on a steady field
for Al-8 layers thick stearic acid film-Al. The insert is a
corresponding log I - log t plot giving a slope of -0.7- [After
Kathoo and Jonscher. ]
Fig. 19 The a.c. conductivity o(u)) vs. frequency u for 11 layer thick film2
with the specimen area ~* Jprm . The conductance scale refers to the
lowest temperature,»higher temperature characteristics being successively
displaced by one decade for clarity. [After Nathoo and Jonacher. 3
-110-
Fig. 20 Frequency domain measurements on two areas la) and (hi of the same
device, Al-5 layer stearic acid film-Al, for a range of electric fields
at room temperature. The dashed line for sample (b) represents
the "true a.c." data after deduction of the direct current. Data for
sample (b) are displaced downwards by two orders of magnitude for
clarity. The curves represent log-log plots of (i/E) vs. f where
i is the current per cm and E is the field in V/cm. [After96),
Jonscher and Nathoo. J•
Fig. 21 Time domain measurements of the discharge currents {symbols indicated)
and charge currents (open circles) for a different area of the same
film as in Fig. 20 , for different charging fields. Charge currents
have d.c. subtracted. The curves are log-log plots of I vs. t
(time). [After Jonscher and Kathoo. J
Fig. 22 Plot of log F vs. log d for Al-barium stearate-Hg devices. Curves
A and B corespond to the thickness range 25-250A (1-10 layers)
and 25O-2OOOA (10-80 layers),respectively. The bars represent the
scatter in the experimental data at a particular thickness and the dots
correspond to the mean values of several observations made on different
samples of the same thickness. [After Agarval and Srivastava. ^' ]
Fig. 23 V-J characteristic in the non-destructive phase for a UO-layer barium
stearate sandwich with area = 0.5cm at room temperature when the
series resistor (- 5K) between the source and device: was connected.
Point A represents the"onset breakdown voltage" and B the
"destructive breakdown voltage" of the device. [After Agarwal and
Srivastava}22h
Fig. 2k Transmission photomicrograph'showing the destruction of the2
film capacitor (area ~0.^cm ) of thickness *v 16 layer at 8^ volts
when the series resistor (= 5K) was used. [After Agarwal and
Srivastava. ]
Fig. 2$ Log-log plot of maximum breakdown strength F vs. film thickness d
for barium stearate films in the thickness range (Uooi - 1000A)
at room temperature. The dots and bars represent the same relationship
as in Fig. 22. . [After Agarwal and Srivastava. *]
Fig. 26. (A) Plot showing breakdown field F,_ vs. temperature T for1 Al-20• — • * • — • • • * U
layer barium stearate-Al sandwich in the temperature range -40 to +40°C.
(B) F^ VS. T for a similar specimen as in (A). :The Tjars and dotsb max
indicate the came relationship as in Fig. 22. [After Agarwal and
-111-
Fig. 27 Plot of a.c. breakdown voltage V . vs. film thickness d of
"barium stearate at fixed frequency " 30KHfc in the thickness range
100A - 1500A . The relationship between dots and bars is the same
as in Fig. 22 . [After Agarval and Srivastava. 1
Fig, 28 a.c.breakdown voltage V , vs. frequency f for "barium stearate film
of thickness 500A in the frequency region 10-200KHz. The relation-
ship between bars and dots is similar to that indicated in Fig. 22 .123)
[After Agarwal and Srivastava. ]
Fig. 29 Current-voltage characteristics of an Al-(Aloxide)- 11 monolayers-Al
sandwich for increasing temperature top electrode "being positive. The
vertical scale is displaced for each curve for clarity. DNR completely
disappears above about 210°K. [After Gundlach and Kadlec. ]
Fig. 30 Current-voltage characteristic of the device as in Fig. 29 at different
voltage cycles. Curve 1: first voltage cycle of an unformed sample.
Curve 2: second voltage cycle, the sample is now formed and exhibits
DWR. Curve 3: smooth characteristic after a few voltage cyclesj
sample temperature was 100°K. Curves h to l'< first four voltage cycles
after temperature increases from 100°K to 190°K. Curve 8: first
voltage cycle after interrupting junction bias for 3 hours, at 190°K.
Curve 9- smooth curve after some more voltage cycles at l°0°K. [After130). .
Gundlach and Kadlec. J
Fig. 31 Current-voltage characteristic and voltage distribution of an Al-9
raonolayer-Al device at 170°K, the plate being positive. [After131),
Gundlach and Kadlec. ]
Fig. 32 Current-voltage characteristic (full curve) and voltage distribution
of the device similar to, that in Fig- 31 at 230°K. The dotted curve is
J-V characteristic at 2O5°K. The spike 'a1 in the curves is probably
just a consequence of new events at 230°K, [After Gundlach and
Kadlec.131 >]
2Fig. 33 C-V curves for MIS structures (area - 125mm ) with 3, 5 and 9
monolayers of orthophenanthroline at a frequency of 1KHz. [After
Tanguy. ]
Fig. 3^ Shift of C-V curves with 5 OF- monolayers deposited on p-type
silicon as a function of temperature. The area of MIS device was2
10mm and measurements taken at f = 1KHz. Normal and anormal type152)
of hystereses are indicated. [After Tanguy. ]
-112-
Fig. ,35 Current voltage characteristics of MIS deyices with different
number of OP, . monolayers deposited on Si, Notice the shift ofI-V curves near zero volts indicating the existence of an e.m. f.
without external "bias. [After Tanguy. ^ ]
Fig. r36 T.S.C. peaks of a M2M device containing 23 monolayers of OP
at fixed V =» 1 volt, t = lOmn.and 3 " O.l6°K/sec. The peaksp p
observed 1 to 6 correspond to various polarization temperatures
T = 281, 287, 291*, 301, 308 and 315°K. [After Tanguy and Hesto,
Fig. 3T. T.S.C. peaks of a MIM device as a function of number of OP
monolayers at fixed T = 293°K, V = 1 volt, t = lOmn. andP P P
8 = 0.16 K/sec. d peaks consisting of several monoenergetic peaks
&-,» <L* **•*» k BJl^ *«; a r e 8^°'WI11 [After Tanguy and Hesto. ]
Fig. 38 Typical curve fittings with 2 superimposed peaks for 21 monolayers of
OP at temperatures Tm = 29O°K and Tm - 297°K, The solid lines
in the figure represent theoretical fits and the circles are
experimental points. [After Tanguy and Hesto."']
Fig. 39, (A) Change of capacitance AC/C vs. applied voltage V for different
number of Ba-stearate monolayers at temperature 77°K and area of
the device 3 0.25mm . (B) Corresponding plot of AC/C vs. square c
applied electric field ( ~ V 2 / N 2 ) . [After Leger et al. ]
TABLE CAPTIONS
TABLE I Electrical data, on unskeletonized (.0) and skeletonized GU films of
cadmium arachidate.' [After' Race and Reynolds.^' i
TABLE II Calculated and measured values of dielectric constant of a few long
chain fatty acid Langmuir films. [After Khanna and Srivastava. >
TABLE III Summary of the models to explain the "forming" process in insulators
[After Dearnley etal. 132^
-113-
TABLE I
Filmlamples
A
B
C
D
(1)
Opticalthickness
75
61
61
59
(2)
53
hi
51
51
(3)
<V*0>0.707 '
J
0.705
0.837
0.86
0-571* :
0.571
0.761*
0.797
(5)
£ 0
Measd.2.1*8
2 ^T
p i Q
2.1*1*
2.1*3
2.1*7
2.50
2.50
2.50
2.1*9
(6)
Measd.
1.69
1.7^
1.71
I.87
1.73
1.62
1.68
I.69
1.96
1.9U
2.02
1.97
2.10
(7)
Calcd.
1.85
1.82
2.15
2.18
(8)Filled wittnon-polar
Measd.
2.29
2.20
2.19
2.25
2.32
2.38
(9)1
oil
Calcd.
2.3^
2.31
2.U2
(10) (11)
Filled with tetra-chloiodiphenyl
€ 3 "£3Measd. Calcd.
3.01*
3.25
1
; 3.63
TABLE I I
(l) ( 2 ) ( 3 ) ( 4 ) ( 5 ) ( 6 )
s •
film substance
Ba-palmitate
Ba-margarate
Ba-stearate
Ba-behenate
monolayerthickness
23.25
2^.05
25.75
30.05
C : (from theory)cala.
for ,monolayers
1.65
1.63
1.67
1.50
multilayers
2.90
2.70
3-10
2.00
£ , (from experimental data whencalc. r
oxide layer effect was inclusive).
for monolayer
1.6T ± 0.12
1.65 ± 0.07
1.62 ± 0.08
1.60 + 0.10
far UO-layers
2.85 ± 0.20
2.59 ± 0.02
2.72 ± 0.07
2,25 ± 0.02
C , (when oxidelayei effect excluded).
for 1monolayer
2.29
2.09
2.10
2.01
formultilayers
2.88
2.77
2.63
2.27
•
t
Capacitance per unitarea ^ iF /cm 2
formonolayer
0.636
0.582
O.551*
0.1*70
formultilayer
0.00988
0,00866
0.0061*2
O.OO60I4
Author
HickmoH
Simmons-Verde rbcr
Dearnalev
Barriac tt al.
Greene el at.
Model proposed
Schottky ionization ofimpurities near middleof bandgap; formationof localized states forimpurity band and*pace charge TO aidinjection of electrons *
Injection of ions fromanode to form broadimpurity band with asharp top. Band bend-ing fr&m space charge
Propagation ofconductingfilament through
fusion of insulatorand electrode, ivithsubsequent ion injec-tion. Also traps formed"
High-field electrolysiswith injection ofvacancies from cathode
Locatorgeneralforming
Uncertain'
Not local
Local
Probably notlocal
Probably local
TABLE I I I
Dependenceon electrode
No detailed model;probably depends onwork function
Impurity levels depend onspecies; band bendingdepends also on workfunction
Forming initiated atanode and can beinhibited jf electrolytic-»lty released gas reactsreadily with anode
Not discussed
If the anode reacts readilywith gas releasedgrowth of insulator at.mode may inhibitforming
Dependenceon insulator
Kp=:i(energy gap);voltage-dependentsince ionization ofimpurities is so
Vr usually less than• (energy gap)'; dependson electrode, throughimpurity levels. Volt-age-dependent becauseimpurity spectrum andbandgap fix \'\-
Vf probably related tocrystal formationenergy or to breakdownstrength. Voltage-dependent becauseforming initiated in
' «mall region - •
Not discussed
free energy offormation of insulator
Dependenceon atmosphere
No detailed model;probably impurity-oxygen reaction
Apparently none since iheinjtcted ions do notreact strongly withoxygen
Oxygen causes competingreactions which preventfilament propagalion orinitiation
Oxygen neutralizes thetons which constitutethe space charge
Oxygen causes other com-peting reactions atcathode
. ft -i