CH 1

‫ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬
ACKNOWLEDGEMENT
The author would like to experess his sincere gratitude to Professor

Dr.A.A.Ammar, formerly Dean of the Faculty of Science, Menufia
University and Professor Dr,M.M.EL-Zaiadia, Dean of the faculty of
Science, Menoufia Univercity, for suggesting this point of research and
for their supervision, fruitful discussions and assistance throughout this
work.
The author wishes to experess his thanks to Dr.Z.I. EL-Badawy for

the valuable assistance during this work.
The author wishes thanks all collegians in the physics department

and all the staff of the Central Research Laboratory, Faculty of Science,
Menoufia University, for their kind cooperation.
‫( ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬1) ‫ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬
CHAPTER (I):
GENERAL I NTRODUCTION:
1-l : INTRODUCTION.
1-2: THE AMORPHOUS WORLD.
CHAPTER (II):
THEORETICAL INTRODUCTION TO THE BAND
STRUCTURE OF AMORPHOUS SEMICONDUCTORS
2-1: THE FAILURE OF' THE BAND THEORY.
2-2: THE STRUCTURAL MODELS.
2-3: LOCAL IZATION.
2-4: BAND MODELS OF AMORPHOUS SEMICONDUC-
TORS.
CHAPTER (III):
EL ECTRICAL TRANSPORT IN AMORPHOUS
SEMICONDUCTORS:
3-1: DIRECT CURRENT CONDUCTION.
3-2: AC. CONDUCTION.
CHAPTER (IV):
SWITCH I NG AND MEMORY PHENOMENA:
4-1: THEORETICAL MODELS FOR SWITCHING AND
MEMORY PHENOMENA.
4-2: FACTORS AFFECTING ON SWITCHING AND
MEMORY PROCESS.
CHAPTER (V):
LITERATURE REVIEW OF PREVIOUS WORK ON
CHALCOGEIDES:
5-1: FILAMENT FORMATION AND ITS
TEMPERATURE.
5-2: SOME PREVIOUS WORK ON CHALCOGENIDES.
5-3: AIM OF THE WORK.
CHAPTER (VI):
EXPERIMENTAL TECHNIQUES OF MEASURE-
MENTS:
6-1: PREPARATION OF AMORPHOUS Ge-Te-Se
SYSTEM.
6-2: D.C. CONDUCTIVITY MEASUREMENTS.
6-3: A.C. CONDUCTIVITY MEASUREMENTS.
6-4: D.T.A. TECHNIQUE.
CHAPTER (VII):
RESULTS AND DISCUSSION:
7-1: RESULTS ON I-V CHARACTERISTICS.
7-1. 1: I-V Characteristics.
7-1. 2: Temperature Effect on the I-V trace.
7-1. 3: Cycling Effect.
7-1. 4: Thickness Effect on the I-V Characteristics.
7-2: PROPOSED MODEL TO ESTIMATE THE
FILAMENT TEMPERATURE.
7-3: DIPOLE MODEL.
7-4: Uniaxial Pressure Effect on The I-V Characteristics.
7-5: D.C. Conductivity at Constant Current.
7-6: GALCULATION OF THE FILAMENT TEMPERA-
TURE.
7-7: A. C. CONDUCTIVITY
CONCLUSION
REFERENCES
ABSTRACT
Amorphous semiconductor draw the attentoin of many researchers

in the field of solid state due to their promising and proved successful
uses recent technological applications.
In this thesis the system Ge Se Te was chosen. Nine ingots with

different compostion were prepared by quenching the melt kept in a
sealed silica tube. The chemical composition of the nine glasses are
(1) [Ge 24 Te 76 x Sex ] where x = 8.11.14
(2) [Ge 21 Te 79 y Se ] where y = 11.14.17
y
(3) [Ge 18 Te 82 z Se ] where z = 14.17.18

z
After quenching few samples were selected from each ingot in glass
transition temperature crystallization temperature melting point … ect
X-ray studies proved that all prepared samples are in the amorphous
state. Using the differential analysis technique reveal 4important
temperatures during heating each sample at a constant rate these are Tg,
Tc1, Tc2 and Tm. These values were found to depend on the composition.
Also activation energies of crystallization were calculated for the nine
samples.
As a preliminary test the I-V curves of 9 samples belonging to the

nine glassy material were measured at room temperature. This test was
necessary to prove that all samples of the system possess differential
negative resistance state and possess switching with a memory
characteristics. All experimental data of I-V curve are found to obey an
empirical relation of the form
V  eCI e I where C  Vth / I th and   1 / I th
are the values of voltange , current and resistance at the

Vth , I th , Rth
point V , I on the I-V curve and will be denoted by "turn over point"
th th
(T.O.P). At this point, the material is converted from the off to the N.R.
state also the results are found to fit another equation
 1  sin  
V  e CI  e CI   . This equation which fits the results is based on
 cos  
the assumption saying that the I-V behavior is a result of interaction

between the applied electric fild and the dipoles present in the material,
 is the angle between the axis of the dipole and the applied electrice
field. It is important to note that
a) The turn over point for all samples measured at any condition is
reached at a constant value of  equal 49.6 . 
b) The resistance of any material when measured at I  0 equal to
(2.718) times the value of the resistance. Rth at T.O.P. Both
equations are identical, as the term e  is found to be equal to  . It I
should be noted that both equations fit results up to a current value

I = 1.25 Ith. The deviation arises from the fact that the Joule heating
at this value of I will cause a change in the crvstal structure of the
material.
Several tests were performed to study the effect of
a) Thickness
b) Ambient temperature
c) Pressure
d) Cycling on the I-V characteristic
Results showed that both Vth and Ith increases on increasing the
thickness of the sample. Increasing ambient temperature T is found to
move the T.O.P. to the up left direction in the I-V plot. Thus T causes a
decrease in Vth while it results in an increase in the Ith. The activation
energies of variation of Vth, Ith and Rth were deduced from the
experimental results. It was suggested that the temperature of the
conduction path (Tf) should be higher than ambient temperature (T ) due
E T
to Joule heating. A general empirical relation Tf 
E  K T  I was
formulated to calculate Tf at any point along the I-Vcureve. At T.O.P,
E T
the last equation reduces to Tf 
E  K T . This equation is nearly similar
to another equation based on the thermal transport equation
Tf was found to depend on
a) Values of Ta
I
b) Values of In 
Ith
, and
c) Activation energy values of conduction
Tf is found to increarly with both T & In but it increases
exponentially with decrease in E. i.e Tf increases sharply with decrease
in E more than its increase with T . The dependence of Tf on I is less
compared to changes in T and E.
The conductivity of the nine samples were also measured as a

function of ambient temperature while keeping the passing the passing
current constant. The range of constant values of current ranges from
2  10 Amp. to 5  10 Amp. Often the relation between In and 1/T could
6 6
be divided into three parts, separated at constant kink temperatures each

part denotes a certain conduction mwchanism.
Several models were used to explain results. The most successful

mechanisms were nearest neighbor hopping mechanism (the range at
low temperature) mechanism and the thermally assisted hopping
mechanism in the middle temperature range. The third one at the higher
temperature range is thought to represent the intrinsic behavior. The
variations i n activation energies and the pre-exponential factor were
used to represent a formule for the temperature Tf of the conduction
path.
The uninxial pressure was found to alter the parameters of the I-V
curve. For example increasing P will cause a decrease in Vth. It is
assumed that P causes a decrease in the activation energy values by a
decrement  . The change  .  dE / dP was calculated for all samples.
The effect of cycling (repeating I-V tests) was also studied. The I-V
characteristic becomes reproducible after performing about 15 tests on
each sample. This experiment should be performed before any test. The
cycling effect (hysteresis) may prove the importance of heat treatment
while applying an electric field.
Above results enables one to understand the mechanisms

dominating in the three states of the material (OFF state, N.P.S and ON
state). The Joule heating and values of applied electric field are the main
factors which control the I-V parameters. At low values of field the I-V
becomes of ohmic nature at all ambient temperature.
It was also found that both ambient temperature and the values of
safety resistance which is always connected with the sample in series,
control the N.R.S range. The latter factor limits the current passing in
the circuit.
It was also possible to calculate from the I-V data the value of load
resistance.
At the end of this work, results on the A.C conductivity of the nine
samples are presented. Measurements tack place in the temperature
range (300-420) Ko and in the frequency range (100Hz-500kHz) using
Tesla BM 507. Results enabled us to suggest equivalent circuit for all
samples. All samples except 6 are represented by a resistance R1
connected in parallel with a capacitor C. The equivalent circuit of
sample 6 consists of a resistance R2 connected to the above circuit. The
presence of R2 which has a value comparable to R1 prevent, this materiai
from converting to a (N.R.S) state. A.C results at different temperature
allowed the calculation of both E &  . Data revealed that the value of
the optical forbidden energy gab is 4 times the activation energy
calculated from the relation between in  and 1/T. This result point to p
the presence of intimate valence alternating pairs (IVAP) which are

necessary for the dominating of the both assisted hopping and nearest
‫‪neighboring sites hopping mechanism. This result agree with the result‬‬
‫‪drawn from the D.C conductivity experiment.‬‬
‫ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ )‪ (9‬ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬
‫)‪CHAPTER (I‬‬
‫‪GENERAL I NTRODUCTION‬‬
‫‪1-l‬‬ ‫‪: INTRODUCTION‬‬

‫‪1-2‬‬ ‫‪: THE AMORPHOUS WORLD‬‬
‫‪1-2.1‬‬ ‫‪: Disorder of amorphous meterias.‬‬
‫‪1-2.2‬‬ ‫‪: Glass transition.‬‬
‫‪1-2.3‬‬ ‫‪: Preparation of amorphous materials.‬‬
‫‪1-2.3(a). : Thermal Evaporation.‬‬
‫‪1-2.3(b). : Sputtering.‬‬
‫‪1-2.3(c). : Glow-discharge decomposition.‬‬
‫‪1-2.3(d). : Chemical vapor deposition.‬‬
‫‪1-2.3(e). : Melt quenching.‬‬
‫‪1-2.3(f). : Electrolytic deposition.‬‬
‫‪1-2.4‬‬ ‫‪: Structure of amorphous sendcanductors.‬‬
‫‪1-2.4(a). : Introduction.‬‬
‫ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ)‪ (10‬ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬
1-2.4(b). : Kinetics of crystallization in amorphous semiconductors.
1-1: INTRODUCTION
The amorphous solids have an intense interest, makes them a

particularly active field in solid state physics, this is because of the
technological importance of their uses in industry. A particular interest
taken in chalcogenide semiconductor glasses for their unique properties
such as switchine and memory effect. S. R. Ovshinsky have always been
defined by the predominant, materials: The Stone Age. Actually, we are
already in the transition period between the age of order and the age of
disorder.
Glass which was previously considered as an optical, dielectric or

passivating materials, can now be used to ceeate devices-switches,
memories, solar cells, catalysts, etc.
Ovshinsky have predicted that "by the year 2004, most of the
alternative energy sources involving materials which convert light heat,
‫( ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬11)‫ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬
or chemical energy into electricity will be made of amorphous or
disordered materials".
Amorphous materials not only can be designed to be very stable, but

they can also be tailored to exhibit structural changes from stable bond
switching to reversible crystallization. The latter include materials which
are the basis of information memories. Their design depends upon an
understanding of how the energy barriers to crystallization can be
overcome by optical, electrical or thermal excitation [2, 3, 4].
The most important, elemental constituents of amorphous

semiconductors are group [IV], (Si, Ge…), group [V], (P, As, Sb,…)
and chalcogenides, group [VI], (Se. S, Te,…) However materials
prepared in the amorphous form by more than one technique often do
not have the same properties.
1-2. THE AMORPHOUS WORLD
Amorphous (or noncrystalline) materials such as glass are not

new and man has been manufacturing glassy materials for thousands of
years. The new is the scientific study of these materials. There has been
an explosion of interest recently as more new materials are produced in
an amorphous form with considerable technological promise.
Amorphous materials possess some degree of structural disorder.

This disorder is not a unique propmrty. It in must be compared to some
standard, and that standard is the perfect crystal. We can define the
perfect crystal as "that in which the atoms (or groups of atoms or
'motifs') are arranged in a pattern that repeats itself periodicaity in three
dimensions to an infinite evfprzt". With this definition, an imperfect
crystal can be simply be one which is finite or one which possesses a
defect. But the defect (or disorder) in amorphous materials is more
complicated than that in an imperfect crystal.
1-2.1: Disorder of amorphous materials.
1-2.1(a). Geornetric disorder is that form of randomness in which three

is no periodicity. There are degrees of geometrical disorder, certain
amorphous materials have considerable Short.-range (or local) order
while others have little, both have no long-range order. So, all
amorphous or glassy solids are distringuished by their lack of
periodicity.
1-2.1(b). Mapnetic disorder in which the atomic site possesses a spin

or magnetic moment which is oriented randomly.
1-2.1(c). Substitutional disorder in which the material with one type of

atom randomly substituting for the other in the lattice.
1-2.1(d).Vibrational disorder in which the random motion of' atoms
about their equilibrium positions destroys the periodicity.
1-2.2: Glass transition
When a liquid is cooled, one of two events may occur. Either

crystallization may take place at the melting point(Tm), or the liquid will
become super cooled for temperature below (Tm) and may form a glass.
These changes can be observed by plotting the (volume or entropy or
enthalpy) as a function of temperature.
Considering the variation of molar volume with temperature the

crystallization process is characterized by an abrupt change in volume at
(Tm) whereas glass formation is characterized by a gradual break in
slope. The region aver which the change of slope occursis termed the
glass-transition temperature (Tg). It is found that the glass-transition
temperature of a particular material depends on its thermal history.
According to the discontinuity of  T (Thermal expansion), KT
(Compressibility) and CP (heat capacity) at the glass-transition
temperature which suggests that this transition is a second-order phase
transition. Haggert [5] has demonstrated that for a number of glasses, CP
reaches the limiting value of 3R at. Tg. (The glasses confirm this
observation). Near the glass-transition temperature the probability of
crystal growth and nucleation increases very rapidly thus (for very long
experimental time-scales) crystallization is more probable than the
relaxation to a lower energy state (fully amorphous state).
1-2.3: Preparation of amorphous materialas
There are at least a dozen different techniquescan be used to prepare

materials in an amorphous state. The different techniques use all the
three phases of matter, vapor, liquid and solid, as the starting materials
in producing amorphous solids.
1-2.3(a). Thermal Evaporation.
This technique is the most widely used method for producing

amorphous thin films. In this technique the starting compound is
vaporized in a vacuum chamber and the material is collected on a
substrate. This method is suitable for relatively low melting-point
compounds. Those materials with higher melting points are best
vaporized by bombardment with high-energy electrons from an electron
gun placed in the chamber.
The principal disadvantage of thermal evaporation as a preparative

technique lies in the variability in purity and composition of the
resulting films. Many factors combine to produce this variability such
as:
a) Substrate temperature.
b) Distance between source and subst.rate and orientation.
c) The Base pressure in the evaporation boat.
e) Boat temperature.
Some of these parameters can be controlled, others cannot be

controlled. So a change from one preparation to another would be
expected for the same material. This was the principal reason why so
many experimental data were apparently contradictory in the early days
of the subject.
1-2.3(b). Sputtering.
The sputtering process is rather more complicated than thermal

evaporation but enjoys the advantage of being far more flexible. It
consists of the bombardment of a target by energetic ions from low-
pressure plasma causing evaporizatian of material, either atom by atom
or as clusters of atoms, which deposit on a substance. The simplest way
to induce sputtering is to apply a high negative voltage to the target
surface to attract the positive ions from the plasma.
1-2.3(c). Glow-discharge decomposition.
Another technique which can produce thin films of amorphous

materials is glow-discharge decomposition in the vapor phase. This
method have been used after the discovry by Spear and Le Comber [6]
certain amorphous semiconductors prepared in this manner can be
substitutionally doped to control their electronic properties.
1-2.3(d). Chemical vapor deposition.
This method is analogous to the glow discharge method in both

depend on the de-composition of a vapor species. Difference is that the
chemical vapor deposition process relies on thermal energy for the
decomposition and the applied r.f. field (if used) simply serves to heat
up the substrate upon which the vapor decomposes. This method is often
used to produce polycrystalline material but under the appropriate
conditions a true amorphous materials can be deposited.
1-2.3(e). Melt quenching.
The oldest, method of producing an amorphotts solid is to cool the

rnolten form of the material sufficiently quickly. So the most important
factor is that the cooling be sufficiently fast to prevent the crystal
nucleation because the crystalline phase is thermodynamically more
stable and crystal growth will always dominate over the formation of the
amorphous phase if allowed to take place. Then a glass is simply a super
cooled liquid in which no time for crystal growth has been allowed.
The most usual way of producing samples is to seal a charge (1-10

gm) in a fused silica ampoule under a good vacuum (10-6 Torr), and keep
the ampoule in a rocking or rotating oven at a sufficiently elevated
temperature that the constituents become molten and can react, the
rocking motion ensures that a mixing of the mixture takes place. The
melt can then be quenched, either slowly by simply switching off the
oven, or more rapidly by bringing the ampoule into the air, or more
rapidly by plunging the ampoule immediately from the oven into a
liquid (a liquid with a high thermal conductivity and high latent heat of
vaporization so that heat is conducted away from the sample as fast as
possible without the formation of a thermally insulating layer around the
ampoule). In this method a cooling rates of the order (102-103 Ko/sec) are
achievable and are sufficient for most purposes. Some pararmeters
which are important in this method are the following:
a) Temperature of the oven.

b) Cooling Rate.
c) Volume of charge in ampoule.
d) Thickness of the wall of the ampoule.
For certain materials even the fastest cooling rates achieved by this
method are insufficient to prevent crystallization, and still faster cooling
methods are required. Such methods have been discovered by Duwez [7]
who was the first to demonstrate cooling rates in the order of 106 Ko/sec.
The technique relies on the rapid cooling afforded by a "chill-block" in
intimate contact with a thin film of the liquid, in which the heat is
continually transported away from the liquid in contact with the "chill-
block".
1-2.3(f ). Electrolytic deposition.
Another technique which uses solutions of the desired material is

electrolytic deposition. In this manner, for example, amorphous Ge films
can be deposited on a Cu cathode by the electrolysis of GeCl4 in be
glycol. A thick film ~ 30  m can be prepared in this way, but
contamination by the solution and reaction products is a problem.
1-2.4: Structure of amorphous semiconductors.

1-2.4(a). Introduction.
Amorphous solids lack the order structure of a crystal. They have an
atomic structure with no long-range periodicity. Despite the absence of
long-range order, there still exists a significant amount of short-range
order determined by chemical bonding in the structure. A great deal of
work has been done to investigate the nature of the shortrange order in
periodicity. Thus k (the wave vector) is no longer conserved in optical
transitions. This has an iniportant beneficial advantage for solar cell
technology, since all amorphous semiconductors are strong absorbers of
photons with an energy above the optical gap, this is why a 1
micrometer amorphous silicon film absorbs as much Visible light as a
50  m sample of crystalline silicon.
As a result of long-range periodicity in crystalline solids the density

of electronic states is characterized by sharp band edges. Introducing
disorder, the periodic potential will be disturbed, the scattering will be
enhanced and the mean free path of the carrier reduced, so sharp band
edges could not exist. Theoretical work [8,9] suggests that the band
structures of amorphous semiconductor is not very different to that in
similar crystalline materials except for the splitting of a tail of localized
states below the conduction band and above the valence band.
Fritzsche [10], measured the differential thermal analysis curve of

such memory glass and come to the conclusion that the current channel
in the conducting state gets warm enough to reach the crystallization
temperature, transforming the channel region into a well conducting
degenerate semiconductor (or semimetal).
The short reset pulse would then bring the crystallized channel
quickly above melting temperature, after which the channel region is
quickly cooled by its environment, establishing by quenching the
origjnal glass state.
1-2.4(b). Kinetics of crystallization in amorphous semiconductors.
The memory effect observed in glass semiconductors is usually

explained on the basis of thermally induced structural transformations in
glasses[11, 12] Thermal inducwd transformations in chalcogenide
glasses have been the subject of many studies [13, 14, 15]. Any change
in the phase constitution of an assembly must be a natural process,
leading to an increase in the entropy of the universe, and to
corresponding changes in the thermodynamic functions of the closed
assembly.
The kinetics of crystallization have been determined from DTA
measurements as a function of heating rate [16]. The crystallization
process of amorphous state occurs by the nucleation and then they are
growing by a certain rate of crystallization. If the crystallization rate
varies with temperature; i.e. prossesses an activation energy, the position
of the peak in DTA thermograms varies with the heating rate if other
experimental conditions are maintained fixed. Kissinger [16, 17] used
this variation in the peak positions during DTA for determining the
activation energy. Also he determined the order of reaction from the
shape of the DTA peak. This technique have been used by several
workers for studying crystallization process on chalcogenide glasses
[14.18, 19, 20]. The rate of crystallization may be expressed as"
df
 C (1  f )
dt
where f is the function of crystallized material and C the rate constant

given by the Arrhenius relation:
C  Co exp ( E / KT )
where E is the activation energy for crystallization and

 Co is the
frequency factor.
If the temperature rises during the transformation the reaction rate,

df/dt, will rise to a maximum value at the peak of crystallization To. At
the miximum rate we have:
d2 f
0 (13)
dt 2
This condition gives:

E 
Co exp ( E / K TC )  (14)
k TC2
In which  s the constant heating rate. By plotting in ( / TC2 )VS .1 / T one

can obtain a straight line from which Co and E can be determined. 
‫‪The reaction order n can be calculated from the DTA curve by‬‬
‫‪measuring the shape index of DTA peak and using the expression:[29].‬‬
‫‪n  1.26 (a / b)1 / 2‬‬ ‫)‪(15‬‬
‫‪where a and b are represented in Fig. (1).‬‬
‫)‪Fig. (1‬‬
‫‪Method for measuring amount of symmetry in D.T.A peak‬‬
‫)‪CHAPTER (II‬‬
‫‪THEORETICAL INTRODUCTION TO THE BAND‬‬

‫‪STRUCTURE OF AMORPHOUS SEMICONDUCTORS‬‬
‫‪CHAPTER (II):‬‬
‫‪2-1: THE FAILURE OF' THE BAND THEORY.‬‬
‫‪2-2: THE STRUCTURAL MODELS.‬‬
‫‪2-3: LOCAL IZATION.‬‬
‫‪2-4: BAND MODELS OF AMORPHOUS SEMICONDUCTORS.‬‬
‫‪2-4.1 : Mott-CFO model.‬‬
‫‪2-4.2 : Davis – Mott Model.‬‬
‫‪2-4.3 : Marshall-Owen Model.‬‬
‫‪2-4.4 : Small polaron Model.‬‬
‫‪2-4.5 : Licciardello Model.‬‬
‫‪2-4.6 : Ovshinsky and Adler Model.‬‬
2-1: THE FAILURE OF THE BAND THEORY.
From the study of the Band theory and its application to solid state
physics an understanding of why certain crystals are metals and others
are insulators have been achieved. The presence of the perfect
periodicity greatly simplifies the mathematical treatment of the behavior
of elements in a solid. In the case of amorphous materials there is no
periodicity and hence there can be no reciprocal space and electrons
cannot be represented as "Bloch waves". The quantity that is equally
valid as a description of electron states for both crystalline and
amorphous solids is the density of states.
Weaire and Thorpe [21] first showed that if short-rang

interactions between electrons are dominant, then it is the short-rang
order which mainly determines the electronic density of states. They
showed using a simple model Hamiltonian for the electron interactions
that a gap is expected for an ideal tetrahedrally coordinated amorphous
solid. The amorphous structure is taken to be that in which each atom is
in a perfect tetrahedral environment with a wide distribution of dihedral
angles which is necessary to generate a random network. The two
interactions considered in the model are the bending interaction V1,
responsible for the width of the bands, and the bonding interaction V2,
responsible for the separation of bonding and antibonding bands.
It has also been found that [21], if V2  2V1 , there is no overlap

between the bands, so a true band gap must exist. Using the Weaire-
Thorpe Hamiltonian a good agreerment is seen to be qualitative for the
valence band, but very poor for the conduction band. A considerable
effort has been expanded in attempts to improve on this rnethod and to
obtain more realistic estimates for the density of states of amorphous
solids. The improvements can be made in two areas:
a) A more realistic Hamiltonian can be used involving more interactions
than just the two inter and intra site terms considered by Weaire and
Thorpe (V2 and V1 respectively).
b) A more realistic structural model than used by Weaire and Thorpe
was introduced (only short - range interactions were included). Thus
the atomic coordinates from a structural model which fits
experimental data such as continuous random network (CRN) or a
dense random packed (DRP) model, are better starting point.
2-2: THE STRUCTURAL MODELS.
In the Continuous Random Network (GRN) the term "random

network" is some what of a misnomer in that such network are not truly
random in a statistical sense, but have a considerable degree of local
order conferred upon the presence of directed bonding. Thus, covalently
bonded amorphous solids generally have a well-defined bond length,
bond angles and nearest-neighbor coordination, often very similar to
those found in corresponding crystalline polymorphs. The fundamental
polyhedra formed by the nearest neighbors are connected together in the
network, and their relative orientation is characterized by the dihedral
angle. If the units were completely randomly arranged the dihedral angle
distribution would be flat and it is found for most CRNs that although all
values of dihedral angle are represented, certain values are more
probable than others. Thus, although a CRN may not be truly random, it
is certainly noncrystalline, and this is achieved by allowing some degree
of bond-angle distortion and more neywork models can be built in this
manner using values for the bond lengths and angles taken from
experimental Radial distribution Functions (RDFs).There are two ways
of constructing structural models.
The first involves the physical building of a model, which for a

covalently bonded material consists of a "ball-and-stick" arrangement, in
which tubes of the appropriate length represent covalent bonds, and
atomic sites are represented by either balls with holds drilled at the
appropriate orientations of units with spokes at the correct relative
angles is arranged. (DRP) Dense Random Packed models can simply be
constructed by mixing balls of the same (or different) diameter and
gluing the resulting non-periodic assembly to form the model.
The second way of constructing structural models is by using a

computer and this technique in itself can take several forms. An
alternativa way of using computational methods to generate structural
models is the modification of existing models usually by altering the
energy of the structure by appropriate algorithms.
From above we conclude that the problem is to obtain the electron

density of state g(E) and the average mobility  (E ) . For amorphous
systems.
2-3: LOCALIZATION.
The presence of disorder manifests itself on the electron states in a

variety of ways. Fluctuations in short-range order, such as bond-angle
distortions, lead to tailing of states into the gap at the band edges. Band
tails are even more pronounced if chemical disorder, is present. Another
cause of band tailing can be the occurrence of short-range disorder in
nun-bonded atom distances. This is particularly important for chalcogen
atoms, which have one non-bonding p-like orbital (forming the top of
the valence band), in addition to two predominantly p-like bonding
orbital per atom. If two such atoms are forced close to each other in the
random network the interaction bet.weFan the non-bonding lone-pair
orbitals causes the energy of such states from t,hE~ valence band edge
into the gap. The result of all these effects gives rise in the extreme case
to the Cuhen-Fritzsche- Ovshinsky [22] model for the density of state in
the gap region of an amorphous chalcogenide semiconductor. In this, the
band tailing is so pronounced that the tails arising from the valence and
conduction bands actually overlap in mid-gap.
Another important consequence of disorder is that the electrons may
become "localized", i.e. Spatially confined to the vicinity of
predominantly a single atomic site. This occurrence is more probable in
the potentials experienced by the electrons, and so is more likely in the
hand tails, since these arise in general from the most distorted sites.
It. should be stressed that localized state are not unique to the
amorphous phase, but can also occur in crystalline materials containing
a degree of disorder. Criteria, which dictate whether electron state
should be localized or non-localized, must be established before any
discussion concerning the amount of disorder necessary to induce it can
be pursued.
A such criterion suggested by Anderson [23] that, if an electron is

placed on an atomic site at T = OK and a time t = 0, then the state is
seemed to be localized if as t   the electron has not diffused away and
has a finite probability, P, where P  C : exp (-2  r) , of remaining at
distance (r) within the atomic volume, if there is a finite chance of
diffusiun at T = OK then the state is extended or delocalized. The
quantity ( 1 ) is known as the localization length, if this length is much
greater than the sample size, it becomes very difficult to determine
whether the state is localized or extended.
2-4:- BAND MODELS OF AMORPHOUS SEMICONDUCTORS.
In review of the large differences in the nature of the various groups

of amorphous semiconductors, it is obvious to note that any single
model can not describe the essential features of all amorphous
materials.Anyhow some of the different models which are presently in
use are described as follows:
2-4.1: Mott-CFO model:
Based on earlier ideas ,proposed by Mott [24,25,26], Cohen,

Fritzsche and Ovshinsky [22] have developed a model for the electronic
structure of amorphous semiconductors, this has come to be known as
the Cohen-Fritzsche-Ovshinsky (CFO) model, as shown in Fig.(2). This
model is derived from the concept of the ideal covalent random network
structure. Simple chemical considerations and some physical
observations suggest that the valence and conduction bands are
separated by a gap. Translational and compositional disorder ara
assumed to cause fluctuations of the potential of sufficient magnitude
that they give rise to localized states extending from the conduction and
valence bands into the gap. These localized states are not associated with
definite imperfections but are the result of the randomness of potential
barrier. Their number and energy depend on the degree of randomness
and strength of scattering.Valence band tail states are assumed to be
neutral when they are occupied Conduction band tail states are neutral
when they are empty. This places the Fermi energy somevvhere near the
gap center. The character of the wave function changes at critical
energies EC and EV which separate the extended and localized states.
Here, electron and hole mobilities drop sharply from a low mobility
band transport.with finite mobility at T = O K to a thermally activated
hopping between localized gep states whichdisappear at T= O K.
Fig. (2) CFO – Model

(a) Localized states extending from the conduction and valence band into the
gap [dashed area]
(b) The mobility gap (Eo-Ev)
The so called mobility edges, define a mobility gap (Ec-Ev) which

contains only localized states. This model is believed to be applied to
alloy glasses which contain compositional as well as positional disorder.
It. was originally intended to apply to tTrulticomponent chalcogenide
glasses. It was able to account for the well- defined electrical activation
energy, the pinned Fermi energy, thermally stimulated currents and
some aspects of the photoluminescence. Its major disadvantages were
the sharp optical absorptsion, the absence of variable range hopping near
the Fermi energy and the complete absence absence of any unpaired-
spin density at all ternperatmres.
2-4.2: Davis - Mott Model:
This model [27] reveals that the tails of localized states should be
rather narrow and should extend a few tenths of an electron volt in the
forbidden gap, Fig [3]. In this model it was distinguished between:
i- The ranges Ec (between Ec and E A ) & EB and EV ) where the localized states
lie in the bands and are due to the lack of' long-range order, and
ii- Hypothetical "taits" due to the defects in the structures. The energies
Ec and Ev separate the ranges of energy where states are localized
and non-localized (extended). The quantity (EC- EV) defines a
mobility gap.
Davis a Mott [27] also assumed that some, and perhaps all
chalcogenide semiconductors contain a band of compensated levels
quite near the middle of the bandgap. Such a band can pin the Fermi
level, so that the semiconductor is extrinsic rather than than
intrinsic.This assumption was made about amorphous germanium [28]
to eaxplain the change of sign in the thermopower at low temperature,
where the levels being acccepters and supposed to be due to dangling
bonds.
Fig (3)
Density of states g(E) suggested by Davis and Mott.
(EC – EV) is the mobility gap
where (EC - EA)= EC & ( E B  EV )  Ev contain localized states
Thermaly assisted hopping may take place in these ranges. Ec & Ev
2-4.3: Marshall-Owen Modal:
In this model (29) the position of the Fermi level is determined by

bands of donors and accepters in the upper and lower halves of the
mobility gap respectively as shown in Fig.(4) . The concentrations of
donors and accepters adjust themselves by self compensation to be
nearly equal, so Fermi level remains near the gap center. At low
temperature it moves to one of the impurity bands because self
compensation is not likely to be complete.
2-4.4: Small polaron Model:
The small polarom model was proposed by Emin [30]:Acr.oruing to

this model, charge carriers enter the self-trapped (small polaron) state
via polarization of the atomic lattic. The field effect screening is
accomplished byredistribution of small polarons within the space charge
layer. The small density of polarons can be calculated, its thermal
activation energy being the separation in energy of the polaron band
from the Fermi level.The difference between its energy and conductivity
activation energy is equal to the hopping energy for polaron movement.
The temperature dependence of the field effect is therefore capable of
testing the small polaron model in a rage where the field effect screening
is controlled by polarons.
2-4.5: Licciardello Model:
Licciardello [31] modified the CFO model by superimposing the

suggestion of Anderson [32] which states that disordered systems are
characterized by sufficiently strong electron-phonon coupling to produce
a negative correlation energy, and this can account for the sharpness of
the absorption edge and the absence of unpaired spins. The model has
major difficulties in explaining the correlations between amorphous and
crystalline solids, the differences between chalcogenides and
tetrahedrally bonded amorphous semiconductors and their luminescence
results.
Fig (4) Density of states g(E) suggested by
Marshal and Owen.
A band of localized acceptor states
Lies below and a band of localized
Donors states above the gap center.
2-4.6 Ovshinsky and Adler Model:

In this model [33] many types of chemical-bonding arrangements
were discussed which can characterize primarily covalent amorphous
Semiconductors.They showed how the removal of crystalline constraints
inherent in all amorphous solids can allow a wealth of unusual chemical
bonding possibilities. These unusual chemical configurations are
necessary to understand recent experimental results on both doped and
chemically modified amorphous semiconductors. This overall chemical
approach provides a unified method for the understanding of the
electronic structure of all amorphous semiconductors. They identified
the normal structural bonds and their low energy states. They described
the deviant electronic Configurations which characterize primarily
amorphous solids. Deviant electronics configurations are that atoms in
the solids which coordinated differently from their normal structures
bonding. The density of states of a chalcogenide glass is sketched in fig.
(5). Cha1cogen atoms have high energy lone pairs, which do not
ordinarily participate in covalent bonding. In an amorphous solid, they
have a wide spectrum of orientations, resulting in diverse mutual lone-
pairs interactions [34].These can give a distribution of states in the gap
which could yield n-type or even metallic conduction under extreme
conditions of strong coupling.This possibility exists in chalcogenide
glasses because it involves interactions among non-bonding electrons
and thus does not significantly increase the energy of the solid.The
resulting localized states are very important because of the large density
of lone pairs in the glass (one per chalcogen atom). In real materials the
additional existence of significant densities of valence alternation pairs
ordinary compensates for any increase of the fermi level and conversion
of some of the positively charged centers, three-fold coordinated
chalcogen atom and a negatively charged centers, singly coordinated
chalcogen atom keeps EF pinned
‫)‪Fig. (5‬‬
‫‪Densty of states g(E) suggested by Ovshinsky and Adler‬‬
CHAPTER (III)
EL ECTRICAL TRANSPORT IN AMORPHOUS

SEMICONDUCTORS
3-1 : DIRECT CURRENT CONDUCTION.

3-1.0 : Interoduction.
3-1.1 : Extended state conduction.
3-1.2 : Thermally assisted tunneling conduction.
3-1.3 : Conduction in localized states near the Fermi energy EF
3-2 : AC. CONDUCTION.
3-2.0 : Interoduction.
3-2.1 : Dielectric relaxation in solids.
3-2.2 : The A.C Conductivity in amorphous semiconductors.
3-2.2(a) : Transport by carriers excited to the extended states near
EC or EV.
3-2.2(b) : Transport by carriers excited into the localized states.
3-2.2(c) : Hopping transport by electrons with energies near the
Fermi level.
3-1: DIRECT CURRENT CONDLICTTON.
3-1.0: Introdmrtion.
The change in electrical prnperties is more significant at the

transition from amorphous to crystalline state [35]. Amorphous alloys
behave as intrinsic semiconductors over a wide range of temperature.
The electric conductivity of most chalcogenide glass is less sensitive to
impurities than the that each atom in these alloys is surrounded by a
number of atoms equal to its valence. In other words, all atoms have
their velence requirements locally satisfied [22.36].
On the basis of the principles suggested by Mott and Davis [37],

there can be three processes leading to conduction in amorphous
semiconductors. Their contribution to the total conductivity will change
markedly in different temperature ranges.
3-1.1: Extended state conduction.
The total conductivity at a finite temperature can be written as:

f ( E )
    E .dE (3.1)
E
1
  E  Ef 
where f ( E )  1  exp  
  kT 
is the Formi-Dirac distribution function.

So, if only the current-carrying states are the extended states beyond
Ev and Ec then
  q  N ( E ). ( E ). f ( E ).[1  f ( E ).dE (3.2)
where  (E ) and N(E) are the mobility and density of states at an energy
(E). Assuming Mott's model for sharp mobility edge at Ev and Ec is
valid, and if conduction takes place far enough away from (Ef) then
  ( Ec  Ef 
 ext  qN ( Ec)  e kT exp  (3.3)
 kT 
where  ext is the conductivity due to extended states and  has been
taken to be zero in the mobility gap, and to be states are not strongly
affected by disorder,  c might describe the motion of nearly free
electrons (or holes) with occasional scattering.
  ( Ec  Ef 
Then  ext   min exp kT


(3.4)
If (Ef) moved from its usual position in mid-gap through the

mobikity edge, one expects the behavior shown in Fig. (6), the physical
meaning of ( min ) is immediately apparent, being the limiting metallic
conductivity, and also the focus of the extraplated [Ln VS. 1/T] curves.
It is important to understand, that the observed activation energy for
conduction (E ) is not (Ec-Ef) because the band gap is itself a function
of temperature, decreasing with increasing temperature since the
phonons contribte a term to the electron self-energy. Optical
measurements show that at room temperature, the dependence is nearly
linear, so
Ec  E f  E   T (3.5)
Then the pre-exponential in equation (3.3) becomes
‫‪ ‬‬
‫‪Co   e.q.N ( Ec).kT . exp  ‬‬ ‫)‪(3.6‬‬
‫‪k‬‬
‫‪Because e  1 / T , then Co is temperature independent.‬‬
Fig. (6)
Scbematic representation of the temperature dependence of the conductivity

as the Fermi level EF is moved away from Ec into the gap in the direction
indicated
3-1.2: Thermally assisted tunneling conduction.
In this mechanism one consider that the leectrical conduativity in

amorphous semiconductor results from excitation of carriers to the
localized tall states below the mobility edge, and hopping at energies
close to EA or EB. Since it is unlikely to find such a level at the same
energy, the tunneling process is usually inelastic, (i.e. it involves the
emission or absorption of a phonon). Hence this thermally assisted
tunneling process involves a hopping energy ( W1 ) in addition to the
activation energy (EA-Ef) needed to raise the electron to the appropriate
localized state.
The conductivity will be thus of the form:

   1 exp [( E A  E f  W1 ) / kT ] (3.7)
It should be mentioned that (W1 ) decreases with decreasing
temperature on account of the variable-range nature of the hopping
transport. So, the conductivity curve is not expected to exhibit a constant
activation energy but a decreasing slope as temperature is lowered. An
estimate of ( 1 ) is not easy to make but it is expected to be several
decades smaller than ( min ) , partly because of a lower effective density
of states near (EA) compared with experimental slope and intercept of
 1 1
 Ln  VS T  at  T   0, will be affected by any temperature dependence of
(EA-Ef) analogous to that affecting the slope and intercept associated
with the with the band conduction process, (3-1.1). Thus, the
predominant temperature dependence is from the exponential term with
exponent, ( E A  EF  ) , and this behavior is expected to be observed at
temperature lower than that at which etitended state conduction occurs.
3-1.3: Conduction in localized states near the Fermi

energy Ef.
In real materials, which may contain a high concentration of defects

and consequently have a high density of state in the gap.electron
transport can take place via such defect states, and the magnitude of this
defect-controlled conductivity may greatly exceed that due to
conventional band conduction.It is an obvious extrension of hopping
conduction model to apply it to the case of conduction in amorphous
semiconductor, in which all the states being in the mobility gap are
localized, as in the case of impurity conduction(e.g donor).The simplest
approach is perhaps that given in the original application of the hopping
mechanism to amorphous solids by Mott [24,25,26]. In which, he
considered two localized sites, one filled and at or slightly below the
fermi energy Ef, and the other empty and above Ef; their energy and
spatial separations are  and R, respectively. The hopping transition
rate ( ) will be determined by three factors:
(i) The probability of existence of a phonon of energy  , given by the


Boltzmann factor [ exp( KT ) ] where, the presence of the phonon is
necessary to conserve energy in the hopping process.
(ii) the probability of electron transfer between the sites if it is assumed
that, each state is exponentially localized, with the same localization
length  1 , where the wave function   exp( r ), and if an electron
transfers by means of tunneling, the probability is given by
[exp ( 2 R ) ].
(iii) a rate term which can be regarded as an attempt frequency (  ) and

which will depend on the strength of the electron-phonon coupling
and the phonon density of states, but will be only weakly dependent
on R and  .
So, the hopping transition rate is given by:
  
   o exp   2  R  
 k T 
the conductivity may be calculated from the Einstein relation:

eD  R2

kT
; D
6 (3.9)
If the density of states at the Fermi level N(Ef) is constant, so that

[kT N(Ef)] is the concentration of gloctrons contributing to the
conductivity which may be rewritten as:
e2 2   
 
6
R  o N ( E f ) exp   2R 


kT 
(3.10)
This equation represent hopping between nearest neighbors if the

distance R nearly equal the radius ( 1 ) of the localized wave function
on each site. The conductivity may represented by:
  W2 
   2 exp   
 kT 
where  2   1   min and (W2 ) is the hopping energy, of the order of half
the width of the band of states.
At temperatures such that (kT) is less than the bandwidth, at low

temperature, where the number and energy of phonons are both small,
hopping to near neighbors is unfavorable because of the large energy
separation encountered on average. Instead, it is more favorable for the
electron to tunnel to more distant sites, since there is a higher probability
that more distant sites will have smaller energy separations. Mott [24]
found that, the final expression for the conductivity is:
 A 
   `2 exp  1 / 4  (3.12)
 T 
1/ 2
e z o  N (E f ) 
where  `2 
2(8 ) 1/ 2 
 kT 
(3.13)
1/ 4
  
and A  2 .1 
kTN ( Ef )  (3.14)
 
where the optimum hopping distance R is given by:

1/ 4
 9 
R 
8N ( Ef ) kT  (3.15)
 
and the optimum energy separation  is given by:

1/ 4
 (kT ) 3 
  5.16   (3.16)
 N ( Ef ) 
The pre-exponential terms  2 and  2` are, as for  1 , not easy to

evaluate mainly because of uncertainties in the term ( o ) , where ( o ) is
the attempt frequency and which will depend on the strength of the
electron-phonon coupling and the phonon density of state. The change
of conductivity mechanism with temperature is shown diagrammatically
in Fig. (7).
So Far, we can see that upon the application of a d.c electric field,
conduction can take place among localized gap states in amorphous
semiconductors by a mechanism of the phonon assisted hopping, which
is simply activated with a constant activation energy if hopping is
between nearest neighbors at constant range [eq. (3.11)], but with a
‫‪temperature activation energy (i.e. T-1/4 behavior) if the hopping is‬‬
‫‪between neighbors at variable ranges [eq (3.12)].‬‬
‫)‪Fig (7‬‬
‫‪The activation energies associated with various processes‬‬

CH 1

Uploaded by

CH 1

Uploaded by

The author would like to experess his sincere gratitude to Professor

The author wishes to experess his thanks to Dr.Z.I. EL-Badawy for

The author wishes thanks all collegians in the physics department

Amorphous semiconductor draw the attentoin of many researchers

In this thesis the system Ge Se Te was chosen. Nine ingots with

(3) [Ge 18 Te 82 z Se ] where z = 14.17.18

As a preliminary test the I-V curves of 9 samples belonging to the

are the values of voltange , current and resistance at the

the assumption saying that the I-V behavior is a result of interaction

should be noted that both equations fit results up to a current value

c) Activation energy values of conduction

The conductivity of the nine samples were also measured as a

be divided into three parts, separated at constant kink temperatures each

Several models were used to explain results. The most successful

Above results enables one to understand the mechanisms

the presence of intimate valence alternating pairs (IVAP) which are

‫‪1-l‬‬ ‫‪: INTRODUCTION‬‬

The amorphous solids have an intense interest, makes them a

Glass which was previously considered as an optical, dielectric or

Amorphous materials not only can be designed to be very stable, but

The most important, elemental constituents of amorphous

Amorphous (or non- crystalline) materials such as glass are not

Amorphous materials possess some degree of structural disorder.

1-2.1: Disorder of amorphous materials.

1-2.1(a). Geornetric disorder is that form of randomness in which three

1-2.1(b). Mapnetic disorder in which the atomic site possesses a spin

1-2.1(c). Substitutional disorder in which the material with one type of

1-2.2: Glass transition

When a liquid is cooled, one of two events may occur. Either

Considering the variation of molar volume with temperature the

1-2.3: Preparation of amorphous materialas

There are at least a dozen different techniquescan be used to prepare

This technique is the most widely used method for producing

The principal disadvantage of thermal evaporation as a preparative

Some of these parameters can be controlled, others cannot be

The sputtering process is rather more complicated than thermal

1-2.3(c). Glow-discharge decomposition.

Another technique which can produce thin films of amorphous

1-2.3(d). Chemical vapor deposition.

This method is analogous to the glow discharge method in both

1-2.3(e). Melt quenching.

The oldest, method of producing an amorphotts solid is to cool the

The most usual way of producing samples is to seal a charge (1-10

a) Temperature of the oven.

1-2.3(f ). Electrolytic deposition.

Another technique which uses solutions of the desired material is

1-2.4: Structure of amorphous semiconductors.

As a result of long-range periodicity in crystalline solids the density

Fritzsche [10], measured the differential thermal analysis curve of

1-2.4(b). Kinetics of crystallization in amorphous semiconductors.

The memory effect observed in glass semiconductors is usually

where f is the function of crystallized material and C the rate constant

where E is the activation energy for crystallization and

If the temperature rises during the transformation the reaction rate,

This condition gives:

In which  s the constant heating rate. By plotting in ( / TC2 )VS .1 / T one

‫‪THEORETICAL INTRODUCTION TO THE BAND‬‬

2-1: THE FAILURE OF THE BAND THEORY.

Weaire and Thorpe [21] first showed that if short-rang

It has also been found that [21], if V2  2V1 , there is no overlap

2-2: THE STRUCTURAL MODELS.

In the Continuous Random Network (GRN) the term "random

The first involves the physical building of a model, which for a

The second way of constructing structural models is by using a

From above we conclude that the problem is to obtain the electron

The presence of disorder manifests itself on the electron states in a

A such criterion suggested by Anderson [23] that, if an electron is

2-4:- BAND MODELS OF AMORPHOUS SEMICONDUCTORS.

In review of the large differences in the nature of the various groups

2-4.1: Mott-CFO model:

Based on earlier ideas ,proposed by Mott [24,25,26], Cohen,

Fig. (2) CFO – Model

The so called mobility edges, define a mobility gap (Ec-Ev) which

2-4.2: Davis - Mott Model:

In this model (29) the position of the Fermi level is determined by

2-4.4: Small polaron Model: