Tntoducleon to c mate aualr k devices

M. [Link]
3
SEMICONDUCTOR
MATERIALS AND THEIR
PROPERTIES

INTRODUCTION

Semiconductors are important because they have found applications in

branches of industry and areas of daily life. Moreover, almost all
the
tors provides an opportunity to apply and test much of the study of semiconduc
obtain information on transport properties which theory of solids and to
study of metals alone. There are two reasons why could not be obtained from the
of electronic processes is easier in semiconductors observation and interpretation
tals. First, semiconductor crystals can be grown than in metals and other crys
has been achieved in the case of metals and with purity far in excess of what
insulators.
Concentrations in a semiconductor are low, so that the Second, the möbile carrier
carriers can be treated as
distinguishable, noninteracting particles like
cussed in Sec. 1.7, for noninteracting particlesmolecules of an ideal gas. As dis
the Pauli exclusion principle is of
little consequence. Thus, instead of the Fermi-Dirac
simpler Maxwell-Boltzmann (MB) statistics. It thenstatistics we can use the much
analytical solutions for many problems that cannot becomes possible to obtain
FD statistics. The condition for which the MB be solved accurately using the
conductor is stated in Sec. 3.5.2. statistics can be used in a semi
This chapter presents a detailed discussion of
types of semiconductors are described, and the semiconductors. The different
oped by taking elemental semiconductors as antheory of these materials is devel
example.
3.1 SEMICONDUCTING MATERIALS

In the previous chapter we saw that

substances in
metals and insulators. Metals are good conductors ofnature can be classified as
tors have very high values of resistivity. The electricity, whereas insula
resistivity scale of various sub
stances in nature is shown in Fig. 3.1. On this scale
that have resisitivity ranging from about 10- to 10° semiconductors
n - cm.
are materials
however, is quite arbitrary because it does not say in what This classification,
are different from metals and insulators. respect semiconductors

61
 SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES ELEMENTAL AND COMPOUND SEMICONDUCTORS []
62
83

the neighboring atoms are bound byy homopolar cohesive forces. These iorces
10 104 10 10 10° 1012 106 1020 termine the two important properties of these materials, namely, the meltiderng
p (0-Cm) point and the energy gap. Some important characteristics of group IV semiso
Metals t - Semiconductors Insulators ductors are given in Table 3.1.
FIGURE 3.1 Resistivity scale of materials found in nature. The cohesive forces are weakened with an increase in atomic number, and this
is reflected by an increase in the size of the atom and a decrease in the melting
Amore satisfactory definition of semiconductors is provided by the band the point and the band gap. Thus, there is agradual transition from strongly insulat.
ory of solids. As shown in Chapter 2, in metals the bonding of atoms is such that ing to essentially metallic behavior. The diamond, for example, is an insulator,
free clectrons become available after the bond formation. However, in an insula whereas near room temperature a-Sn behaves almost like a metal.
tor all the valence electrons are used to form the bonds, and the resulting valence
band is separated from the conduction band by an energy gap. Consequently, mo 3.2.2 Compound Semiconductors
bile carriers in insulators can be produced only when energy is supplied from an
external source (such as thermal radiation) to raise electrons from the valence In recent years anumber of compound semiconductors have found applications in
band to the conduction band. Semiconductors have an energy band structure simi various devices. The most important among them are the intermetallic II-V and
lar to that of insulators. However, the energy gap is smaller (generally less than IIVI compounds. Some IV-VIcompounds also exhibit semiconducting prope
2.5 eV), so that a significant number of electrons can be thermally excited from ties, but most of them have a small band gap that limits their use to infrared de
the valence band to the conduction band at room temperature. tectors and lasers.
III-V Compound Semiconductors These semiconductors are compounds formed
3.1.1 Classification of Semiconductors [] from the elements in group IIA (all metals) with those in group VA. Ignoring the
compounds of thallium and bismuth, we present the remaining sixteen:
Semiconductors can be classified in several ways. One classification may be ac
cording to the nature of the current carriers. Thus, we have ionic and electronic
semiconductors. In ionic semiconductors, the conduction takes place through the Group Group
movement of ions and is accompanied by mass transport. In an electronic semi III A ! VA’
conductor, current is carried by electrons and no mass transport is involved.
Semiconductors may also be classified as elemental and compound semicon B
P As St
ductors. In an elemental semiconductor, all the constituent atoms are of the same BN BP BAS BSb
AI AIN AIP AlAs
kind,whereas acompound semiconductor is composed of atoms of two or more Ga GaN GaP GaAs
AISb
different elements. GaSb
In InN InP InAs InSb
Yet another way of classifying semiconductors is according to their structure.
Accordingly, asemiconductor may be amorphous, polycrystalline, or single crys
tal. The nature of the crystalline structure has asignificant influence on the prop Not all these compounds are potential semiconductors. Boron and nitrogen
erties of the semiconductor. Amorphous semiconductors have ashort-range order com
pounds, for example, have been included only for the sake of completeness. Alu
and can be considered to be quasi-periodic. They display poor characteristics in minum compounds are not very stable and usually disintegrate with time. The six
terms of carrier mobility and lifetime. In the majority of the semiconductor
most important semiconductors are printed in bold type.
devices, single-crystal materials are used because of their superior electrical char Most of the III-V semiconductors have the zinc blende crystal structure. Eight
acteristics. Polycrystalline semiconductors show electrical behavior similar to valence electrons are shared between a pair of nearest atoms, so that, on an aver
that of single crystals, but their conductivity is significantly lower because of the
presence of grain boundaries that obstruct the flow of mobile carriers. Therefore,
they have found only limited use. In this book we will be concerned mainly with TABLE 3.1
single-crystal electronic semiconductors.
Some Properites of Group IV Elements

3.2 ELEMENTAL AND COMPOUND SEMICONDUCTORS [1] Band Gap E

Atomic at 300 K Melting Covalent
Element Number (eV)
3.2.1 Elemental Semiconductors
Point (°C) Radius (Å)
Carbon (C) 6 5.30 3800 0.77
Two important semiconductors are germanium and silicon, which belong to Silicon (Si) 14 " 1.12 1417 1.17
Germanium (Ge) 32
group IV of the periodic table. All the group IV elements have two electrons in Tin (a-Sn) 50
0.67 9317 12
0.08
their outermost p subshell and crystallize in a diamond-type structure in which 232 1.40
64
SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES ELEMENTAL AND COMPOUND SEMICONDUCTORS 11 65

age, cach atom has four valence clectrons. This

valent character, and to a first approximation,suggests that the bonding has a co
the cohesion between the atoms is
Horizontal Direction Both the cohesive energy and the energy gap increase in
homopolar. Therefore, we would expect the properties horizontal transition from elemental to compound semiconductors. This happens
of these compounds to be
similar to those of corresponding group IV elements. However,
because of increased cohesion due to the ionic contribution.
of group Illare more electropositive, and those of group Vare since the elements The technical importance of IIIV semiconductors is that they provide a wider
more electronega choice of band gap than do elemental semiconductors. Some of these compounds
tive than the group IV elements, the bonding in IIIV compounds has a partial have energy gaps corresponding to the visible region of spectrum and have found
ionic character as well. Therefore, the cohesive force between atoms represents
the cohesive force of the covalent bonding plus an additional term because of applications in light-emitting diodes. Others, like InSb and InAs, are used in gal
vanomagnetic devices.
ionic contribution. As a result, the cohesive force and the strength with which Amajor problem with compound semiconductors is that their preparation in
the valence electrons are bound to the atoms are higher for these crystals than single-crystal form is more difficult. The bond formation process usually encour
those for the corresponding group IV semiconductors. This is evidenced by the ages a uniform growth, and we would, therefore, expect the stoichiometry of the
fact that each of these compounds has a higher melting point and a larger band compound to be established during crystal growth. However, because of the dif
gap compared to group IV elements with the same lattice spacing, atomic num ferent vapor pressures of the constituents near the melting point, one component
ber, and density. As an example, consider InSb and a-Sn. These substances may vaporize more rapidly, causing an excess of the other type of atoms. These ex
have the same lattice constant and the same average atomic number and atomic ces atoms may either get trapped interstitially in the lattice or may precipitate out
weight. On the basis of purely covalent bonding, the two should exhibit the same to form a second phase. Either of these possibilities destroys the stoichiometry.
properties. However, because of the partial ionic character of bonds in InSb, it The III-V compounds have a general formula A"BY. However, several of these
has a higher melting point (525°C) and a larger band gap (E, = 0.17 eV at 300 K). con1pounds have identical structures and similar lattice constants. Such com
Because of their predominantly covalent character, the IIIV semiconductors pounds exhibit complete solid solubility in each other and structures of the form
form a link between group IV elements and IIVI compounds. Therefore, most of A,B",C are possible, with x ranging continuously from 0 to 1. The properties
their characteristics can be deduced from those of neighboring materials. To ap of such compounds vary smoothly with composition. Thus, it is possible to grow
ply these considerations, Table 3.2 is drawn by assigning isoelectronic III-V com crystals of a compound like Al,Ga,-As. Similarly, the growth of crystals having
pounds to their group IV clements. Here, the average total number of clectrons two different group V elements like GaP,As1-y is also possible.
(core plus valence electrons) per atom in a row is nearly the same, and this is II-VICompound Semiconductors The[e semiconductorsS are formed by com
what is meant by the term isoelectronic. It is seen that certain material properties bining the elements in groups IIB and VIA in the periodic table. Alist of all the
change gradually as one moves from one end to the other in horizontal and verti 0I-VIcompounds is given below:
cal directions. These characteristics were predicted by H.J. Welker (2] even
before n1any of the compounds were obtained in the laboratory. They can be
summarized as follows [1: Group Group
Vertical Direction The cohesive energy and the energy gap E, decrease with IIB | VIA’

increasing atomic number. This is because in an atom with a large number of Se Te

electrons, the core electrons repel each other with a greater force. This results in Zn ZnO ZnS ZnSe ZnTe
weak cohesion of atoms and a loose binding of valence electrons. Decreasing Cd CdO CaS CaSe CaTe
binding energy results in wider allowed bands and narrower band gaps. Hg HgO HgS HgSe HgTe

TABLE 3.2 Among the oxides, ZnO has found use in both electrophotography as a photo
Variation in the Properties of lsoelectronic Group IV and llI-V Compounds sensitive material and in paints and varnishes. Little information is available
Group IV Semiconductors IIIV Compounds about the semiconducting properties of CdO and Hgo. The remaining three
group VI elements are called chalcogenides. Among the chalcogenides of mer
BN cury both HgSe and HgTe are semimetals, whereas HgS is a semiconductor. The
SiC BP AIN remaining six compounds printed in bold letters form a group of potential IIVI
S BAS AIP GaN semiconductors.
AlAS GaP InN
SiGe BSb
InP
The crystal structure of these compounds is rather complex. CdS and CdSe
Ge AISb GaAs crystallize in the hcp wurtzite structure, and ZnTe and CßTe in the zinc blende
GeSn GaSb InAs
InSb
structure, whereas ZnS and ZnSe can ekist in both these forms. The bonding in
a-Sn
these compounds is a mixture of covalent and ionic types. This is because the av
Cohesive energy erage number of electrons per atom is still 4, but VIatoms are considerably
more electronegative than group II atoms. and this ints
Band gap L introduces ionicity. The ionic
character varies considerably over the whole range and increases as the atomic
66 SEMICONDUCTOR MATERIALS AND THEIA PROPERTIES
THE VALENCE BOND MODEL OF THE SEMICONDUCTOR 67

size decreases. Thus, while ZnS is predominantly jonic, the bonding is nearly co 3. A layer of stable native oxide (SiO,) can be grown thermally on the surface
valent in HTe. The ionic character has the cffect of binding the valence electrons of Si. This oxide layer can be used for selective masking of the silicon sur
rather tightly to the lattice atoms. Thus, the band gaps of these compounds are face. The grown SiO, also makes the surface of the device passive and can
larger than those of the covalent semiconductors of comparable atomic weights,
although their hardness does not match that of covalent semiconductors. Because be used for protecting the device against the effects of the ambient. No
other semiconductor is known to form a stable oxide.
of their large band gaps, these materials are considered potential sources of elec
troluminescent devices.
Like IV compounds,ternary compounds of the form A'BY'CcY, can also be Two approaches can be employed to study semiconductors: the valence bond
obtained from many of these semiconductors. Thus, it has been possible to grow model and the energy band model. The valence bond model emphasizes events
crystals of ZnS,Se ,-s, with xranging from 0to 1. Here again the properties of the in space and time, whereas the energy band model does the same in energy and
resulting compound vary gradually with the fraction x. momentum. Each of these models gives söme information that the other cannot,
and in some respects, they complement cach other.
Narrow Gap Semiconductors Another class of semiconductors that have re
ceivedconsiderable attention in recent years because of their applications as in 3.3.1 Intrinsic Semiconductors
frared detectors and radiation sources are IV-VI narrow gap semiconductors.
Four of these compounds, namely, PbS, PbSe, PbTe, and SnTe, crystallize in the
simple cubic NaCl crystal structure. The bonding in these compounds is mainly As we saw in the previous chapter, Ge and Si crystallize in a diamond lattice,
ionic with some covalent contribution. with each atom bound to its four nearest neighbors by covalent bonds. These
With the exception of SnTe, all of these compounds have asmall band gap and neighbors are allequidistant from the central atom and lie at the four corners of
low effective mass of the carriers. In IVVI compounds and their alloys, single a tetrahedron (Fig. 3.2). The distance between the nearest neighbors is V3a/4
crystals of somewhat inferior quality can be easily prepared. These crystals have where a is the side of a unit cell (Fig. 2.7). Each bond has two electrons with
a number of defects such as voids, inclusions, dislocations, and point defects. opposite spins, so that the central atom appears to have eight electrons in its va
High-quality crystals are difficult to prepare. lence orbit which it shares with its four nearest neighbors. A large crystal of sili
con can be generated by repeating this unit cell in space.
Other Semiconductors Several other compounds have been found to have semi
Let us consider a pure crystal of Si containing no impurities and no lattice de
conducting properties. These include a number of simple metallic oxides (such as fects. A two-dimensional representation of such a crystal is shown in Fig. 3.3.
MgO and Cu,0), a large number of transition metal oxides (TiO2, V,0, Fe,03, Here the circles represent the atomic core with four positive charges, and the
and NiO), and several other oxides having the general formula AB,O4, where A dots represent the valence electrons. At 0°K all the electrons are tightly bound in
is a second group element and B is a transition metal. Typical examples of this the bonds, and the crystal is a perfect insulator. At higher temperatures lattice
class are MgFe,0, and ZnFe,04. Some of these compounds have found use in atoms vibrate around their equilibrium positions. These vibrations
devices such as thermistors, but many others are only of academic interest.
occasionally
impart enough energy to some of the bound electrons, enabling them to leave
bond and move freely inside the crystal. Fig. 3.3a illustrates the creation ofthea
3.3 THE VALENCE BOND MODEL OF THE SEMICONDUCTOR
free electron by breaking a covalent bond at position A. The vacant site created
by the broken bond has a net positive charge left on the atomic core. This va
After presenting a brief account of different types of semiconductors, our next
endeavor will be to study the basic principles of scmiconductor theory. In doing
so, we willbe concerned mainly with the elemental semiconductors. There are
two reasons for making this choice. First, elemental semiconductors have the (0, V2, V2)
simplest structure and their properties are easy to understand. A second and
more compelling reason is that these materials are widely used in semiconductor | (V4, V4, V4)
devices. During the early days of transistor development, Ge was the dominant (V2, 0, v2)
material used in making diodes and transistors. However, most of the present-day
devices employ Si. Next comes GaAs which has found applications in.a number
of devices where it has established its superiority over Si. However, GaAs tech 10947°
nology is more involved, and the material is much more expensive than Si. There
are three main reasons why Si has become the device material of choice:
(0, 0, 0)
1. Si is the most readily available material and is found in its oxides and sili
cates which form about 25 percent of the earth's crust. (v2. V2, 0)
2. Si has a wider band gap than Ge. Thus, Si devices can be operated at tem
peratures up to 200C compared to Ge devices which cannot be operated
above 80C FIGURE 3.2 Tetrahedron con figuration of Si crystal showing the four nearest neighbors.
68 SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES THE VALENCE BOND MODEL OF THE SEMICONDUCTOR 69

Section 3.8.1 willshow that at room temperature the wave function of a hole
(or an clectron) extends over aregion of space that encompasses several thousand
lattice atoms. Thus, the size of ahole is orders of magnitude larger than the inter
atomic spacing. This being the case, any notion of ahole movement being created
Conduction by the jumping of a bound clectron bet ween neighboring bonds is meaningless.
electron
Another difficulty is encountered when one tries to explain the Hall effect
(see Sec. 4.5). If the motion of a hole can be represented by the motion of an elec
tron in the opposite direction, then the Hall voltage will have the same negative
sign regardless of whether electrons or holes are moving. However, the measured
Hall voltage is positive when the current is carried by holes.
Thus, it is clear that the mental picture of a hole provided by the valence bond
model, though helpful in visualizing how ahole is created, is not useful in under
standing the motion of a hole as aparticle. This should not be surprising inas
much as neither a hole nor an electron is a classical particle.
la) In the above process, the electron-hole pairs were produced by the thermal
vibrations of lattice atoms. This process of pair generation is called the thermal
generation process. In addition to this process, there is a reverse process of re
combination in which a freely moving electron loses its ionization energy by
reentering into a broken covalent bond. In this way an electron-hole pair disap
pears. When asemiconductor crystal is in equilibrium at agiven temperature, a
certain number of bonds always remain broken. Any additional breaking of bonds
above their thermal equilibrium value is exactly balanced by the reverse process
of recombination.
The semiconductor under discussion will have as many vacant sites as the
number of free electrons. Such a semiconductor is called an intrinsic semicon
Electron ductor, and the holes and electrons created in this way are called intrinsic charge
Hole carriers. The total number of holes and electrons in a piece of semiconductor is
rarely of any interest. What is usually important is the number of carriers per unit
volume, known as carrier density or carrier concentration, and it is expressed as
the number of carriers per cm'. Thus, for an intrinsic semiconductor
(b)
p =n= n; (3.1)
FIGURE 3.3 A [wo-dimensional schematic representation of silicon crystal showing the atom
cores and valence electrons. (a) Thermal generation of an electron-hole pair and (b) motion of a where p and n represent the hole and electron concentrations, respectively, and
hole from location A toB. (From S. M. Sze, Semiconductor Devices: Physics and Technology, p. 9.
the subscript i is used to denote that the semiconductor is intrinsic. The intrinsic
Copyright O1985. Reprinted by permission of John Wiley &Sons, Inc., New York.)
concentration n,varies with the temperature and is different for different semi
conductors. The values of n, for Si, Ge, and GaAs at 300 Kare given in Chapter 4
cancy behaves like a particle of positive charge and mass cqual to that of an elec Table 4.2.
tron and is called ahole. Like afree electron, ahole can also move freely through
the crystal.
It might seem quite distubing that a vacant site, short of an electron, can 3.3.2 Extrinsic Semiconductors
move through the crystal. One way to visualize the movement of a hole is shown
in Fig. [Link], as aresult of thermal motion, an electron in acovalent bond at Apart from thermal generation, charge carriers in semiconductors can also be
Badjacent to the hole jumps into the original vacant site A, thus transferring the created by adding certain types of impurity atoms. For silicon and germanium,
location of the hole to the site from where the electron came. In this way, the mo the impurities of most interest are the elements of groups III and V shown in the
tion of a hole can be regarded as the transfer of ionization from one atom to an periodic table. These impurities are intentionally added after purifying the crys
other by the jumping of the bound electron from acovalent bond to a neighboring tal. The process of adding impurities to a semiconductor is called doping, and the
vacant site. This process does not affect the motion of afree electron. Thus, elec impurity-added semiconductors are known as doped semiconductors. By suitable
trons and holes can move freely through the crystal independent of each other. doping, a crystal can be made to have unequal values of electron and hole con
This picture of ahole has anumber of serious limitations which will become clear centrations. The resulting semiconductor is said to be extrinsic. Almost all the
from the following discussion. semiconductor devices employ extrinsic semiconductors.
 71
SEMICONDUCTOR
70 OF THE
SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES THE VALENCE BOND MODEL

temperatures the thermal vibrations

of the lattice
(a) N-TYPE SEMICONDUCTORS tion energy. Thus, at ordinary electron to shake it free, whereupon it be.
atoms will impart enough energy to the positive
Let us assume that a very small amount of a group Velement (As, P, Sb) is added impurity atom left behind becomes a
to an otherwise pure crystal of silicon. Upon incorporation into the crystal, comes a conduction electron. The electric field because it is bound to
these ion is immobile under an applied
ion. This
impurities occupy lattice sites that are normally occupied by Si atoms. Thérefore,
neighboring atoms by the usual covalent
bonds. When a freely moving electron
these dopants are called substitutional impurities. Fig. 3.4 shows the situation electrostatic attraction, the electronwill not
stay long
when an As atom replaces a Si atom in the lattice. An As atom has five valence gets attracted to it owing to
again shake it free. Since the
electrons. Four of these make covalent bonds with the Si atoms, but the fifth because the thermal vibrations of the lattice will
crystal, it is called a donor atom or
clectron cannot be accommodated in the bonding arrangement and is bound to impurity atom donates one free electron to the
the impurity atom only by weak electrostatic forces. At 0°K this electron will simply a donor. distributed in the semi
move around the impurity atom just like an clectron in a hydrogen atom. There Now consider a large number of donor atoms randomly atoms donates one
fore, we can use the Bohr theory of the hydrogen atom to calculate the radius of conductor. At a sufficiently high temperature, each of these
concentration in the
its ground state orbit (r) and the ionization energy (E). However, two changes conduction electron to the crystal. Therefore, the electron concentration.
must be made. First, the effective mass of the electron mhas to be used in place semiconductor rises above n, and becomes higher than the hole
Such a semiconductor is known as an n-type semiconductor. In an n-type semi
of the free electron mass m,. Second, the free space permittivity (e) has to
replaced by the permittivity of the semiconductors (e,). Thus, we can write be conductor, the electrons are in the majority and are termed the majority carriers,
whereas the holes, which are in the minority, are the minority carriers.
rise in the electron concentration above n,depresses the hole concentration
A
(3.2a) below n,. This happens because of the electron-hole pair recombination process.
The rate of recombination R is proportional to the electron and hole concentra
and
tions (7 and p), and can be written as R = a,pn, where a, is a constant of pro
portionality. For an intrinsic semiconductor in thermal equilibrium p = n = n,
and R = a,n. Since this rate must be balanced by the thermal generation rate
E, = (3.2b) Gh, We have

R= Gh = a,n? (3.3)
where 0.53 Å is the ground state Bohr radius and 13.6 eV is the ionization energy
of the electron in the hydrogen atom. For germanium, m# 0.5m, and e, = At agiven temperature, G is found to be independent of doping of the semicon
16e,. Substituting these values in Eqs. (3.2a) and (3.2b) yields [Link] state ra ductor provided that the dopant concentration remains small (i.e., less than 1do
dius of approximately 16 and an ionization energy of 0.026 eV. For silicon, the pant atom for 10° atoms). Now consider a crystal with. the equilibrium
corresponding values are 6.3 and 0.1 eV, respectively. concentrations n, and po, where the subscript o is used to denote thermal equi
From the above discussion, it is clear that a very small energy is required to librium. For this crystal R = a, D,n, = Gh. Thus, if n, is increased aboven, by
ionize the extra electron attached to the impurity atom. The magnitude of ther doping the semiconductor, P, must be decreased to keepp.n, = n. The physical
mal energy kTIg at 290 Kis 0.025 eV, which is comparable to the electron ioniza process is as follows: The donor atoms contribute additional electrons to the
crystal. Some of these electrons fall back into the broken bonds causing a de
crease in the hole concentration below n,.

(b) P-TYPE SEMICONDUCTORS

An atom of group IIelements (B, Al, In, Ga) also occupies a substitutional posi
tion in the Si (or Ge) lattice but has only three valence electrons. These electrons
Conduction
electron
complete three of the covalent bonds, leaving one bond vacant which accepts an
electron from aneighboring atom and creates a hole in the crystal as shown in
Fig. 3.5a for an Al atom in the Si lattice. When this happens, a negative charge
is left on the Al atom. The hole is attracted toward this negative charge and
moves around it in a Bohr-like orbit as shown in Fig. 3.5b. At 0°K the hole re
mains bound to the impurity atom but breaks away from it at higher tempera
tures, whereupon it wanders freely in the crystal. The binding energy of the hole
can be calculated from Eg. (3.2b) after replacing m; by the hole effective mass
m. The calculated value is in reasonably good agreement with the experimen
tally determined values in Ge, but only an order of magnitude agreement is seen
FIGURE 3.4 A drnor atom in a silicon crystal. in the case of Si.
72
SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES THE ENERGY BAND MODEL 73

a hole that becomes free only at high temperatures. Similarly. a much less elec
tronegative isovalent atom, like Bi. substituted for P in GaP acts as a donor. In a
III-V semiconductor (such as GaAs) the presence of a Ga atom at an As site (or
vice versa) would create an electrically active defect. However, in covalently
e
bonded compounds no more than one atom out of 10 atoms is situated on the
wrong sublattice, and the effect of this type of defect is not very significant.
Hole In II-VIsemiconductors, group IIIelements replace the group II elements and
act as donors. Elements of group VII (CI, Br, and I) also act as donors by replac
Hole
+q le
ing group VI elements. Similarly, ap-type semiconductor can be cbtained by dop
ing either with group Ior with group V elements. Semiconductors that can be
made bothp- and n-type are said to be ambipolar. Elemental semiconductors (Ge
and Si) and most of the III-V compounds are ambipolar. II-VI semiconductors
that have a larger ionic contribution in bonding tend to be unipolar. Some of
them are found only n-type semiconductors while others are only the p-type. The
(a) (6) mechanism that gives rise to this behavior is known as self-compensation.
To understand self-compensation, let us assume that ZnTe is to be doped with
FIGURE 3.5 (a) An accepton atom in silicon crystal and (b) hole motion around the negatively iodine (|) to make it n-type. For an I atom to replace a Te atom, the semiconduc
charged acceptor atom.
tor temperature has to be raised, and this creates zinc vacancies by evaporation
of Zn atoms. Each Zn vacancy acts as a double acceptor by accepting two free
An impurity atom that contributes a hole is called an acceptor atom because it electrons. Let AE(Zn) be the energy required to create a Zn vacancy, and let AE()
accepts a bound electron from the covalent bond. When a number of acceptor be the energy released when a donor electron drops into the acceptor level cre
atoms are added to the semiconductor, most of these atoms ionize at room tem ated by the Zn vacancy. Since the binding energy of donors and acceptors is
perature, contributing one hole per impurity atom. The resulting crystal has an ex small, AE(I) E,. Now suppose that a Zn vacancy is created in ZnTe doped
cess of holes compared to the electrons andis called ap-type seniconductor. Like with I. If the two donor electrons (from two Iatoms) can drop into the Zn va
the donor atom, the ionized acceptor atom remains immobile in an electric field. cancy, an energy of 2E, is regained in the process. If 2E, > AE(Zn), the energy
In a semiconductor crystal that is doped with both donors and acceptors, the of the system is lowered by the creation of Zn vacancies, and the trapping of the
extra electrons attached to the donor atoms fall into the incomplete bonds of donor electrons by these vacancies will frustrate all efforts to dope ZnTe with
the acceptor atoms so that neither electrons nor holes will be produced. This
Iatoms. Thus, ZnTe always remains p-type. This is because the Zn atom is
process is known as compensation, and the semiconductor having both donors
smaller than the Te atom, and its binding energy is lower than that of a Te atom.
and acceptors is said to be compensated. If the number of donors exceeds the Just the opposite is true for CdS. Here the small-sized S atoms have smaller bind
number of acceptors, the semiconductor becomes the n-type, and vice versa for a ing energies than the larger Cd ions. Thus, CdS always has an excess of sulphur
p-type semiconductor. When the two types of impurities are equal in number, vacancies, which act as donors, and is always n-type. The same is true for CdSe
the material is said to be fully compensated. The electron-hole concentrations in
and ZnSe. However, in CTe the energy E, is small, and self-compensation is not
such a semiconductor are equal, but many of the properties (e.g., resistivity) are complete, so CdTe is ambipolar.
different from those of the intrinsic crystal. In narrow gap semiconductors PbS, PbSe, and PbTe, doping is easily affected
(c) DOPING OF COMPOUND SEMICONDUCTORS by deviation from stoichiometric composition. Excess metal atoms act as donors,
whereas excess chalcogenide atoms act as acceptors. Impurity doping is possible
Compound semiconductors have more than one kind of atom. Therefore, the role with Tl as an acceptor and Bi as a donor.
of an impurity in these semiconductors depends largely on which kind of atom is
replaced by the impurity atom. In IIV semiconductors, group II elements re
place the trivalent atoms and act as acceptors. The most important donors are 3.4 THE ENERGY BAND MODEL
atoms from group VI which replace the pentavalent atoms. Atoms of group IV
are called amphoteric impurities. They act as donors when they are built into the 3.4.1 The Energy Band Structure of Semiconductors
sites occupied by group lII elements; they act as acceptors when they replace the
pentavalent atoms. In a three-dimensional crystal k has three components-k, ky, k;-and a two
In some compound semiconductors like GaP, it has been observed that even dimensional representation of the E versus krelation is difficult. Because of crys
isovalent atoms (those having the same number of valence electrons) act as
donors and acceptors. For example, nitrogen (N) is an isovalent acceptor in GaP.
tal symmetry, it becomes more meaningful to express Eas afunction of kalong
the principal directions of the crystal. Fig. 3.6 shows the E-k plots for two semi
The electronegativity of an Natom is greater than that of the P atom. When the conductors along the [100] and [11] directions. In Fig. 3.6a it is seen that the
Natom replaces P in GaP, it becomes negatively charged by attracting an extra lowest conduction band minimum and the valence band maximum are situated at
electron to itself. The electrostatic field of this negative charge can loosely bind the same value of k. Semiconductors with this type of band structure are called
74 THE ENERGY BAND MODEL
SEMICONDUCTOR MATERIALS AND THEIR PAOPERTIES
E,(D)
[111) (111]
[100) (100)
fer e n -(orduétor
FIGURE 3:7 Energy band diagram for a semimetal.
(a) (6)
FIGURE 3.6 E-k plots for (a) direct gap and (b) indirect gap semiconductors. tion is made between the direct and indirect gap semiconductors. The veriical
distance in Fig. 3.8 represents electron energy that increases upwards, and there
direct gap semiconductors. The significance of this term is that a photon of en fore, the electrostatic potential increases downwards. The horizontal coordinate
ergy hv = E,can excite an electron from the top of the valence band directly in the diagram has no significance since it is a point diagram that is valid at each
into a state at the bottom of the conduction band, as shown by the vertical line in point in the interior of the semiconductor (but not at the surface). We recall that
Fig. 3.6a. If the photon energy hu > E, the excess energy is shared between the the energy band picture of a solid is based on the coupling of identical
that are arranged in a periodic atoms
electron and the hole, and both have the same value of k (dashed line). Fig. 3.6b array. Thus, the energy band diagram is a char
is the E-k diagram for an indirect gap semiconductor. Here the valence band acteristic of the material, provided that the volume of the
maximum and the lowest conduction band minimum are situated at different val a sufficiently large number of unit cells-a requirement thatsample contains
is met in all
ues of k. In such semiconductors, the direct transition of the electron from the laboratory-size samples.
valence band to the conduction band by a photon of energy E, is not possible. Let us now consider the intrinsic semiconductor shown in
This happens becausea photon has very small momentum, whereas in a nonverti valence band is completely filled, the conduction band has [Link], 3.8. At 0°K the
cal transition over E,, an electron suffers alarge change in momentum. However, material is an insulator. At a finite temperature, thermal no and the
vibrations of lattice
a nonvertical transition can occur indirectly by the cooperation of a lattice pho atoms produce a certain concentration nof electrons in the conduction
an equal number of empty states in the valence band. In the presence ofband and
non which can supply the required momentum. In this case, the minimum photon nlly applied field, electrons will conduct in the an exter
energy required to produce an electron-hole pair will be larger than E,. Vertical
transitions are also possible for hu> E,(D), where E,(D) is the direct gap energy. sites (1.., the holes) will conduct in the valenceconduction
band
band, and the vacant
Elemental semiconductors Ge and Si have an indirect band gap, as do GaP and all How an empty state in a valence band acts as a hole can be seen as follows:
Alcompounds. Other IIV and most IIVIsemiconductors have direct band gap. Suppose there are 4N states in the valence band and the current arising
from
Figure 3.7 shows the E-k diagram of asubstance in which the conduction and
valence bands overlap slightly. Because of this overlap, some states in the conduc
tion band get filled leaving behind an equal number of empty states in the valence
band, thus, making available a finite number of electron-hole pairs,even at 0°K. Conduction
Such substances are called semimetals. Typical examples are Bi, As, Sb, and
some of their alloys.
eEnlecrtgoyn band
3.4.2 Energy Band Representation of Semiconductors E
(a) INTRINSIC SEMICONDUCTOR Valence
band
The energy band scheme in Fig. 3.6 is too detailed. One generally represents the
electron energy in the valence and conduction bands separated by a band gap as
shown in Fig. 3.8 for an intrinsic semiconductor. Here Ey represents the valence FIGURE 3.8 A tw0-dimensional energy band diagram of an intrinsic
band maximum, and Eç represents the conduction band minimum. No distinc semiconductor.
 THE ENERGY BAND MODEL 77
SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES

cach of these states is qv, where v, is the velocity of electron in the ith energy Itis now easy to see what happens in a semiconductor that is doped with both
state. Thus, for the completely filled band, the current density J can be written as donors and acceptors. In such a semiconductor, electrons from the level E, in
stead of going to the conduction band will drop into the acceptor states at E.;
cach such jump eliminates one electron-hole pair that would have been there,
J=-,- 0 (3.4) provided this jump could be prevented. If there are more donors than acceptorS,
then allthe states at E, will be filled by the electrons from the donor levels and
the leftover electrons of the donor atoms will be excited to the conduction band
where Vis the volume. Now assume that some jth state near the top of the va to make the semiconductor n-type. If there is an excess of acceptors, then the
lence band is empty. The current density J, arising from the band is then given by semiconductor will become p-type.
(3.5) -34:3 The Energy Levels of Impurities in Si and GaAs
and this can be written as
Our discussion thus far has been confined to the donor and acceptor atoms that
give rise to energy levels close to band edges. These levels are called shallow levels.
Several impurities give rise to energy levels deep in the band gap; they are known
(3.6) as deep impurities.
Figure 3.10 depicts the energy levels of the commonly known impurities in Si
and GaAs. The ionization energies for important impurity atoms are also given.
Thus, in the absence of the ith electron, the current density is given by qv,. Note that many of these impurities give rise to more than one energy level.
(b) EXTRINSIC SEMICONDUCTORS
3.4.4 The Energy-Momentum Relation in Valence and Conduction Bands
The energy band diagram in Fig. 3.8 can be easily modified to include the effect
atoms. We have seen that the
of the electron energy levels of donor au From the energy band diagram of a semiconductor, it is clear that an electron is
fifth electron of a donor atom is loosely bound dcan be set free by supplying
Thus, the electron not available in the conduction band unless its total energy E exceeds Ec. An
an energy E,which is only about tens of milli electron volt. conduction
electron with energy E = Ec is said to have zero kinetic energy. If the energy E
attached to a donor atom must have an energy level E, below the
Obvi
band edge Ec such that Ec - E, =E;. This level is shown in Fig. 3.9a.
donor
ously, Ez lies in the otherwise forbidden region. It is customary to show the Sb As T C Pt Au
the crystal
level by a dashed line because donor atoms are randomly distributed in 0.039 0.045 05A 0.21 0.16
to be noted that the 0.25 0.25
and do not form a periodic structure like the host atoms. It is 0.38
electron attached
level represented by Eg corresponds to the ground state of the
between E, and Ec, but 0.54 0.51
to the donor atom. There could be many excited states 1.12
temperatures almost
these states are of little interest to us because at ordinary 0.34 0.35 0.36
0.41

all the donors are ionized and their electrons are set free into the conduction 0.16
0.29
shown in Fig. 3.95.
band. The energy band diagram for a p-type semiconductor is
0.045 0.067 0.072 03

The acceptor energy level E, lies in the forbidden gap close to the valence band Al Ga Ir P Ey
edge Ey. Se So Te

0.006 0.006 0.006 0 0.0058 0.006 Ec

GaAs
0.4

0.63
1.42
(Donor level) (Acceptor level) E
0 44

0.24
0.028 0.028 0.031 0.035 0.0352s
Be Mg Zn Cd Si Cu Cr Ey
(a) FIGURE 3.10 Energy levels of impurities in Si and GaAs. (From S. M. Sze, Semiconductor Devices:
FIGURE 3.9 Energy band iagrams of extrinsic semiconductor showing (a) the donor level and Physics and Technology, p. 23. Copyright © 1985. Reprinted by permission of John Wiley & Sons,
Inc., New York.)
(b) the acceptor level.
78
SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES CONCENTRATIONS OF ELECTRONS AND HOLES INSIDE THE ENERGY BANDS

is greater than Ec, the difference (E - Ec) will appear as the kinetic energy of
the electron in the conduction band. Assuming the clectron to be a classical par Consider a volume in k space with components k, ky, and k, as shown in
ticle with effective mass m and momentum p, we can write Fig. 3.11. Since k? = k? + k; + kË, Eq. (3.8b) can be written as
energy E as

E = Ec + = Ee+mv?
2m (3.7a) E- (k} + k; + k;) (3.9)

Similarly, a hole at E, has a zero kinetic energy, and its energy E in the valence
band can be expressed as Using the method of separation of variables, we can write

w(x,yz) = yr)u(yb(z) (3.10)

2m
(3.7b)

These relations tell us how the energy of clectrons and holes in the Substituting for yx, yz) from Eg. (3.10) and for k from Eq. (3.9) into Eq. (3.8a),
and valence bands conduction we can show that y(x) has the solution
depends on their momenta.
(x) = B, exp(jk,x) + B, exp(- jk,x) (3.11)
3.5 EQUILIBRIUM CONCENTRATIONS OF ELECTRONS
INSIDE THE ENERGY BANDS AND HOLES Since a laboratory-size crystal is large compared to the dimensions of a
we assume y(*) to be periodic with periodicity L, so that* unit cell,
Now that the general features of energy bands in semiconductors have been de
scribed, the important problem to investigate is how the clectrons and holes are bx + L) = u(x)
distributed among the various available states inside the bands when thermal (3.12)
equilibrium prevails. Note that there are no allowed states in the forbidden gap. With this condition, Eg. (3.11) will be satisfied
To calculate the electron and hole concentrations in the conduction and valence only when
bands, we need to know how many energy states are available in a given energy
interval and how these states are occupied by the electrons. The product of the
density of states and the probability of number
i n t e r y a l : e y ortheSe states give the (3.13)
Of electrons in the energv

3.5.1 The Density of States Function

A function that gives the number of available electron states per unit volume in
an energy interval between Eand E + dE is designated by N(E)aE, where N(E)
is known as the density of states function and represents the number of states per
unit volume per eV.
To derive an expression for N(E), we need to consider a crystal of finite vol
ume. We assume that the constant energy surfaces are spherical so that the effec
tive mass m* is a scalar quantity and the relation between energy and momen
tum is parabolic. Let us now consider a free carrier (electron or hole) of mass m*
in the solid. Since the potential energy is arbitrary, we assume V = 0. The time 2/I. 6T/L. 10x/L
independent Schrödinger equation can then be written as 2z/L.

6/L
V'u + k' = 0 (3.8a)
with 10r/L

2m*
k? = 2 E (3.8b) FIGURE 3.11 The k-space for a free
carrier of mass m with a unit cell.
"Equation (3.12) is known as periodic
boundary condition.
 SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES
cONCENTRATIONS OF ELECTRONSs AND HOLES INSIDE THE ENERGY BANDS 81

where n, is a positive integer. Similar solutions can be found for (y) and (z). The density of electrons in the conduction band can be obtained by integrating
The vector k thus has the components
Eq. (3.18) from Ec to some value Eop at the top of the conduction band. The inte
2r 2T
gral, however, becomes complicated. For ease of calculation, we limit the range of
k, = k, =n,, k, = L -n, (3.14) application of Eq. (3.18) by assuming that (Ee - Er) > 34T. With this limitation,
the FD distribution function can be approximated by the Boltzmann distribution:
where n, and n, are also positive integers. From Fig. 3.11 the volume of a unit cell
in kspace is (2m/L = 8m/L'. Therefore, the volume of aunit cell in kspace for
a crystal of unit volume is 8. For a crystal of unit volume, the volume of a
spherical shell in k space having radius between k and (k + dk) is given by The integral of Eq. (3.18) can now be written as
(Etop
k+ dk} - *] - 4nk'dk n, -2myaE- E) exp - (E kT
- Ep) dE (3.19)

Thus, the number of unit cells in this volume is k² dk/2. The density of avail In a typical semiconductor, the width of the conduction band is several eV, but
able electron states is twice as large as this because each cell can have two elec most of the electrons are confined in the vicinity of Ec so that the upper limit of
trons with different spins. Hence, the number of electron states per unit volume integration can be replaced by infinity. The resulting integral can be easily evalu
of the crystal within an infinitesimal range dk of k is given by ated after substituting

N(K)dk =
k
(3.15) E- EC y.
kT

Using Eg. (3.8b), we can write this as Thus

N(E)dE = (2mE dE (3.16) no

em:kTY" exp- exp(-y) d
This equation gives the density of states for a particle of mass m* with energy
betweenE and E + dE. This result can be applied to electrons in the conduction (3.20)
band after replacing m* by m:. Since the energy E is measured from the conduc
tion band edge Ec, E is replaced by (E - Ec). Thus, for electrons in the conduc where
tion band

N(E)dE =
h
2mE-Ec"dE, for E > Ec (3.17a) - (3.21)

Nc iscalled the effective density of states in the conduction band. It

A similar expression can be written for holes in the valence band: number of states required to be placed at the energy Ec, which, represents the
with the probability of occupation of this level, gives the numberafter multiplying
of electrons in
4
N(E)dE = (2m(Ey - E) dE, for E < Ey (3.17b) the conduction band. Note that in reality the density of
states at Ec is zero and
Eq. (3.20) is just an artifice to obtain a simple expression for n,.
To compute the equilibrium hole
tion function fA(E) of a hole is that ofconcentration po, we note that the distribu
a state not occupied by an electron. Thus
3.5.2 The Effective Density of States
Under thermal equilibrium, the number of electrons dn, in an energy interval dE fi(E) = 1 - f(E) =
of the conduction band can be obtained by multiplying the function N(E)dE with (3.22)
the FD function f(E). Thus, 1+
ap)
dn, = N(E)f(E)dE A derivation similar to that implemented for
the electrons leads to (see Prob. 3.12)
(3.18)
p=N expl-E) (3.23)
 83
82 ENERGY BANDS
HOLES INSIDE THE
SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES CONCENTRATIONS OF ELECTRONS AND
equilibrium, the prod
for any semiconductor
in thermal
(3.24) tion since it shows that, concentrations is independent of doping.
uct of electron and hole (3.29) are valid only for
that Eqs. (3.19) through small. Consequently,
where Ny represents the effective density of states in the valence band. We would like to emphasize mobile carrier concentration is
semiconductors in which the Max well-Boltzmann distri
Equations (3.21)and (3.24) show that both Nc and Ny depend on can be replaced by the
Since the effective masses of electrons and holes are not the same, Nctemperature.
and Ny for the FD distribution function
the number of mobile carriers
in the bands is so
asemiconductor are different (see Table 4.2). bution. Physically, this means that available states is occupied. This is known
of
Small that only a very small fraction electrons (solute) form a dilute solution in
the
as the diluteness condition because dilute electron gas is said to be nondegen
3.5.3 The Equilibrium Electron-Hole Product the available energy states (solvent). A a nondegenerate semiconductor.
erate, and the resulting semiconductor is called
Since the expressions for Nç and Ny are independent of the doping of the semi
conducto, relations (3.20) and (3.23) are valid for any semiconductor provided
that the Fermi level E, lies in the forbidden gap and is at least 3kT away from the 3.5.4 The Temperature Dependence of Intrinsic Carrier Concentration
band edges. These relations are also valid for an intrinsic semiconductor. If E;
From Eq. (3.29) the intrinsic concentration n, can be written as
denotes the energy of the Fermi level in an intrinsic semiconductor, then Eqs.
(3.20) and (3.23) can be written as n, = K,T exp(-E/2kT) (3.30)

n,= Ne exp EcKT- E. (3.25) In Ge and Si, the band gap E, decreases monotonically with temperature and can
be expressed as
and (3.31)
E,(T) = Ep - b,I
where E is the extrapolated value of the band gap at 0°K and bË is the rate of
P. = n, = Ny (3.26) decrease of the energy gap with temperature. The linear form of Eq. (3.31) is fairly
accurate from 100 Kto 400 K, but is grossly inaccurate at low temperatures. Sub
Combining Eq. (3.25) with (3.20) and Eq. (3.26) with (3.23) leads to stituting Eg. (3.31) into Eq. (3.30) leads to the equation
n, = K,T' exp(-E/2kT) (3.32)
n, = n, (3.27)
where

and
K, = KË exp(b,/2k)

(3.28)
By substituting the numerical values of m;, m~, bi, and Em, n, as a function of
P. = n, temperature can be determined. However, because there are many difficulties in
the independent measurements of these quantities, such an effort does not lead
These relations show that p¡n, = [Link] is clear from Eq. (3.27) tht for an n-type to results in close agreement with the experimentally determined values. From
semiconductor (n, > n) the Fermi level E, lies above E, and hence is close to the measurements, it is found that for temperatures above 50 K[41
conduction band edge Ec. Similarly, for a p-type semiconductor the Fermi level
lies close to the valence band edge Ey. Multiplying Eq. (3.20) by Eq. (3.23) yields n,(T) = 1.76 x 109732 lcm- for Ge (3.33a)
p.n, = n/= KiT' exp(-E,/kT) (3.29)
and
where
n,(T) = 3. cm- for Si (3.33b)
KË =
3.5.5 lonization of Impurities
The effective masses m and m; which enter into the expressions for Nc and Ny We can calculate the values of the carrier concentration n, and p, from Eqs. (3.20)
are called density of states effective masses. Equation (3.29) is an important rela and (3.23) if the Fermi energy E; is known. In asemiconductor, E; is determined
 SEMICONDUCTOR MATERIALS AND THEIR PROPERTIES CONCENTRATIONS OF ELECTRONS AND HOLES INSIDE THE ENERGY BANDS
85

Com the charge neutrality condition which requires a knowledge of ionization of the right-hand side in
This fraction will remain small if the second term on
donor and acceptor impuritics. Thus, it is necessary to examine the subtle feature
Eq. (3.37) is very large compared to unity, which means
of theoccupation statistics of these impurities. In this discussion, we will assume
anondegenerate semiconductor. Let us first consider the case of donor levels. (3.38)
When a donor atom (e.g., As) is ionized, it loses an electron to the conduction
band and becomes positively charged. This situation can be represented by
kT in both Ge and Si: thus, the expo
Near 300K, (E - E) is on the order of
order of unity and the above condition re
Aa -eS As* nential factor in Eg. (3.38) is on the
cmat 300 Kfor Si, this condition is
duces to N, < Nc/2. Since Nc = 2.8 x 10
satisfied for all values of N, which are orders of magnitude lower than this num
If N represents the concentration of ionized donors and Nù that of the neu ber. This means that the fraction of unionized donors under this condition is very
tral donors, the total donor concentration N, is given by N, = NË + NË. One are almost completely ionized. The same is also true for
small, and the donors
might concBude that the probability of occupation of the donor level E by an acceptors.
electron can be obtained simply by substituting E = E, in the FD distribution silicon sample at 300 K doped with
To clarify these ideas, let us consider a
function. However, this is not correct because of the spin degeneracy of the donor cm. Assuming that all of them are ionized, we can write Ni =
2x 10" donors value of Ncand n, in Eq. (3.20)
levels. When a donor atom is ionized, there are two possible quantum states cor Nn, = 2x 10 and N? 0. Substituting the
responding to each of the two allowed spins. The electron can occupy any one of we obtain
these states with the condition that as soon as one of them is occupied, the occu
pancy of the other is precluded. Taking this into account, the occupation proba Ne =exo[(EC - E)JkT]- 10
bility of level E, is given by

fE) = (3.34)
(Ec - Ep) = 6.9kT
Hence, the concentration of ionized donors is
Now
(3.35) 6kT, since (Ec - E) = kT.
N = (N - N) = (E, - Ep) = (Ec- E;) - (Ec - E)
electron, and it is Substituting this value in Eq. (3.34) gives
filled when it is occupied by an
A donor state is said to be an acceptor state is empty when
ionized. In contrast,
said to be empty when it is atom 1
The transition process at an acceptor N [1+exp(6)] = 2exp(-6) = 200
neutral and is filled when ionized.
it is f(E.) ==
N
(e.g., Al) is
has its electron attached to the level
Al° + e 5 A Thus, out of 200 donor atoms, only one atom
percent ionized. Since there are Nc =
[Link], the donors are more than 99 dilute
spin degeneracy is 10 are occupied, the
of the acceptor level at E4, if 2.8 x 10"states at energy Ec, out of which only 2 x
The probability of occupation ness condition is also satisfied.
taken into consideration, is given by
(3.36)
Concentrations
3.5.6 Equilibrium Electron-Hole
total acceptor nondegenerate semiconductor to which
of ionized acceptors and N, is the - N,. Be Let us consider a homogeneous piece of
where N; isthe concentration been added. Since the semiconductor
concentration of neutral acceptors is N = N, N donors cm andN, acceptors cm have
concentration. The and Ge, the factor 2 in
of the valence band in Si as a whole remains space-charge neutral
cause of two-fold degeneracy
by 4.
Eq. (3.36) should be replaced the ratio of the concentration of electrons bound NË+ p, - N, - n, = 0
In an n-type semiconductor, electron concentration (n, + N2) can be ob
total
to donor atoms (N) to the the unity term in Eq. (3.34). That
is
tained by using Eq. (3.20) and by neglecting or

n, - P, = Ni - N,
(3.39)
NG (3.37)
n, + N
86 SEMICONDUCTOR MATERIALS AND THEIA PROPERTIES THE FERMI LEVEL AND ENERGY DISTRIBUTION OF CARRIERS
87

The left-hand side of this relation represents the mobile charge and the right As an example, consider a Ge sample having n, = 2 x 10 cm and p, =
hand side the immobile charge. The value of N¿ can be obtained from Eq. (3.35) 3.5 X10 cm at 200 Kwhere all the dopants are ionized. Let us calculate the
provided E; and T are known, and the value of N, can be similarly determined values of n, and p, at 300 K.
using Eq. (3.36), so that the difference (n, - p.) is known. At temperatures
above 100 K, practically all the donors and acceptors are ionized and Eq. (3.39)
can be rewritten as At 200 K, (N, - N) = (n, - p.) 2 x 10 cm

n, - p, = NË - N, (3.40) This is the net dopant concentration in the sample and will remain the same at
300 K. The value of n, at 300K as obtained from Table 4.2 is 2.4 x 10 cm
From Eq. (3.29) Thus, we can write

Pon, = n M,- P, = 2 x 10
From the identity (p. + n,} = (p, - n,} + 4pono, We obtain and

(p. + no) = +[(N - N,} + 4n2 (3.41) n,p, = (2.4 x 10 = 576 x 10s
In this equation, only the positive sign has physical significance because (p. + Solving these equations yields n, = 3.6 x 10 cm and p, = 1.6 x 10 cm.
n,) is always positive. From Eqs. (3.40) and (3.41) we obtain

(N, - N,) + VN - ,) +4n] 3.6 THE FERMI LEVEL AND ENERGY DISTRIBUTION OF CARRIERS
THE BANDS (3] INSIDE

P, =(N, - N) + V(N, - N, +4ni] (3.42) 3.6.1 Calculation of the Fermi Level

These relations show that if the net impurity concentration N - N, = 0, then The position of the Fermi level in a semiconductor is determined from the charge
p. = n, = n,and the semiconductor is intrinsic. If (N, - N) is positive, n, > neutrality condition
p, and the semiconductor is n-type. On the other hand, when N, > NËthe semi
conductor is p-type. Po - n, + NË - N; = 0
For an n-type semiconductor having (N N, > 4n',
For a nondegenerate semiconductor, n, is given by Eq. (3.20) and p, by Eq. (3.23).
n, = (Nå - N,) = N if N, = 0 Substituting for N from Eq. (3.35) and for N; from (3.36), we can
above relation as write the
and

Po = n}/N, (3.43)

Similarly, for ap-type semiconductor with (N, - N> 4ni, = 0 (3.45)

P. (N, - N) = N, if N = 0
This equation can be solved for Es provided that all the other quantities are
and known. We will now apply Eg. (3.45) for some simple cases.
n, = nN, (3.44) (a) AN INTRINSIC SEMICONDUCTOR
Semiconductors that satisfy Eqs. (3.43) and (3.44) are considered to be strongly In the case of an intrinsic semiconductor, (N; - N;) = 0, and denoting E, by
extrinsic. As the temperature of an extrinsic semiconductor is gradually raised, E, we can write Eq. (3.45) as
n, increases with temperature and eventually a point is reached where the semi
conductor starts behaving like an intrinsic semiconductor. We will assume the
condition |N. Nl= Sn, as the onset point of intrinsic behavior.
Ey - 2E,)
exp/c+kT