Chenming Hu - ch1 3

Hu_ch01v4.
fm Page 1 Thursday, February 12, 2009 10:14 AM
1
Electrons and Holes
in Semiconductors
CHAPTER OBJECTIVES
This chapter provides the basic concepts and terminology for understanding
semiconductors. Of particular importance are the concepts of energy band, the two
kinds of electrical charge carriers called electrons and holes, and how the carrier
concentrations can be controlled with the addition of dopants. Another group of
valuable facts and tools is the Fermi distribution function and the concept of the Fermi
level. The electron and hole concentrations are closely linked to the Fermi level. The
materials introduced in this chapter will be used repeatedly as each new device topic is
introduced in the subsequent chapters. When studying this chapter, please pay
attention to (1) concepts, (2) terminology, (3) typical values for Si, and (4) all boxed
equations such as Eq. (1.7.1).
T
he title and many of the ideas of this chapter come from a pioneering book,
Electrons and Holes in Semiconductors by William Shockley [1], published
in 1950, two years after the invention of the transistor. In 1956, Shockley
shared the Nobel Prize in physics for the invention of the transistor with Brattain
and Bardeen (Fig. 1–1).
The materials to be presented in this and the next chapter have been found
over the years to be useful and necessary for gaining a deep understanding of a
large variety of semiconductor devices. Mastery of the terms, concepts, and models
presented here will prepare you for understanding not only the many
semiconductor devices that are in existence today but also many more that will be
invented in the future. It will also enable you to communicate knowledgeably with
others working in the field of semiconductor devices.
1.1 ● SILICON CRYSTAL STRUCTURE ●
A crystalline solid consists of atoms arranged in a repetitive structure. The

periodic structure can be determined by means of X-ray diffraction and electron
microscopy. The large cubic unit shown in Fig. 1–2 is the unit cell of the silicon
1
Hu_ch01v4.fm Page 2 Thursday, February 12, 2009 10:14 AM
2 Chapter 1 ● Electrons and Holes in Semiconductors
● Inventors of the Transistor ●

Born on three different continents (Brattain in Amoy, China; Bardeen in Madison,
Wisconsin, USA; and Shockley in London, England), they all grew up in the United
States and invented the transistor in 1947–1948 at Bell Telephone Laboratories.
Brattain was an experimentalist while Bardeen and Shockley contributed more to
the concepts and theories. Their reflections on that historic event:
“... after fourteen years of work, I was beginning to give up ...”
—Walter H. Brattain (1902–1987)
“Experiments that led to the invention of the point-contact transistor by Walter
Brattain and me were done in November and December, 1947, followed closely by
the invention of the junction transistor by Shockley.”
—John Bardeen (1908–1991)
“All of us who were involved had no doubt that we had opened a door to a new
important technology. ”
—William B. Shockley (1910–1988)
FIGURE 1–1 Transistor inventors John Bardeen, William Shockley, and Walter Brattain (left
to right) at Bell Telephone Laboratories. (Courtesy of Corbis/Bettmann.)
crystal. Each sphere represents a silicon atom. This unit cell is repeated in all
three directions many times to form a silicon crystal. The length of the unit cell,
e.g., 5.43 Å in Fig. 1–2, is called the lattice constant.
The most important information from Fig. 1–2 is the simple fact that each and
every silicon atom has four other silicon atoms as its nearest neighbor atoms. This
fact is illustrated in Fig. 1–2 with the darkened cluster of a center atom having four
neighboring atoms. This cluster is called the primitive cell. Silicon is a group IV
element in the periodic table and has four valence electrons. These four electrons
are shared with the nearest neighbors so that eight covalent electrons are associated
1.1 ● Silicon Crystal Structure 3
5.43 Å
FIGURE 1–2 The unit cell of the silicon crystal. Each sphere is a Si atom. Each Si atom has
four nearest neighbors as illustrated in the small cube with darkened atoms. (Adapted from
Shockley [1].) For an interactive model of the unit cell, see http://jas.eng.buffalo.edu/
with each atom. The structure shown in Fig. 1–2 is known as the diamond structure
because it is also the unit cell of the diamond crystal with each sphere representing
a carbon atom. Germanium, the semiconductor with which the first transistor was
made, also has the diamond crystal structure.
Figure 1–3 introduces a useful system of denoting the orientation of the
silicon crystal. The cube in Fig. 1–3a represents the Si unit cell shown in Fig. 1–2 and
each darkened surface is a crystal plane. The (100) crystal plane in the leftmost
drawing in Fig. 1–3a, for example, is simply the plane in Fig. 1–2 closest to the
reader. It intersects the x axis at 1 lattice constant and the y and z axes at infinity.
One might refer to this plane as the 1 ∞ plane. However, it is standard practice to
refer to it as the (1/1 1/∞ 1/∞), or the (100), plane. In general, the (abc) plane
intersects the x, y, and z axes at 1/a, 1/b, and 1/c lattice constants. For example, the
(011) plane in the middle drawing in Fig. 1–3a intersects the x axis at infinity and
the y and z axes at 1 lattice constant. The numerals in the parentheses are called the
Miller indices. The related symbol [abc] indicates the direction in the crystal normal
to the (abc) plane. For example, when an electron travels in the [100] direction, it
travels perpendicular to the (100) plane, i.e., along the x axis.
Figure 1–3b shows that the silicon wafers are usually cut along the (100) plane to
obtain uniformity and good device performance. A flat or a notch is cut along the (011)
plane in order to precisely and consistently orient the wafer as desired during device
fabrication. Different surface orientations have different properties such as the rate of
oxidation and the electronic quality of the oxide/semiconductor interface. Both the
surface orientation and the direction of current flow along the surface affect the speed
performance of a surface-base device such as metal-oxide-semiconductor field-effect
transistor (MOSFET, see Section 6.3.1). The most important semiconductor materials
used in microelectronics are crystalline. However, most everyday solids are not single
crystals as explained in the sidebar in Section 3.7.
z z z
y y y
x x x
(100) (011) (111)
(a)
(100)
Plane
(011)
Flat
(b)
(c)
FIGURE 1–3 (a) A system for describing the crystal planes. Each cube represents the unit
cell in Fig. 1–2. (b) Silicon wafers are usually cut along the (100) plane. This sample has a
(011) flat to identify wafer orientation during device fabrication. (c) Scanning tunneling
microscope view of the individual atoms of silicon (111) plane.
1.2 ● BOND MODEL OF ELECTRONS AND HOLES ●
Each silicon atom is surrounded by four nearest neighbors as illustrated by the shaded
cluster in Fig. 1–2. We can represent the silicon crystal structure with the two-
dimensional drawing shown in Fig. 1–4. An Si atom is connected to each neighbor
with two dots representing the two shared electrons in the covalent bond. Figure 1–4
suggests that there are no free electrons to conduct electric current. This is strictly true
1.2 ● Bond Model of Electrons and Holes 5
Si Si Si
Si Si Si
Si Si Si
FIGURE 1–4 The silicon crystal structure in a two-dimensional representation.
only at the absolute zero temperature. At any other temperature, thermal energy will
cause a small fraction of the covalent electrons to break loose and become conduction
electrons as illustrated in Fig. 1–5a. Conduction electrons can move around in a
crystal and therefore can carry electrical currents. For this reason, the conduction
electrons are of more interest to the operation of devices than valence electrons.
An interesting thing happens when an electron breaks loose and becomes
free. It leaves behind a void, or a hole indicated by the open circle in Fig. 1–5a. The
hole can readily accept a new electron as shown in Fig. 1–5b. This provides another
means for electrons to move about and conduct currents. An alternative way to
think of this process is that the hole moves to a new location. It is much easier to
think of this second means of current conduction as the motion of a positive hole
than the motion of negative electrons moving in the opposite direction just as it is
much easier to think about the motion of a bubble in liquid than the liquid
movement that creates the moving bubble.
In semiconductors, current conduction by holes is as important as electron
conduction in general. It is important to become familiar with thinking of the holes as
mobile particles carrying positive charge, just as real as conduction electrons are
mobile particles carrying negative charge. It takes about 1.1 eV of energy to free a
covalent electron to create a conduction electron and a hole. This energy can be
determined, for example, from a photoconductivity experiment. When light shines on a
Si sample, its conductivity increases because of the generation of mobile electrons and
holes. The minimum photon energy required to induce photoconductivity is 1.1 eV.
The densities of thermally generated electrons and holes in semiconductors
are generally very small at room temperature given that the thermal energy, kT, is
26 meV at room temperature. A much larger number of conduction electrons can be
introduced if desired by introducing suitable impurity atoms—a process called doping
Si Si Si Si Si Si
Si Si Si Si Si Si
Si Si Si Si Si Si
(a) (b)
FIGURE 1–5 (a) When a covalent electron breaks loose, it becomes mobile and can conduct
electrical current. It also creates a void or a hole represented by the open circle. The hole can
also move about as indicated by the arrow in (b) and thus conduct electrical current.
Si Si Si Si Si Si
Si As Si Si B Si
Si Si Si Si Si Si
(a) (b)
FIGURE 1–6 Doping of a semiconductor is illustrated with the bond model. (a) As is a
donor. (b) B is an acceptor.
the semiconductor. For example, group V elements such as As shown in Fig. 1–6a
bring five valence electrons with each atom. While four electrons are shared with the
neighboring Si atoms, the fifth electron may escape to become a mobile electron,
leaving behind a positive As ion. Such impurities are called donors for they donate
electrons. Notice that in this case, no hole is created in conjunction with the creation of
a conduction electron. Semiconductors containing many mobile electrons and few
holes are called N-type semiconductors because electrons carry negative (N) charge.
As and P are the most commonly used donors in Si.
Similarly, when boron, a group III impurity, is introduced into Si as shown in
Fig. 1–6b, each B atom can accept an extra electron to satisfy the covalent bonds,
thus creating a hole. Such dopants are called acceptors, for they accept electrons.
Semiconductors doped with acceptors have many holes and few mobile electrons,
and are called P type because holes carry positive (P) charge. Boron is the most
commonly used acceptor in Si. In and Al are occasionally used.
The energy required to ionize a donor atom (i.e., to free the extra electron and
leave a positive ion behind) may be estimated by modifying the theory of the
ionization energy of a hydrogen atom,
m0q 4
E ion = -------------
- = 13.6 eV (1.2.1)
2
8 ε0h 2
where m0 is the free electron mass, ε0 is the permittivity of free space, and h is
Planck’s constant. The modification involves replacing ε0 with 12ε0 (where 12 is the
relative permittivity of silicon) and replacing m0 with an electron effective mass, mn ,
which is a few times smaller than m0 as explained later. The result is about 50 meV.
Because donors have such small ionization energies, they are usually fully ionized at
room temperature. For example, 1017cm–3 of donor atoms would lead to 1017cm–3 of
conduction electrons. The same conclusion also applies to the acceptors.
● GaAs, III–V Compound Semiconductors and Their Dopants ●

GaAs and similar compound semiconductors, such as InP and GaN, are dominant
in optoelectronic devices such as light-emitting diodes and semiconductor lasers
(see Sections 4.13 and 4.14). GaAs also plays a role in high-frequency electronics (see
Sections 6.3.2 and 6.3.3). Its crystal structure is shown in Fig. 1–7 and Fig. 1–8. The
similarity to the Si crystal is obvious. The shaded spheres represent As atoms and the
1.2 ● Bond Model of Electrons and Holes 7
light spheres represent Ga atoms. Each Ga atom has four As neighbors and each
As atom has four Ga neighbors. The lattice constant is 5.65 Å. Ga is a group III
element and As is a group V element. GaAs is known as a III–V compound
semiconductor, as are GaP and A1As, which also have the same crystal structure as
illustrated in Fig. 1–7.
It is probably obvious that group VI elements such as S and Se can replace the
group V As and serve as donors in GaAs. Similarly, group II elements such as Zn can
replace Ga and serve as acceptors.
But, are group IV elements such as Si and Ge donors or acceptors in GaAs?
The answer is that they can be either donors or acceptors, depending on whether they
substitute for Ga atoms (which have three valence electrons) or As atoms (which
have five valence electrons). Such impurities are called amphoteric dopants. It turns
out that Si is a donor and Ge is an acceptor in GaAs because it is energetically more
favorable for the small Si atoms to substitute for the small Ga atoms and for the
larger Ge to substitute for the larger As.
Ga
As
FIGURE 1–7 The GaAs crystal structure.
Ga As Ga
As Ga As
Ga As Ga
FIGURE 1–8 Bond model of GaAs.

Empty upper bands
Lowest (nearly) empty band

2p (conduction band)
Highest (nearly) filled band

2s (valence band)
Filled lower bands
(a) (b)
FIGURE 1–9 The discrete energy states of a Si atom (a) are replaced by the energy bands in
a Si crystal (b).
1.3 ● ENERGY BAND MODEL ●
While the bond model described in the previous section is conceptually simple, it is
not complete enough for understanding semiconductor devices. The most useful
model involves the concept of energy bands.
Recall that electrons in an atom occupy discrete energy levels as shown in
Fig. 1–9a. If two atoms are in close proximity, each energy level will split into two
due to the Pauli exclusion principle that states that each quantum state can be
occupied by no more than one electron in an electron system such as an atom
molecule, or crystal. When many atoms are brought into close proximity as in a
crystal, the discrete energy levels are replaced with bands of energy states
separated by gaps between the bands as shown in Fig. 1–9b. One may think of an
energy band as a semicontinuum of a very large number of energy states.
Naturally, the electrons tend to fill up the low energy bands first. The lower the
energy, the more completely a band is filled. In a semiconductor, most of the energy
bands will be basically totally filled (completely filled at absolute zero), while the
higher energy bands are basically totally empty. Between the (basically) totally filled
and totally empty bands lie two bands that are only nearly filled and nearly empty as
shown in Fig. 1–9b. They are of utmost interest to us. The top nearly filled band is called
the valence band and the lowest nearly empty band is called the conduction band. The
gap between them is called the band gap. The electrons in a totally filled band do not
have a net velocity and do not conduct current, just as the water in a totally filled bottle
does not slosh about. Similarly, a totally empty band cannot contribute to current
conduction. These are the reasons the valence band and the conduction band are the
only energy bands that contribute to current flows in a semiconductor.
1.3.1 Energy Band Diagram

Figure 1–10 is the energy band diagram of a semiconductor, a small portion of Fig. 1–9.
It shows the top edge of the valence band, denoted by Ev, and the bottom edge of
the conduction band, denoted by Ec . The difference between Ec and Ev is the band-
gap energy or energy gap, Eg . Clearly, Eg = Ec – Ev. For silicon, the energy gap is
about 1.1 eV. The electrons in the valence band are those associated with the
1.3 ● Energy Band Model 9
Conduction band
Ec
Band gap
Eg
Ev
Valence band
FIGURE 1–10 The energy band diagram of a semiconductor.
● Photoconductor as Light Detector ●

When light is absorbed by a semiconductor sample and electron–hole pairs are created
as shown in Fig. 1–11, the number of electrons and holes (and therefore the conductivity
of the semiconductor) increase in proportion to the light intensity. By putting two
electrodes on the semiconductor and applying a voltage between the electrodes, one
can measure the change in the semiconductor conductance and thus detect changes in
light intensity. This simple yet practical photodetector is called a photoconductor.
Electron
Ec
Photons
Eg
Photon energy : hv Eg
Ev
Hole
FIGURE 1–11 Eg can be determined from the minimum energy (hν) of photons that are
absorbed by the semiconductor.
covalent bonds in the bond model discussed in the previous section, and the
electrons in the conduction band are the conduction or mobile electrons. The band-
gap energies of some semiconductors are listed in Table 1–1 to illustrate their wide
range. The band-gap energy has strong influence on the characteristics and
performance of optoelectronic devices (see Section 4.12.4 and Table 4–1). By mixing
multiple semiconductors, the band-gap energy can be precisely tuned to desired
values. This is widely practiced for optical semiconductor devices (see Section 4.13).
The band-gap energy can be determined by measuring the absorption of light
by the semiconductor as a function of the photon energy, hv. The light is strongly
absorbed only when hv is larger than Eg . The absorbed photon energy is consumed
to create an electron–hole pair as shown in Fig. 1–12. As hv is reduced below Eg ,
the specimen becomes transparent to the light. Eg can be determined by observing
this critical hv. Values of Eg listed in Table 1–1 are basically obtained in this way.
TABLE 1–1 • Band-gap energies of selected semiconductors.
Semiconductor InSb Ge Si GaAs GaP ZnSe Diamond
Eg (eV) 0.18 0.67 1.12 1.42 2.25 2.7 6.0

EXAMPLE 1–1 Measuring the Band-Gap Energy

If a semiconductor is transparent to light with a wavelength longer than
0.87 µm, what is its band-gap energy?
SOLUTION:
Photon energy of light with 0.87 µm wavelength is, with c being the speed of
light
–34 8 –19
c 6.63 × 10 ( J ⋅ s ) × 3 × 10 m/s 1.99 × 10 ( J ⋅ µ m )
h ν = h --- = ------------------------------------------------------------------------------ = ----------------------------------------------------
λ 0.87 µ m 0.87 µ m
–19
1.99 × 10 ( eV ⋅ µ m ) 1.24 ( eV ⋅ µ m )
= -------------------------------------------------------- = -------------------------------------- = 1.42 eV
1.6 × 10 × 0.87 µ m
–19 0.87 µ m
Therefore, the band gap of the semiconductor is 1.42 eV. The semiconductor is
perhaps GaAs (see Table 1–1).
1.24
USEFUL RELATIONSHIP: h ν ( eV ) = -------------------
λ (µm)
The visible spectrum is between 0.5 and 0.7 µm. (Silicon and GaAs have band
gaps corresponding to the hv of infrared light. Therefore they absorb visible
light strongly and are opaque.) Some semiconductors such as indium and tin
oxides have sufficiently large Eg’s to be transparent to the visible light and be
used as the transparent electrode in LCD (liquid crystal display) flat panel
displays.
Conduction band
Ec
Ed
Donor level
Donor ionization
energy
Acceptor level
Ea
Ev
Valence band
FIGURE 1–12 Energy levels of donors and acceptors.
1.3.2 Donors and Acceptors in the Band Model

The concept of donors and acceptors is expressed in the energy band model in the
following manner. Although less important than Ec and Ev , two other energy
levels are present in the energy band diagram, the donor energy level Ed and the
acceptor energy level Ea (Fig. 1–12). Recall that it takes the donor ionization
energy (about 50 meV) to free the extra electron from the donor atom into a
conduction electron. Therefore, the donor electron, before the electron is freed,
must occupy a state at about 50 meV below Ec. That is to say, Ec – Ed is the donor
ionization energy. Similarly, Ea – Ev is the acceptor ionization energy (i.e., the
1.4 ● Semiconductors, Insulators, and Conductors 11
TABLE 1–2 • Ionization energy of selected donors and acceptors in silicon.
Donors Acceptors
Dopant Sb P As B Al In
Ionization energy, Ec – Ed or Ea – Ev (meV) 39 44 54 45 57 160
energy it takes for an acceptor atom to receive an extra electron from the valence
band, creating a hole there). Some donor and acceptor ionization energies in
silicon are listed in Table 1–2 for reference. As, P, Sb, and B are the most
commonly used dopants for silicon. Acceptor and donor levels with small
ionization energies, such as these four, are called shallow levels. Deep levels can be
created with other impurities such as copper and gold, and they affect silicon
properties in very different ways (see Section 2.6).
1.4 ● SEMICONDUCTORS, INSULATORS, AND CONDUCTORS ●
Based on the energy band model, we can now understand the differences among
semiconductors, insulators, and conductors. A semiconductor has a nearly filled
valence band and a nearly empty conduction band separated by a band gap as
illustrated in Fig. 1–13a. The band diagram of an insulator is similar to that of a
semiconductor except for a larger Eg , which separates a completely filled band and
a completely empty band (see Fig. 1–13b). Totally filled bands and totally empty
bands do not contribute to current conduction, just as there can be no motion of
liquid in totally filled jars and totally empty jars. A conductor has a quite different
energy band diagram. As depicted in Fig. 1–13c, a conductor has a partially filled
band. This is the conduction band of the conductor and it holds the conduction
electrons. The abundance of the conduction electrons makes the resistivity of a
typical conductor much lower than that of semiconductors and insulators.
Why do some materials have a partially filled band and are therefore
conductors? Each energy band can hold two electrons per atom.1 This is why
Ec
Top of
conduction band
Eg 9 eV Empty
Ec
Eg 1.1 eV Filled
Ev Ev Ec
(a) Si, semiconductor (b) SiO2, insulator (c) Conductor

FIGURE 1–13 Energy band diagrams for a semiconductor (a), an insulator (b),
and a conductor (c).
1 This is a simplified picture. Actually, each band can hold two electrons per primitive cell, which may
contain several atoms. A primitive cell is the smallest repeating structure that makes up a crystal. The
darkened part of Fig. 1–2 is the primitive cell of the Si crystal.
elemental solids with odd atomic numbers (and therefore odd numbers of
electrons) such as Au, Al, and Ag are conductors. Elements with even atomic
numbers such as Zn and Pb can still be conductors because a filled band and an
empty band may overlap in energy, thus leaving the combined band partially filled.
These elements are known as semimetals.
An insulator has a filled valence band and an empty conduction band that
are separated by a larger Eg. How large an Eg is large enough for the material to
be classified as an insulator? There is no clear boundary, although 4 eV would be
an acceptable answer. However, even diamond, with Eg ~ 6 eV (often cited as a
textbook example of an insulator) exhibits semiconductor characteristics. It can be
doped N type and P type, and electronic devices such as rectifiers and transistors
have been made with diamond.
One may say that semiconductors differ from insulators in that
semiconductors can be made N type or P type with low resistivities through
impurity doping. This characteristic of the semiconductors is very important for
device applications.
1.5 ● ELECTRONS AND HOLES ●
Although the term electrons can be used in conjunction with the valence band as in
“nearly all the energy states in the valence band are filled with electrons,” we
should assume that the term usually means conduction-band electrons. Holes are
the electron voids in the valence band. Electrons and holes carry negative and
positive charge (±q) respectively. As shown in Fig. 1–14, a higher position in the
energy band diagram represents a higher electron energy. The minimum
conduction electron energy is Ec. Any energy above Ec is the electron kinetic
energy. Electrons may gain energy by getting accelerated in an electric field and
may lose energy through collisions with imperfections in the crystal.
A lower location in the energy diagram represents a higher hole energy as
shown in Fig. 1–14. It requires energy to move a hole “downward” because that is
equivalent to moving an electron upward. Ev is the minimum hole energy. We may
think of holes as bubbles in liquid, floating up in the energy band. Similarly, one may
think of electrons as water drops that tend to fall to the lowest energy states in the
energy band.
1.5.1 Effective Mass

When an electric field, Ᏹ, is applied, an electron or a hole will accelerate according to
– qᏱ
Acceleration = ---------- electrons (1.5.1)
mn
qᏱ
Acceleration = ------- holes (1.5.2)
mp
In order to describe the motion of electrons and holes with the laws of motion of
the classical particles, we must assign effective masses (mn and mp) to them. The
electron and hole effective masses of a few semiconductors are listed in Table 1–3.
1.5 ● Electrons and Holes 13
Increasing electron energy

Electron kinetic energy
Increasing hole energy

Ec
Ev
Hole kinetic energy
FIGURE 1–14 Both electrons and holes tend to seek their lowest energy positions.
Electrons tend to fall in the energy band diagram. Holes float up like bubbles in water.
TABLE 1–3 • Electron and hole effective masses, mn and mp , normalized to the free
electron mass.
Si Ge GaAs InAs AlAs
mn /m0 0.26 0.12 0.068 0.023 2.0

mp /m0 0.39 0.30 0.50 0.30 0.3
The electrons and holes in a crystal interact with a periodic coulombic field in
the crystal. They surf over the periodic potential of the crystal, and therefore mn
and mp are not the same as the free electron mass.
A complete description of the electrons in a crystal must be based on their
wave characteristics, not just the particle characteristics. The electron wave function
is the solution of the three-dimensional Schrödinger wave equation [2].
h 2
– ---------- ∇ ψ + V ( r )ψ = E ψ (1.5.3)
2m 0
where h = h ⁄ 2 π is the reduced Planck constant, m0 is the free electron mass, V(r)
is the potential energy field that the crystal presents to the electron in the three-
dimensional space, and E is the energy of the electron. The solution is of the form
exp ( ± k ⋅ r ) , which represents an electron wave k, called the wave vector, is equal
at 2 π /electron wavelength and is a function of E. In other words, for each k there is
a corresponding E (see Fig. 4–27 for a schematic E–k diagram). It can further be
shown [2] that, assuming the E–k relationship has spherical symmetry, an electric
field, Ᏹ, would accelerate an electron wave packet with
2
qᏱ d E
Acceleration = – -------
2
---------- (1.5.4)
h dk 2
In order to interpret the acceleration in the form of F/m, it is convenient to
introduce the concept of the effective mass
2
h
Effective mass ≡ ------------------------ (1.5.5)
2 2
d E ⁄ dk
Each semiconductor material has a unique E–k relationship (due to the

unique V(r)) for its conduction band and another unique E–k relationship for its
valence band. Therefore, each semiconductor material has its unique mn and mp.
The values listed in Table 1–3 are experimentally measured values. These
values agree well with the effective masses obtained by solving the Schrödinger
wave equation with computers.
1.5.2 How to Measure the Effective Mass2

If you wonder how one may measure the effective mass of electrons or holes in a
semiconductor, let us study a powerful technique called cyclotron resonance.
Consider an electron in an N-type semiconductor located in a magnetic field,
B, as shown in Fig. 1–15. A moving electron will trace out a circular path in a plane
normal to B. (In addition, there may be electron motion parallel to B. We may
ignore this velocity component for the present analysis). The magnetic field exerts a
Lorentzian force of qvB, where v is the electron velocity and B is the magnetic flux
density. Equating this force to the centripetal force corresponding to the circular
motion with radius r, we obtain
2
mnv
------------ = qvB (1.5.6)
r
qBr
v = --------- (1.5.7)
mn
The frequency of the circular motion is
v qB
f cr = -------- = ------------- (1.5.8)
2πr 2 π mn
This is the cyclotron resonance frequency. Notice that the resonance frequency
is independent of r and ν. Now, if a circularly polarized electric field of the same
frequency fcr (typically in the gigahertz range) is applied to this semiconductor, the
Microwave
FIGURE 1–15 The motion of electrons in an N-type semiconductor in the presence of a

magnetic field, B, and a microwave with rotating electric field (the direction of rotation is
indicated by the arrow).
2 This section may be omitted in an accelerated course.

1.6 ● Density of States 15
electrons will strongly absorb the microwave energy. They do so by accelerating to a

higher velocity and tracing circles of increasing radius [see Eq. (1.5.7)] without
changing their frequency of circular motion [see Eq. (1.5.8)], losing the energies
through collisions, and starting the acceleration process again. Obviously, the
absorption would be much weaker if the frequency of the applied field is not equal
to fcr, i.e., when the applied field is out of sync with the electron motion.
By varying the frequency of the electric field or varying B until a peak in
absorption is observed, one can calculate mn using Eq. (1.5.8). One can also perform
the same measurement on a P-type semiconductor to measure the effective mass of
holes.
1.6 ● DENSITY OF STATES ●
It is useful to think of an energy band as a collection of discrete energy states.

Figure 1–16a emphasizes this picture. In quantum mechanics terms, each state
represents a unique spin (up and down) and unique solution to the Schrodinger’s
wave equation for the periodic electric potential function of the semiconductor [3].
Each state can hold either one electron or none. If we count the number of states in
a small range of energy, ∆Ε, in the conduction band, we can find the density of states:
number of states in ∆ E
D c ( E ) ≡ ------------------------------------------------------ (1.6.1)
∆ E × volume
This conduction-band density of states is a function of E (i.e., where ∆Ε is
located). Similarly, there is a valence-band density of states, Dv(E). Dc and Dv,
graphically illustrated in Fig. 1–16b, can be shown to be proportional to E – E c
and E v – E at least for a range of E. The derivation is presented in Appendix I,
“Derivation of the Density of States.”
8 π m n 2m n ( E – E c )
D c ( E ) = -----------------------------------------------
-, E ≥ Ec (1.6.2a)
3
h
8 π m p 2m p ( E v – E )
D v ( E ) = -----------------------------------------------
-, E ≤ Ev (1.6.2b)
3
h
E
Dc
E
Ec Ec
Ev Ev
Dv
(a) (b)
FIGURE 1–16 (a) Energy band as a collection of discrete energy states. (b) D is the density
of the energy states.
Dc(E) and Dv(E) have the dimensions of number per cubic centimeter per
electronvolt.
It follows from Eq. (1.6.1) that the product Dc(E) dE and Dv(E) dE are the
numbers of energy states located in the energy range between E and E + dE per cubic
centimeter of the semiconductor volume. We will use this concept in Section 1.8.1.
1.7 ● THERMAL EQUILIBRIUM AND THE FERMI FUNCTION ●
We have mentioned in Section 1.6 that most of the conduction-band electrons will
be found near Ec, where the electron energy is the lowest. Most holes will “float”
toward Ev, where the hole energy is the lowest. In this section, we will examine the
distribution of electrons and holes in greater detail.
1.7.1 An Analogy for Thermal Equilibrium

Let us perform the following mental experiment. Spread a thin layer of sand on the
bottom of a shallow dish sitting on a table as shown in Fig. 1–17. The sand particles
represent the electrons in the conduction band. A machine shakes the table and
therefore the dish up and down. The vibration of the dish represents the thermal
agitation experienced by the atoms, electrons and holes at any temperature above
absolute zero. The sand will move and shift until a more or less level surface is
created. This is the equilibrium condition. The equilibrium condition is the lowest
energy configuration in the presence of the thermal agitation. If a small sand dune
exists as a protrusion above the flat surface, it would not be the lowest energy
condition because the gravitational energy of the system can be reduced by
flattening the dune. This is the equivalent of electrons preferring to occupy the
lower energy states of the conduction band.
Notice, however, that the agitation prevents the particles from taking only the
lowest possible energy positions. This fact can be dramatized by vibrating the dish
more vigorously. Now, some sand particles jump up into the air and fall back. The
system is at the lowest possible energy consistent with the presence of the
“thermal” agitation. This is thermal equilibrium.
Because the sand particles bounce up and down, some higher energy states, in
the air, are occupied, too. There is a lower probability (smaller fraction of time) for
the sand particles to be higher in the air, i.e., to occupy higher energy states.
Similarly, electrons and holes in semiconductors receive and exchange energy from
Dish
Vibrating table
FIGURE 1–17 Elevations of the sand particles in the dish on a vibrating table represent the
energies of the electrons in the conduction band under the agitation of thermal energy.
1.7 ● Thermal Equilibrium and the Fermi Function 17
or with the crystal and one another and every energy state in the conduction and
valence bands has a certain probability of being occupied by an electron. This
probability is quantified in the next section.
1.7.2 Fermi Function—The Probability of an Energy State

Being Occupied by an Electron
One can derive an expression to describe the probability of finding an electron at a
certain energy. A statistical thermodynamic analysis without regard to the specifics
of how particles bounce off the atoms or one another and only assuming that the
number of particles and the total system energy are held constant yield the following
result. (The derivation is presented in Appendix II, “Derivation of the Fermi–Dirac
Distribution Function.)
1
f ( E ) = -------------------------------------- (1.7.1)
1 + e ( E – EF ) ⁄ kT
An equation that is highlighted with a box, such as Eq. (1.7.1), is particularly
important and often cited. Equation (1.7.1) is the Fermi function, or the
Fermi–Dirac distribution function, or the Fermi–Dirac statistics. EF is called the
Fermi energy or the Fermi level. f(E) is the probability of a state at energy E being
occupied by an electron. Figure 1–18 depicts the Fermi function. At large E (i.e.,
E – EF >> kT) the probability of a state being occupied decreases exponentially
with increasing E. In this energy region, Eq. (1.7.1) can be approximated with
f ( E ) ≈ e –( E – EF ) ⁄ kT (1.7.2)
Equation (1.7.2) is known as the Boltzmann approximation. In the low energy

region (i.e., E – EF << –kT), the occupation probability approaches 1. In other words,
the low energy states tend to be fully occupied. Here Eq. (1.7.1) can be approximated as
–( EF – E) ⁄ kT
f( E) ≈ 1 – e (1.7.3)
f(E) ⬵ e(E EF)/kT

EF 3kT
EF 2kT
EF kT
EF
EF kT
EF 2kT
EF 3kT
f(E) ⬵ 1 e(EF E)/kT
f(E)
0.5 1
FIGURE 1–18 The Fermi function diagram. For an interactive illustration of the Fermi function,
see http://jas.eng.buffalo.edu/education/semicon/fermi/functionAndStates/functionAndStates.html
In this energy region, the probability of a state not being occupied, i.e., being
occupied by a hole is
–( EF – E) ⁄ kT
1 – f(E) ≈ e (1.7.4)
From Eq. (1.7.1) and Fig. 1–18, the probability of occupation at EF is 1/2. The
probability approaches unity if E is much lower than EF , and approaches zero at E
much higher than EF. A very important fact to remember about EF is that there is
only one Fermi level in a system at thermal equilibrium. This fact will be used often
in the rest of this book. Later, you will see that the value of EF depends on the
number of electrons or holes present in the system.
● What Determines EF ? ●
This example is designed to show that the Fermi level, EF, is determined by the available
electrons and states in the system. Figure 1–19 shows the energy states of an electron
system at room temperature. Each energy state can hold either one electron or none, i.e.,
be either occupied or empty. It is known that there is only one electron in this system.
Since there is only one electron in the system, the sum of the probabilities that each state
is occupied must be equal to 1. By trial and error, convince yourself that only one specific
EF, somewhere above E1 and below E2, can satisfy this condition. Do not calculate this
value. Hint: Ask yourself how many electrons will be in the system if EF is, say, above E3
and how many if EF is below E1.
E3 0 eV
E2 1 eV
EF
E1 4 eV
FIGURE 1–19 A simple electron system at room temperature for illustration of what
determines the Fermi energy, EF.
EXAMPLE 1–2 Oxygen Concentration versus Altitude

We all know that there is less oxygen in the air at higher altitudes. What is the
ratio of the oxygen concentration at 10 km above sea level, Nh, to the
concentration at sea level, N0, assuming a constant temperature of 0°C?
SOLUTION:
There are fewer oxygen molecules at higher altitudes because the
gravitational potential energy of an oxygen molecule at the higher altitude,
Eh, is larger than at sea level, E0. According to Eq. (1.7.2)
–E ⁄ kT
Nh e h –( Eh – E0 ) ⁄ kT
------- = ------------------- = e
N0 – E 0 ⁄ kT
e
1.8 ● Electron and Hole Concentrations 19
E0 –Eh is the potential energy difference, i.e., the energy needed to lift an
oxygen molecule from sea level to 10 km.
E h – E 0 = altitude × weight of O 2 molecule × acceleration of gravity

4 –2
= 10 m × O 2 molecular weight × atomic mass unit × 9.8 m ⋅ s
4 –27 –2
= 10 m × 32 × 1.66 × 10 kg × 9.8 m ⋅ s
–21
= 5.2 × 10 J
N – 21 – 10 –1
–5.2 ×10 J ⁄ 1.38 ×10 J ⋅ K × 273K
∴-----h- = e
N0
–1.38
= e = 0.25
So, the oxygen concentration at 10 km is 25% of the sea level concentration.

This example and the sand-in-a-dish analogy are presented to demystify
the concept of equilibrium, and to emphasize that each electron energy state
has a probability of being occupied that is governed by the Fermi function.
Additional question: See Problem 1.4 for a follow-up question.
1.8 ● ELECTRON AND HOLE CONCENTRATIONS ●
We have stated that if a semiconductor is doped with 1016 donors per cubic
centimeter, the electron concentration will be 1016cm–3. But, what would the hole
concentration be? What are the carrier concentrations in undoped semiconductors?
These questions will be answered after the important relationships between the
Fermi level and the carrier concentrations are derived in this section.
1.8.1 Derivation of n and p from D(E) and f(E)

First, we will derive the concentration of electrons in the conduction band, known
as the electron concentration. Since Dc(E) dE is the number of energy states
between E and E + dE for each cubic centimeter, the product f(E)Dc(E) dE is then
the number of electrons between E and E + dE per cubic centimeter of the
semiconductor. Therefore, the number of electrons per cubic centimeter in the
entire conduction band is
Top of conduction band
n = ∫E f ( E ) D c ( E ) dE (1.8.1)
c
Graphically, this integration gives the shaded area in Fig. 1–20, which yields the
density of electrons. We now substitute Eqs. (1.6.2a) and (1.7.2) into Eq. (1.8.1) and
set the upper limit of integration at infinity. Resetting the upper limit is acceptable
because of the rapid fall of f(E) with increasing E as shown in Fig. 1–20. This allows
us to obtain a closed form expression for n
Energy band Density Occupancy Carrier

diagram of states probability distributions
E E E E
Dc(E)
Electrons
Ec
EF EF
Eg
Ev
Holes
Dv(E)
0.5 1
Electron N(E) f(E)
Hole
FIGURE 1–20 Schematic band diagram, density of states, Fermi–Dirac distribution,
and carrier distributions versus energy.
8 π m n 2m n ∞ – ( E – EF ) ⁄ kT
3
-∫
n = --------------------------- E – Ec e dE (1.8.2)
h E c
8 π m n 2m n – ( Ec – EF ) ⁄ kT ∞ – ( E – Ec ) ⁄ kT
= ---------------------------
-e ∫0 E – Ec e d( E – E c ) (1.8.3)
3
h
Introducing a new variable
x = ( E – E c ) ⁄ kT
reveals that the integral in Eq. (1.8.3) is of a form known as a gamma function and
is equal to π / 2, i.e.,
∞ –x
∫0 x e dx = π⁄2 (1.8.4)
Applying Eq. (1.8.4) to Eq. (1.8.3) leads to the following two equations:
– ( Ec – EF ) ⁄ kT
n = Nc e (1.8.5)
2 π m n kT 3⁄2
N c ≡ 2 -------------------- (1.8.6)
2
h
Nc is called the effective density of states. Equation (1.8.5) is an important equation

and should be memorized. It is easy to remember this equation if we understand
why Nc is called the effective density of states. It is as if all the energy states in the
TABLE 1–4 • Values of Nc and Nv for Ge, Si, and GaAs at 300 K.
Ge Si GaAs
Nc (cm–3) 1.04 × 1019 2.8 × 1019 4.7 × 1017

Nv (cm–3) 6.0 × 1018 1.04 × 1019 7.0 × 1018
conduction band were effectively squeezed into a single energy level, Ec, which can
hold Nc electrons (per cubic centimeter). As a result, the electron concentration in
Eq. (1.8.5) is simply the product of Nc and the probability that an energy state at Ec
is occupied.
An expression for the hole concentration can be derived in the same way. The
probability of an energy state being occupied by a hole is the probability of it not
being occupied by an electron, i.e., 1– f (E). Therefore,
Ev
p = ∫ D v ( E ) (1 – f ( E ) ) dE (1.8.7)
Valence band
bottom
Substituting Eqs. (1.6.2b) and (1.7.4) into Eq. (1.8.7) yields
– ( EF – Ev ) ⁄ kT
p = Nv e (1.8.8)
2 π m p kT 3⁄2
N v ≡ 2 -------------------- (1.8.9)
2
h
Nv is the effective density of states of the valence band. (The full name of Nc is the
effective density of states of the conduction band.) The values of Nc and Nv , both
about 1019cm–3, differ only because mn and mp are different. Nc and Nv vary
somewhat from one semiconductor to another because of the variation in the
effective masses, too. Nc and Nv for Ge, Si, and GaAs are listed in Table 1–4.3
1.8.2 Fermi Level and the Carrier Concentrations

We will use Eqs. (1.8.5) and (1.8.8) time and again. Right now, they can help to
remove the mystery of the Fermi level by linking EF to the electron and hole
concentrations.
Figure 1–21 shows the location of the Fermi level as a function of the carrier
concentration. Note that the solid lines stop when EF is about 20 meV (~kT) from
3 The effective mass in Table 1–3 is called the conductivity effective mass and is an average over quantum
mechanical wave vectors appropriate for describing carrier motions [3]. The effective mass in Eqs. (1.6.2a)
and (1.6.2b), (1.8.6), and (1.8.9) is called the density-of-states effective mass and has a somewhat different
value (because it is the result of a different way of averaging that is appropriate for describing the density
of states).
Ec
300 K
400 K
r-dop ed
EF , Dono
EF , Ac
ceptor-d
oped
400 K
300 K
Ev
1013 1014 1015 1016 1017 1018 1019 1020
3
Na or Nd (cm )
FIGURE 1–21 Location of Fermi level vs. dopant concentration in Si at 300 and 400 K.
EXAMPLE 1–3 Finding the Fermi Level in Si

Where is EF located in the energy band of silicon, at 300K with n = 1017cm–3?
And for p = 1014cm–3?
SOLUTION: From Eq. (1.8.5)
E c – E F = kT ⋅ ln ( N c ⁄ n )
19 17
= 0.026 ln ( 2.8 × 10 ⁄ 10 )
= 0.146 eV
Therefore, EF is located at 146 meV below Ec, as shown in Fig. 1–22a.
For p = 1014cm–3, from Eq. (1.8.8),
E F – E v = kT ⋅ ln ( N v ⁄ p )
19 14
= 0.026 ln ( 1.04 × 10 ⁄ 10 )
= 0.31 eV
Therefore EF is located at 0.31 eV above Ev.
0.146 eV
Ec Ec
EF
EF
0.31 eV
Ev Ev
(a) (b)
FIGURE 1–22 Location of EF when n = 1017cm–3 (a), and p = 1014cm–3 (b).

Ec or Ev. Beyond this range, the use of the Boltzmann approximation in the
derivation of Eqs. (1.8.5) and (1.8.8) is not quantitatively valid. Those equations are
not accurate when the semiconductor is heavily doped (>1019cm–3) or degenerate.
Please remember that n decreases as EF moves farther below Ec , and vice versa; p
decreases as EF moves farther above Ev.
1.8.3 The np Product and the Intrinsic Carrier Concentration

Since EF cannot be close to both Ec and Ev, n and p cannot both be large numbers
at the same time. When Eqs. (1.8.5) and (1.8.8) are multiplied together, we obtain
– ( Ec – Ev ) ⁄ kT – Eg ⁄ kT
np = N c N v e = NcNve (1.8.10)
Equation (1.8.10) states that the np product is a constant for a given semiconductor
and T, independent of the dopant concentrations. It is an important relationship
and is usually expressed in the following form:
2
np = n i (1.8.11)
– Eg ⁄ 2kT
ni = Nc Nve (1.8.12)
According to Eq. (1.8.11), there are always some electrons and holes
present—whether dopants are present or not. If there are no dopants present, the
semiconductor is said to be intrinsic. In an intrinsic semiconductor, the nonzero n
and p are the results of thermal excitation, which moves some electrons from the
valence band into the conduction band. Since such movements create electrons and
holes in pairs, n = p in intrinsic semiconductors. This fact and Eq. (1.8.11)
immediately suggest that, in intrinsic semiconductors,
n = p = ni (1.8.13)
Therefore, ni is called the intrinsic carrier concentration. ni is a strong function of

Eg and T according to Eq. (1.8.12), but is independent of the dopant
concentration. ni at room temperature is roughly 1010cm–3 for Si and 107cm–3 for
GaAs, which has a larger band gap than Si. For silicon, the np product is
therefore 1020cm–6 regardless of the conductivity type (P type or N type) and
the dopant concentrations.
One may explain why the np product is a constant this way: the electron–hole
recombination rate is proportional to the np product. When np = ni2, the
recombination rate happens to be equal to the rate of thermal generation of
electron–hole pairs. This is the same mass action principle that keeps the product of
the concentrations of [H+] and [OH–] constant in aqueous solutions, whether
strongly acidic, strongly alkaline, or neutral.
EXAMPLE 1–4 Carrier Concentrations

What is the hole concentration in an N-type semiconductor with
QUESTION:
15 –3
10 cm of donors?
SOLUTION: For each ionized donor, an electron is created. Therefore n = 1015cm–3.
2 20 –3
n 10 cm 5 –3
p = ----i- ≈ ---------------------- = 10 cm
n 10 15 cm –3
With a modest temperature increase of 60°C, n remains the same at 1015cm–3,

while p increases by about a factor of 2300 because ni2 increases according to
Eq. (1.8.12).
QUESTION: What is n if p = 1017cm–3 in a P-type silicon wafer?
2 20
n 10 3 –3
SOLUTION: n = ----i- = --------- = 10 cm
p 17
10
The electron and hole concentrations in a semiconductor are usually very

different. In an N-type semiconductor, the abundant electrons are called the majority
carriers and the almost nonexistent holes are the minority carriers. In P-type
semiconductors, the holes are the majority carriers and the electrons are the minority
carriers.
EXAMPLE 1–5 The Intrinsic Fermi Level

In an intrinsic semiconductor, n = p. Therefore Ec – EF ≈ EF – Ev and the
Fermi level is nearly at the middle of the band gap, i.e., EF ≈ Ec – Eg/2. This
level is called the intrinsic Fermi level, Ei. Here we derive a more exact
expression for Ei. Rewriting Eq. (1.8.12) for lnni, yields
ln n i = ln N c N v – E g ⁄ 2kT
Writing Eq. (1.8.5) for the intrinsic condition where n = ni ,

– ( Ec – Ei ) ⁄ kT
ni = Nc e (1.8.14)
N E N
∴ E i = E c – kT ln ------c = E c + kT ln n i – kT ln N c = E c – -----g- – kT ln ------c (1.8.15)
ni 2 Nv
We see that Ei would be at the midgap, Ec – Eg/2, if Nc = Nv. For silicon, Ei is

very close to the midgap and the small last term in Eq. (1.8.15) is only of
academic interest.
1.9 ● General Theory of n and p 25
1.9 ● GENERAL THEORY OF n AND p ●
The shallow donor and acceptor levels (Ed and Ea) in Fig. 1–12 are energy states
and their occupancy by electrons is governed by the Fermi function. Since Ed is
usually a few kT above EF, the donor level is nearly empty of electrons. We say that
nearly all the donor atoms are ionized (have lost the extra electrons). Similarly, all
the acceptor atoms are ionized. For simplicity, we can assume that all shallow donors
and acceptors are ionized.
EXAMPLE 1–6 Complete Ionization of the Dopant Atoms

In a silicon sample doped with 1017cm–3 of phosphorus atoms, what fraction of
the donors are not ionized (i.e., what fraction are occupied by the “extra”
electrons)?
SOLUTION:
First assume that all the donors are ionized and each donor donates an
electron to the conduction band.
17 –3
n = N d = 10 cm
From Fig. 1–20, Example 1–3, EF is located at 146 meV below Ec. The
donor level Ed is located at 45 meV below Ec for phosphorus (see Table 1–2
and Figure 1–23).
The probability that a donor is not ionized, i.e., the probability that it is
occupied by the “extra” electron, according to Eq. (1.7.1), is
1
Probability of non-ionization ≈ -------------------------------------------
1 ( Ed –EF ) ⁄ kT
1 + --- e
2
1
------------------------------------------------------------------- = 3.9%
1 ( ( 146 –45) meV) ⁄ 26meV
1 + --- e
2
(The factor 1/2 in the denominators stems from the complication that a donor
atom can hold an electron with upspin or downspin. This increases the
probability that a donor state is occupied by an electron.)
Therefore, it is reasonable to assume complete ionization, i.e., n = Nd .
45 meV 146 meV

E d Ec
EF
Ev
FIGURE 1–23 Location of EF and Ed. Not to scale.

DISCUSSION: You may have noticed that as Nd increases, EF rises toward Ed

and the probability of nonionization can become quite large. In reality, the
impurity level broadens into an impurity band that merges with the
conduction band in heavily doped semiconductor (i.e., when donors or
acceptors are close to one another). This happens for the same reason energy
levels broaden into bands when atoms are brought close to one another to
form a crystal (see Fig. 1–9). The electrons in the impurity band are also in the
conduction band. Therefore, the assumption of n = Nd (or complete ionization)
is reasonable even at very high doping densities. The same holds true in P-type
materials.
There are four types of charged species in a semiconductor: electrons, holes,

positive donor ions, and negative acceptor ions. Their densities are represented by
the symbols n, p, Nd, and Na. In general, all samples are free of net charge. Charge
neutrality requires that the densities of the negative particles and positive particles
are equal:
n + Na = p + Nd (1.9.1)
Equations (1.8.11) and (1.9.1) can be solved for n and p:
Nd – Na N d – N a 2 1/2
- +  ------------------
2
n = ------------------ - + ni (1.9.2a)
2  2 
Na – Nd N a – N d 2 1/2
- +  ------------------
2
p = ------------------ - + ni (1.9.2b)
2  2 
Although it is interesting to know that n and p can be calculated for arbitrary

values of Na and Nd, the complicated Eq. (1.9.2) is rarely used. Instead, one of the
following two cases is almost always valid:
1. Nd – Na >> ni (i.e., N type),
n = Nd – Na (1.9.3a)
2
p = ni ⁄ n (1.9.3b)
If, furthermore, Nd >> Na, then

2
n = Nd and p = ni ⁄ Nd (1.9.4)
2. Na – Nd >> ni (i.e., P type),
p = Na – Nd (1.9.5a)
2
n = ni ⁄ p (1.9.5b)
1.9 ● General Theory of n and p 27
If, furthermore, Na >> Nd, then

2
p = Na and n = ni ⁄ Na (1.9.6)
We have intuitively assumed Eqs. (1.9.4) and (1.9.6) to be true in the

previous sections. It is worthwhile to remember that Eqs. (1.9.3) and (1.9.5) as the
more exact expressions. We see that an acceptor can effectively negate the effect
of a donor in Eq. (1.9.3a) and vice versa in Eq. (1.9.5a). This fact is known as
dopant compensation. One can even start with P-type Si and convert a portion of
it into N-type simply by adding enough donors. This is one of the techniques
employed to make complex devices.
EXAMPLE 1–7 Dopant Compensation

What are n and p in a Si sample with Nd = 6 × 1016cm–3 and Na = 2 × 1016cm–3 ?
With additional 6 × 1016cm–3 of acceptors?
SOLUTION: As shown in Fig. 1–24a:
16 –3
n = N d – N a = 4 × 10 cm
2 20 16 3 –3
p = n i ⁄ n = 10 ⁄ 4 × 10 = 2.5 × 10 cm
With the additional acceptors, Na = 2 × 1016 + 6 × 1016 = 8 × 1016cm–3, holes

become the majority,
16 16 16 –3
p = N a – N d = 8 × 10 – 6 × 10 = 2 × 10 cm
2 20 16 3 –3
n = n i ⁄ p = 10 ⁄ ( 2 × 10 ) = 5 × 10 cm
The addition of acceptors has converted the Si to P-type as shown in Fig. 1–24b.
n 4 1016 cm3
Nd 6 1016 cm3 Nd 6 1016 cm3
Na 8 1016 cm3
Na 2 1016 cm3
p 2 1016 cm3
(a) (b)
Electron; Hole; donor ions; acceptor ions
FIGURE 1–24 Graphical illustration of dopant compensation.
1.10 ● CARRIER CONCENTRATIONS AT EXTREMELY HIGH

AND LOW TEMPERATURES4 ●
At very high temperatures, ni is large [see Eq. (1.8.12)], and it is possible to have
ni >> N d – N a . In that case, Eq. (1.9.2) becomes
n = p = ni (1.10.1)
In other words, the semiconductor becomes “intrinsic” at very high temperatures.

At the other extreme of very low temperature, EF may rise above Ed, and
most of the donor (or acceptor, in the case of P-type material) atoms can remain
nonionized. The fifth electrons stay with the donor. This phenomenon is called
freeze-out. In this case, if the doping is not heavy enough to form an impurity band
(see Section 1.9), the dopants are not totally ionized. The carrier concentration may
be significantly less than the dopant concentration. The exact analysis is
complicated, but the result is [4]
Nc Nd 1⁄2
n = ------------- e – ( Ec –Ed ) ⁄ 2kT (1.10.2)
2
Freeze-out is a concern when semiconductor devices are operated at, for example,
the liquid–nitrogen temperature (77 K) in order to achieve low noise and high
speed.
Figure 1–25 summarizes the temperature dependence of majority carrier
concentrations. The slope of the curve in the intrinsic regime is Eg/2k, and the slope
in the freeze-out portion is (Ec – Ed)/2k (according to Eq. (1.10.2)). These facts may
be used to determine Eg and Ec – Ed.
Intrinsic regime
n Nd
ln n
Freeze-out regime
1/T
High Room Cryogenic
temp- temperature temperature
erature
FIGURE 1–25 Variation of carrier concentration in an N-type semiconductor over a wide
range of temperature.
4 This section may be omitted in an accelerated course.

1.11 ● Chapter Summary 29
● Infrared Detector Based on Freeze-Out ●

Often it is desirable to detect or image the black-body radiation emitted by warm
objects, e.g., to detect tumors (which restrict blood flow and produce cold spots), to
identify inadequately insulated building windows, to detect people and vehicles at night,
etc. This requires a photodetector that responds to photon energies around 0.1 eV. For
this purpose, one can use a semiconductor photoconductor with Eg less than 0.1 eV, such
as HgPbTe operating in the mode shown in Fig. 1–11. Alternatively, one can use a more
common semiconductor such as doped Si operating in the freeze-out mode shown in
Fig. 1–26. In Fig. 1–26, conduction electrons are created when the infrared photons
provide the energy to ionize the donor atoms, which are otherwise frozen-out. The
result is a lowering of the detector’s electrical resistance, i.e., photoconductivity.
At long enough wavelength or low enough photon energy hν, light will no
longer be absorbed by the specimen shown in Fig. 1–26. That critical hν corresponds
to Ec – Ed. This is a method of measuring the dopant ionization energy, Ec – Ed.
Electron
Photon Ec
Ed
Ev
FIGURE 1–26 Infrared photons can ionize the frozen-out donors and produce conduction
electrons.
1.11 ● CHAPTER SUMMARY ●
In a silicon crystal, each Si atom forms covalent bonds with its four neighbors. In an
intrinsic Si crystal, there are few mobile electrons and holes. Their concentrations
are equal to ni (~1010cm–3 for Si), the intrinsic carrier concentration. The bond
model explains why group V atoms can serve as donors and introduce electrons,
and group III atoms can serve as acceptors and introduce holes. Electrons and holes
should be thought of as particles of equal importance but carrying negative and
positive charge, respectively, and having effective masses, mn and mp, which are a
fraction of the free electron mass.
The band model is needed for quantitative analysis of semiconductors and
devices. The valence band and conduction band are separated by an energy gap.
This band gap is 1.12 eV for Si. Thermal agitation gives each energy state a certain
probability of being occupied by an electron. That probability is expressed by the
Fermi function, which reduces to simple exponential functions of (E – EF)/kT for E
more than a few kT’s above the Fermi level, EF. EF is related to the density of
electrons and holes in the following manner:
– ( Ec – EF ) ⁄ kT
n = Nc e (1.8.5)
– ( EF – Ev ) ⁄ kT
p = Nv e (1.8.8)
The effective densities of states, Nc and Nv, are around 1019cm–3. The majority
carrier concentrations are
n = Nd – Na for N-type semiconductor (1.9.3a)
p = Na – Nd for P-type semiconductor (1.9.5a)

where Nd and Na are the concentrations of donors and acceptors, which are usually
assumed to be completely ionized. The minority carrier concentrations can be
found from
2
np = ni (1.8.11)
ni is the intrinsic carrier concentration, about 1010cm–3 for Si at 300 K. It is a

function of Eg and T.
You are now ready to study how electrons and holes move and produce
current in the next chapter.
● PROBLEMS ●
● Visualization of the Silicon Crystal ●
1.1 (a) How many silicon atoms are there in each unit cell?
(b) How many silicon atoms are there in one cubic centimeter?
(c) Knowing that the length of a side of the unit cell (the silicon lattice constant) is
5.43 Å, Si atomic weight is 28.1, and the Avogdaro’s number is 6.02 × 1023
atoms/mole, find the silicon density in g/cm3.
● Fermi Function ●
1.2 (a) Under equilibrium condition, what is the probability of an electron state being
occupied if it is located at the Fermi level?
(b) If EF is positioned at Ec, determine the probability of finding electrons in states at
Ec + kT. (A numerical answer is required.)
(c) The probability of a state being filled at Ec + kT is equal to the probability of a
state being empty at Ec + 3 kT. Where is the Fermi level located?
1.3 (a) What is the probability of an electron state being filled if it is located at the Fermi
level?
(b) If the probability that a state being filled at the conduction band edge (Ec) is
precisely equal to the probability that a state is empty at the valence band edge
(Ev), where is the Fermi level located?
(c) The Maxwell–Boltzmann distribution is often used to approximate the Fermi-
Dirac distribution function. On the same set of axes, sketch both distributions
as a function of (E – EF)/kT. Consider only positive values of E – EF . For what
range of (E – EF)/kT is the Maxwell–Boltzmann approximation accurate to
within 10%?
Problems 31
1.4 Refer to the oxygen concentration example in Sec. 1.7.2.

(a) Given that nitrogen is lighter in weight than oxygen, is N2 concentration at 10 km
more or less that 25% of the sea level N2 concentration?
(b) What is the ratio of N2 concentration to O2 at 10 km? At sea level, the ratio is 4 to 1.
1.5 Show that the probability of an energy state being occupied ∆E above the Fermi level is
the same as the probability of a state being empty ∆E below the Fermi level.
f(EF + ∆E) = 1 – f(EF – ∆E)
1.6 (a) Sketch the Fermi–Dirac distribution f(E) at room temperature (300 K) and at a
lower temperature such as 150 K. (Qualitative hand drawing.)
(b) The state distribution in a system is given in Fig. 1–27, where each circle represents
two electron states (one is spin-up; one is spin-down). Each electron state can be
occupied by one electron. There is no state below Emin. The Fermi level at 0 K is
given in Fig. 1–27. How many electrons are there in the system?
Energy
EF
Emin
FIGURE 1–27
● Energy: Density of States ●
1.7 The carrier distributions in the conduction and valence bands were noted to peak at
energies close to the band edges. (Refer to carrier distribution in Fig. 1–20.) Using
Boltzmann approximation, show that the energy at which the carrier distribution peaks
is Ec + kT/2 and Ev – kT/2 for the conduction and valence bands, respectively.
1.8 For a certain semiconductor, the densities of states in the conduction and valance bands are
constants A and B, respectively. Assume non-degeneracy, i.e., EF is not close to Ec or Ev.
(a) Derive expressions for electron and hole concentrations.
(b) If A = 2B, determine the location of the intrinsic Fermi energy (Ei) at 300 K with
respect to the mid-bandgap of the semiconductor.
Hint: These relationships may be useful:
∞ –x
∫0 x n – 1 e dx = Γ ( n ) (Gamma function)
Γ(2) = Γ(1) = 1, Γ(3) = 2, Γ(4) = 6

Γ(1/2) = π , Γ(3/2) = 1 ⁄ 2 π , Γ(5/2) = 1 ⁄ 3 π .
1.9 For a certain semiconductor, the densities of states in the conduction and valence bands
are: Dc(E) = A · (E – Ec) · u(E – Ec) and Dv(E) = B · (Ev – E) · u(Ev – E), respectively.
u(x), the unit step function, is defined as u(x) = 0 if x < 0 and u(x) = 1 if x > 0. Assume
nondegeneracy, i.e. not too highly doped. You may find this fact useful:
∞ –x
∫0 x e dx = 1
(a) Derive expressions for electron and hole concentrations as functions of the Fermi
energy, EF.
(b) If A = 2B, compute the intrinsic Fermi energy at 300 K.
–( E – Ef ) ⁄ kT
1.10 The Maxwell–Boltzmann distribution function f ( E ) = e is often used as an
approximation to the Fermi–Dirac function. Use this approximation and the densities
1⁄2
of the states in the conduction band D c ( E ) = A ( E – E c ) to find:
(a) The energy at which one finds the most electrons (1/cm³ · eV).
(b) The conduction-band electron concentration (explain any approximation made).
(c) The ratio of the peak electron concentration at the energy of (a) to the electron
concentration at E = Ec + 40 kT (about 1eV above Ec at 300 K). Does this result
justify one of the approximations in part(b)?
(d) The average kinetic energy, E – Ec of the electrons.
Hint: These relationships may be useful:
∞ –x
∫0 x n – 1 e dx = Γ ( n ) (Gamma function)
Γ(2) = Γ(1) = 1, Γ(3) = 2, Γ(4) = 6

Γ(1/2) = π , Γ(3/2) = 1 ⁄ 2 π , Γ(5/2) = 3 ⁄ 4 π .
● Electron and Hole Concentrations ●
1.11 (a) The electron concentration in a piece of Si at 300 K is 105 cm–3. What is the hole
concentration?
(b) A semiconductor is doped with impurity concentrations Nd and Na such that
Nd – Na >> ni and all the impurities are ionized. Determine n and p.
(c) In a silicon sample at T = 300 K, the Fermi level is located at 0.26 eV (10 kT) above
the intrinsic Fermi level. What are the hole and electron concentrations?
(d) What are the hole and electron concentration at T = 800 K for the sample in part
(c), and where approximately is EF? Comment on your results.
● Nearly Intrinsic Semiconductor ●
1.12 For a germanium sample at room temperature, it is known that ni = 1013 cm–3, n = 2p,
and Na = 0. Determine n and Nd.
1.13 Boron atoms are added to a Si film resulting in an impurity density of 4 × 1016 cm–3.
(a) What is the conductivity type (N-type or P-type) of this film?
(b) What are the equilibrium electron and hole densities at 300 K and 600 K?
(c) Why does the mobile carrier concentration increase at high temperatures?
(d) Where is the Fermi level located if T = 600 K?
References 33
● Incomplete Ionization of Dopants and Freeze-Out ●
1.14 Suppose you have samples of Si, Ge, and Ge, and GaAs at T = 300 K, all with the same
doping level of Nd+ – Na– = 3 × 1015 /cm3. Assuming all dopants are ionized, for which
material is p most sensitive to temperature (the sensitivity of p is define by δp/δT)? What is
your conclusion regarding the relation between Eg and temperature sensitivity of minority
carrier concentration? Repeat the problem using (δp/δT)/p as the definition of sensitivity.
1.15 An N-type sample of silicon has uniform density (Nd = 1019/cm–3) of arsenic, and a P-type
silicon sample has a uniform density (Na = 1015 /cm–3) of boron. For each sample, determine
the following:
(a) The temperature at which the intrinsic concentration ni exceeds the impurity
density by factor of 10.
(b) The equilibrium minority-carrier concentrations at 300 K. Assume full ionization
of impurities.
(c) The Fermi level relative to the valence–band edge Ev in each material at 300 K.
(d) The electron and hole concentrations and the Fermi level if both types of
impurities are present in the same sample.
1.16 A silicon sample is doped with Nd = 1017cm–3 of As atoms.
(a) What are the electron and hole concentrations and the Fermi level position
(relative to Ec or Ev) at 300 K? (Assume full ionization of impurities.)
(b) Check the full ionization assumption using the calculated Fermi level, (i.e., find
the probability of donor states being occupied by electrons and therefore not
ionized.) Assume that the donor level lies 50 meV below the conduction band,
(i.e., Ec – ED = 50 meV.)
(c) Repeat (a) and (b) for Nd = 1019cm–3. (Discussion: when the doping concentration is
high, donor (or acceptor) band is formed and that allows all dopant atoms to
contribute to conduction such that “full ionization” is a good approximation after all).
(d) Repeat (a) and (b) for Nd = 1017cm-3 but T = 30K. (This situation is called dopant
freeze-out.)
1.17 Given N-type silicon sample with uniform donor doping of (a) Nd = 1018/cm3, (b) Nd =
1019/cm3, and (c) Nd = 106/cm3, calculate the Fermi levels at room temperature
assuming full ionization for all cases. Check whether the above assumption of full
ionization of each case is correct with the calculated Fermi level. When this is not
correct, what is the relative position of EF and ED? Assume that
Ec – ED = 0.05 eV.
● REFERENCES ●
1. Shockley, W. Electrons and Holes in Semiconductors. Princeton, NJ: Van Nostrand, 1950.
2. Shur, M. Physics of Semiconductor Devices. Englewood Cliffs, NJ: Prentice-Hall, Inc., 1990.
3. Neamen, D. Semiconductor Physics and Devices, 3rd ed. New York: McGraw-Hill, 2003.
4. Smith, R.A. Semiconductors, 2nd ed. London: Cambridge University Press, 1979.
5. Streetman, B.G. Solid State Electronic Devices, 6th ed. Upper Saddle River, NJ: Prentice-
Hall, Inc., 2006.
6. Pierret, R. F. Modular Series on Solid State Devices, Vol. I. Reading, MA: Addison-Wesley
Publishing Co., 1983.
7. Sze, S. M. Physics of Semiconductor Devices, 2nd ed. New York: John Wiley & Sons, 1981.
8. Taur, Y., and T. H. Ning. Fundamentals of Modern VLSI Devices. Cambridge, UK:
Cambridge University Press, 1998.
● GENERAL REFERENCES ●
1. Neamen, D. Semiconductor Physics and Devices, 3rd ed. New York: McGraw-Hill, 2003.
2. Streetman, B.G. Solid State Electronic Devices, 6th ed. Upper Saddle River, NJ: Prentice-
Hall, Inc., 2006.

Chenming Hu - ch1 3

Uploaded by

Copyright:

Available Formats

Chenming Hu - ch1 3

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chenming Hu - ch1 3

Uploaded by

Copyright:

Available Formats

Hu_ch01v4.

fm Page 1 Thursday, February 12, 2009 10:14 AM

1.1 ● SILICON CRYSTAL STRUCTURE ●

A crystalline solid consists of atoms arranged in a repetitive structure. The

2 Chapter 1 ● Electrons and Holes in Semiconductors

● Inventors of the Transistor ●

1.1 ● Silicon Crystal Structure 3

4 Chapter 1 ● Electrons and Holes in Semiconductors

1.2 ● BOND MODEL OF ELECTRONS AND HOLES ●

1.2 ● Bond Model of Electrons and Holes 5

6 Chapter 1 ● Electrons and Holes in Semiconductors

● GaAs, III–V Compound Semiconductors and Their Dopants ●

1.2 ● Bond Model of Electrons and Holes 7

FIGURE 1–7 The GaAs crystal structure.

FIGURE 1–8 Bond model of GaAs.

8 Chapter 1 ● Electrons and Holes in Semiconductors

Empty upper bands

Lowest (nearly) empty band

Highest (nearly) filled band

Filled lower bands

1.3 ● ENERGY BAND MODEL ●

1.3.1 Energy Band Diagram

1.3 ● Energy Band Model 9

FIGURE 1–10 The energy band diagram of a semiconductor.

● Photoconductor as Light Detector ●

Semiconductor InSb Ge Si GaAs GaP ZnSe Diamond

Eg (eV) 0.18 0.67 1.12 1.42 2.25 2.7 6.0

10 Chapter 1 ● Electrons and Holes in Semiconductors

EXAMPLE 1–1 Measuring the Band-Gap Energy

1.3.2 Donors and Acceptors in the Band Model

1.4 ● Semiconductors, Insulators, and Conductors 11

TABLE 1–2 • Ionization energy of selected donors and acceptors in silicon.

Ionization energy, Ec – Ed or Ea – Ev (meV) 39 44 54 45 57 160

1.4 ● SEMICONDUCTORS, INSULATORS, AND CONDUCTORS ●

(a) Si, semiconductor (b) SiO2, insulator (c) Conductor

12 Chapter 1 ● Electrons and Holes in Semiconductors

1.5 ● ELECTRONS AND HOLES ●

1.5.1 Effective Mass

1.5 ● Electrons and Holes 13

Increasing electron energy

Increasing hole energy

Si Ge GaAs InAs AlAs

mn /m0 0.26 0.12 0.068 0.023 2.0

14 Chapter 1 ● Electrons and Holes in Semiconductors

Each semiconductor material has a unique E–k relationship (due to the

1.5.2 How to Measure the Effective Mass2

FIGURE 1–15 The motion of electrons in an N-type semiconductor in the presence of a

2 This section may be omitted in an accelerated course.

1.6 ● Density of States 15

electrons will strongly absorb the microwave energy. They do so by accelerating to a

1.6 ● DENSITY OF STATES ●

It is useful to think of an energy band as a collection of discrete energy states.

16 Chapter 1 ● Electrons and Holes in Semiconductors

1.7 ● THERMAL EQUILIBRIUM AND THE FERMI FUNCTION ●

1.7.1 An Analogy for Thermal Equilibrium

1.7 ● Thermal Equilibrium and the Fermi Function 17

1.7.2 Fermi Function—The Probability of an Energy State

Equation (1.7.2) is known as the Boltzmann approximation. In the low energy

f(E) ⬵ e(E EF)/kT

Nd 6 1016 cm3 Nd 6 1016 cm3