ECE102: Semiconductor

Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L1 0
 Course Overview
• Course handout, grading scheme, etc
• What students will learn from this course (First ask the students!)
• Broad outline of course contents:
• Fundamentals of semiconductor materials
• Energy bands, Electrons and holes as charge carriers
• Excess minority carriers
• Carrier transport mechanisms
• Semiconductor junctions
• MOS capacitors (Metal Oxide Semiconductor)
• Field Effect Transistors (FET)
• Bipolar Junction Transistor (BJT)
• Integrated Circuits (ICs)
• Power semiconductor devices
ECE102 - VH - L1 1
Semiconductor Devices
• Building blocks of today’s electronic gadgets
• Laptops
• Mobile phones
• TVs (smart or otherwise!)
• Automobiles (ECUs, ADAS, AVs, etc)
• USB pen drive
• LED (Light Emitting Diode)
• The list goes on …
• Form factor:
• Integrated circuits (IC): VLSI
• Discrete devices
ECE102 - VH - L1 2
 Broad Learnings from this Course
• What are the electrical characteristics of key
semiconductor devices
• What happens when some voltage is applied at some
terminals
• Why these devices exhibit these characteristics
• The exact physics of what happens inside them, that make
them do the things they do
• Sample applications of such devices
ECE102 - VH - L1 3
 Why Integrated Circuits
• More transistors on die => more functionality
• Lower power
• Good matching (especially important for analog circuits)
• Low cost
• For integrated circuits, before packaging stage is reached, the process cost is
amortized over many xtors in many dies in many wafers in many lots across many
customers (in case of foundry model). For example: Today’s digital designs have
about 50M xtors (ie. 50x106 transistors) per mm2
• For same circuit made using discrete components, every part is packaged before use
• Other benefits of integrated circuit versus making using discrete
components:
• Better reliability when an IC chip is surface mounted on a PCB
• Lesser weight when an IC chip is surface mounted on a PCB

ECE102 - VH - L1 4
Moore’s Law

Typical size-related statistics in 2020:

• GPU die size (typically bigger than
CPU): 600 𝑚𝑚2 (think of it as 2 cm x 3
cm die)
• Xtor count in such a die: 30 billion
(30 x 109 )
• => Xtor density: 50 million per 𝑚𝑚2
ECE102 - VH - L1 5
Integrated Circuits
1960: First planar IC (by Fairchild
Semiconductors Inc), had about 4 xtors
Source: https://www.computerhistory.org/revolution/digital-logic/12/329

What a wafer fab looks like

Source: https://fortune.com/2016/06/11/raytheon-next-gen-chips/

2007: Intel fab engineer holding a 300 mm

wafer with 45 nm test chips
Source: https://en.wikichip.org/wiki/45_nm_lithography_process

ECE102 - VH - L1 6
 Semiconductors: From Sand to Chip
• Key enablers:
• Process:
• Photolithography (immersion, dual/quad patterning, etc)
• Ion implantation and/or diffusion
• Etching (dry, wet, plasma, etc)
• Epitaxial growth
• ALD (atomic layer deposition)
• Metallization (evaporation, (dual)damascene, etc)
• CMP (chemical mechanical polishing)
• Tools
• TCAD tools (process simulation, device simulation, EM field-solvers for on-chip
interconnects)
• EDA/CAD tools: Circuit simulation, parasitic extraction, reliability analysis, etc. And
in the case of digital circuits: HDL simulation, synthesis, place and route, static
timing analysis, etc
• Each of the above are active areas of research
ECE102 - VH - L1 7
Most Impactful Device: MOSFET
• MOSFET: Metal Oxide Semiconductor Field Effect
Transistor (Xtor)

Will learn more about

them later in this course

Common Schematic Symbol (there are several other symbols in use, depending on use in analog
versus digital, enhancement versus depletion mode, etc)
ECE102 - VH - L1 8
Cross-section: Planar N-MOSFET

Schematic cross-section Schematic cross-section that’s a bit more realistic

ECE102 - VH - L1 9
 Typical Layout and Other Views
Chip-design activity final deliverables: Layout polygons
for each layer that help make the masks for that layer. In
picture below, each color is a different layer/mask.
Similar to multi-layer PCB design

ECE102 - VH - L1 10
END OF LECTURE

ECE102 - VH - L1 11
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L2 0
Agenda
• What is a semiconductor
• What is a crystal
• Material properties of Silicon

ECE102 - VH - L2 1
 What is a Semiconductor?
• Has electrical conductivity (𝜎,
pronounced as sigma) between
that of a conductor and insulator Table (a) shows a fragment of the periodic table

• 𝜎 can be significantly altered (by

doping with specific impurities,
heating, optical excitation, etc)
• Elemental semicons are in Group
IV of the Periodic Table, with
Silicon (Si) being the most popular
• Compound semicons (eg. GaAs)
also possible (used for LEDs, lasers,
high-speed devices, etc)
ECE102 - VH - L2 2
 Atomic Structure
• Regular arrangement in a specific orientation => Crystalline
• Random => Amorphous
• Regular in select regions after which orientation changes =>
Polycrystalline (the boundaries are called grain boundaries)
• Most semiconductor devices for electronics are made from all
three kinds, but most action happens in crystalline material!

Crystalline Amorphous Polycrystalline

ECE102 - VH - L2 3
•
Crystal Lattice
Lattice (a geometric concept)≜ Periodic arrangement
of <something>
• Basis≜ What arrangement of which atoms within
that <something>
• ՜ ՜
𝒂 and 𝒃 are called basis vectors

• Thus: Lattice + Basis = Crystal 2D lattice showing translation of a unit cell by r = 3a + 2b

• Unit Cell: Any group of atoms (not necessarily the

smallest) that can be repeated to form the crystal
structure
• Smallest => Primitive (e.g. ODEF is a rhombic primitive
cell)
• Else non-primitive (e.g. PQRST is a rectangular non-
primitive cell). Often more convenient to work with
• Mechanical properties (eg. density) as well as
electrical properties can be inferred Face-centered cubic (FCC) Packing of hard spheres in an FCC lattice
ECE102 - VH - L2 4
 Miller Indices: Planes
• 3 integers (ℎ𝑘𝑙) used to refer to a plane in the
lattice (notice the parenthesis). Method:
• Find inverse of intercepts along crystal axes wrt basis
vectors
• Scale them proportionately to integral values
• Thus, parallel planes have the same Miller indices
• Eg.: Consider the inclined plane. 2, 4 and 1 are the
intercepts. Their inverses are 1Τ2 , 1Τ4 , 1. When
scaled in same proportion, we get 2, 1, 4. Hence
referred to as a (214) plane.
• The set of crystallographically equivalent planes
(ie. indistinguishable from symmetry standpoint,
even when not parallel to each other) are
indicated in curly braces {} The 6 faces of the cubic lattice can be
referred to as {100} planes
ECE102 - VH - L2 5
 Miller Indices: Directions
• 3 integers [ℎ𝑘𝑙] used to refer to a direction in the
lattice (notice the square brackets). Method: Find
vector components along crystal axes wrt basis
vectors (similar to vector decomposition)
• Thus, parallel directions have the same Miller
indices
• Eg.: Consider a direction along the cube diagonal.
Its components are 1, 1 and 1. Hence referred to
as a [111] direction.
• The set of crystallographically equivalent
directions (ie. indistinguishable from symmetry
standpoint, even when not parallel to each other)
are indicated in angular braces <> The 3 directions of the cubic lattice along
a,b,c can be referred to as <100> directions
ECE102 - VH - L2 6
 Lattice structure of Si, etc Top view along any < 100 >
direction of an extended diamond
• Si and Ge have diamond lattice lattice, showing the two
interpenetrating sublattices

• Most compound semicons have

zincblende lattice
• Both have same lattice structure
but different basis
• Both lattices can be visualized as an FCC structure with an extra
atom placed at 1/4th the lattice constant (lattice constant: size ′𝑎′
of a cubic unit cell)
• In diamond lattice, both sublattices have the same atoms (e.g. Si)
• In zincblende lattice, they alternate (Ga in blue sites, As in black
sites)
ECE102 - VH - L2 7
 Silicon Valence Structure
• A silicon atom has 14 electrons. Electrons in the
outermost shell (n=3) are called the valence electrons,
they take part in bonding with other Si atoms
• Using Quantum Mechanics to solve the Schrodinger
equation results in the 2following electronic
configuration for Si: 1𝑠 2𝑠 2𝑝 3𝑠 2 3𝑝2
2 6

• The p-orbital can hold 6 electrons but holds only 2, and

is thus incompletely filled (the s- and p-orbitals3in n=3
actually hybridize into 4 tetrahedral-shaped 𝑠𝑝
orbitals when bonding with nearby Si atoms, and are
responsible for the diamond lattice)
• A silicon atom and its 4 nearest neighbors is shown in
red in the diamond cubic lattice, with−10
lattice constant =
5.431 Angstroms (1 Angstrom = 10 𝑚 = 0.1 𝑛𝑚)

ECE102 - VH - L2 8
 Crystal Growth
• Pure and single-crystal material is key to device performance. Defects
(e.g. missing or foreign atom at a lattice site) degrades device
performance
• Good device behaviour requires less than 1 defect in 109 atoms (1 part
per billion)!
• Technique:
• Purity: Using chemical reactions starting with sand (SiO2),
one can get EGS purity (Electronic Grade Silicon, ie.
9
less
than 1 ppb foreign elements (parts per billion (10 )), albeit
polycrystalline in structure
• Single-crystal: Czochralski method is then used where pure
and single-crystal Si grows around a seed material slowly
rotated and raised above a molten bath (1412 °C) of the
above mentioned EGS purity polycrystalline Si
ECE102 - VH - L2 9
 Wafers
• The grown crystal comes out about 1 meter long and
weighs over 100 kgs, called an ingot (picture on the right)
• Its tapered edges are sawed off so as to let the cylindrical
portion remain (1 foot diameter, ie. 300 mm)
• Crystal planes in the ingot are identified using X-ray
crystallography, and a small notch is ground on one side
of the cylinder to indicate the crystal-plane of the top face
of the ingot
• The ingot is then sawed into thin wafers less than 800 um
thick
• The wafers are highly polished (to avoid depth-of-focus
(DoF) problems in subsequent lithographic steps), and
edges rounded (to avoid chipping off during wafer
handling)
ECE102 - VH - L2 10
 Summary: Silicon Characteristics
• Abundantly available (> 70% of earth’s crust contains silicon in some form)
• Has an excellent oxide (SiO2) that is very compatible-with and useful-in IC
process technologies
• Very good (ie. high) thermal conductivity, hence is able to effectively carry
heat away from active areas in IC chips, which is desirable
• All these have been responsible for it being the candidate semiconductor of
choice for active devices in discrete devices and ICs
• Some material properties (more will be covered in a later lecture):
• Atomic density = 5x1022 atoms/cm3
• Intrinsic (ie. undoped) carrier concentration 𝑛𝑖 = 1.5x1010 electrons/cm3
• This is the #electrons available to take part in conduction. This is much lower than a conductor but
much higher than an insulator
• When doped, it is called extrinsic, and it has a much higher carrier concentration that 𝑛𝑖
• Bandgap of silicon 𝐸𝑔 = 1.11 𝑒𝑉 (will learn what this means in a later lecture)
ECE102 - VH - L2 11
Physical Constants

Source: Appendix from Streetman & Banerjee text

ECE102 - VH - L2 12
Properties of Some Semicons
Note:
1. All values at 300 K
2. The 2nd column indicates the band
structure type and crystal
structure, viz.:
• i = indirect band structure
• d = direct band structure
• D = diamond
• Z = zinc blende
• W = wurtzite
• H = halite (NaCl)
3. Values of mobility are for material
of high purity (undoped)
4. The gaps indicate that the values
are unknown
5. Vacuum permittivity:
𝜖0 = 8.8542x10−12 𝐹/𝑚

Source: Appendix from Streetman & Banerjee text

ECE102 - VH - L2 13
 Unit of Energy: Joules vs eV
• 1 eV = 1.6 × 10−19 Joules ≜ Energy gained-by an
electron in a potential difference of 1 Volt
• Note: 𝑘𝑇 is an energy (Joules or eV), but 𝑘𝑇/𝑞 is a
potential (Volts). Thus, a 𝑘𝑇/𝑞 value of 26 mV
(milli-Volts) corresponds to 𝑘𝑇 of 26 meV (milli-
electron-Volts)
• Why I chose these peculiar value of 26 mV, will
become clear in later lectures
ECE102 - VH - L2 14
END OF LECTURE

ECE102 - VH - L2 15
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L3 0
Agenda
• Energy bands in crystalline
semiconductors

ECE102 - VH - L3 1
 Implications of Quantum Mechanics
• Quantum Mechanics describes the laws of Physics governing
subatomic particles, and the Schrodinger Equation (SE) is a key
equation that encapsulates these laws
• These laws are key to explaining why the key devices in today’s gadgets
(mobile phones, laptops, etc) such as diodes, transistors, LEDs, USB
drives, lasers, solar cells, etc work the way they do! No kidding!
• With suitable boundary conditions, the final outcomes of this
equation are:
• A set of wavefunctions 𝜓 as a function of space (ie. 𝑥, 𝑦, 𝑧), which are
deemed as the eigenvectors of the SE. These are often complex functions
(ie have real & imaginary ∗parts) and by themselves don’t have any
physical meaning. But 𝜓 𝜓 represents the square of the probability of
finding an electron at that point in space (probability density)
• Eigenvalues pertaining to those wavefunctions, which correspond to
energy values
ECE102 - VH - L3 2
 Implications of Quantum Mechanics
• When solved for electrons of an isolated atom (e.g.
Hydrogen, Silicon, etc) in a potential well, the results
show that the electrons can only exist in certain
discrete, allowed energy levels
• When solved for electrons of very many atoms of a
crystalline solid (such as a realistic Si crystalline sample
may contain), the results show that the wavefunctions
of electrons of the individual atoms interact and the
allowed energies are no longer discrete levels, but
rather a range of energies; that is, a band of energies
ECE102 - VH - L3 3
Energy Levels → Bands as Atoms get Closer
2 electrons per
• N atoms brought atom in 3p2 level
closer at 0° 𝐾 (max capacity 3p6)
• Left edge of figure
corresponds to
atoms getting close
• Right end 2 electrons per
corresponds to atom in 3s2 level
isolated atoms (max capacity 3s2
=> fully filled)

ECE102 - VH - L3 4
 Energy Bands
• In a typical crystalline semiconductor, there are multiple
such bands (each containing so many finely spaced
discrete energy levels that they’re imperceptible, so a
continuous band is a better representation)
• These bands are separated by a bandgap (ie. forbidden
energy range), whose value is typically in 𝑒𝑉, ie. electron
volts

ECE102 - VH - L3 5
 Energy Bands
• For semiconductors at 0° K (Kelvin), it turns out that there are
bands of allowed energies that are fully filled with electrons, and
beyond that (ie. higher in electron energy) there are bands that
are fully empty. The highest such fully-filled band is called the
valence band and the lowest empty band is called the conduction
band
• Actually all the empty bands are conduction bands,
but the lowest empty band is most impactful in
device phenomena, and hence the term conduction
band usually refers to the lowest empty band
• The situation at the top of the valence band and the
bottom of the conduction band is key to explaining
several device phenomena
ECE102 - VH - L3 6
 Energy Bands
• For current to flow (ie. electron transport to occur), two conditions
need to be satisfied:
A. There must be electrons available in a band
B. Vacant (ie. unfilled), allowed energy states must exist in that band
• In a semiconductor at 0° K, both conditions are not simultaneously
satisfied. Thus, a semiconductor at 0° K acts like an insulator
• However, unlike an insulator, the bandgap 𝐸𝑔 of a semiconductor
isn’t very high. Therefore, at nearabout room temperature, a few
electrons in the valence band get sufficient thermal energy to get
excited to the conduction band, thereby satisfying conditions A
and B above, and hence having some conductivity (much better
than insulator but much worse than conductor)
ECE102 - VH - L3 7
 Band Structure vs Band Diagram
• Band structure (top figure) refers to the depiction of
allowed band of energies versus wave-vector 𝑘
(introduced when solving the Schrodinger equation), an
abstract quantity that is related to momentum
• The lattice structure and periodicity of most materials is
different in different directions, hence the full (𝐸, 𝑘) band
structure is a complex 3D surface
• Materials (eg. GaAs) that have the valence band maximum
and conduction band minimum at the same value of 𝑘, are
called direct semiconductors. Very useful for LEDs, etc
• Materials (eg. Si) that have the valence band maximum
and conduction band minimum at different 𝑘 values, are
called indirect semiconductors
• Band diagram (bottom figure at 0° K) refers to the
depiction of allowed band of energies versus spatial
location (ie. position)
ECE102 - VH - L3 8
Band Structure of Compound Semicons
• Band structures often depict symbols like 𝐿, Γ, 𝑋, etc. These correspond (in
𝑘-space) to high-symmetry points of the crystal structure, and conduction
band minima can occur at or near these 𝑘-values (for example, 𝐿
corresponds to 𝑘𝑥 = 𝑘𝑦 = 𝑘𝑧 = 0)
• The bandgap (𝐸𝑔 ) is the separation along the E-axis (in an E-k band structure)
between the top of the valence band and the lowest point of the conduction
band
• Materials like 𝐴𝑙𝑥 𝐺𝑎1−𝑥 𝐴𝑠, 𝐺𝑎𝐴𝑠1−𝑥 𝑃𝑥 where 𝑥 is the mole fraction, etc are
called compound semiconductors. As the mole fraction 𝑥 varies, the
bandgap (𝐸𝑔 ) also varies, as also the characteristic of the material from being
a direct to an indirect semiconductor
• 𝐺𝑎𝐴𝑠1−𝑥 𝑃𝑥 remains a direct semiconductor from 𝑥 = 0 to about 𝑥 = 0.45,
and is therefore used in visible LEDs with mole-fractions in this range
• Most red LEDs are made with 𝐺𝑎𝐴𝑠1−𝑥 𝑃𝑥 with 𝑥 = 0.4, as its 𝐸𝑔 corresponds to the
wavelength of red light (recall, 𝐸 = ℎ𝜈 = ℎ𝑐/𝝀)
ECE102 - VH - L3 9
END OF LECTURE

ECE102 - VH - L3 10
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L4 0
 Agenda
• Electrons and holes as charge carriers in
semiconductors
• Band structure
• Band-structure-effective-mass (BSEM) of
carriers (also called inertial effective
mass)
• Doping and extrinsic semiconductors
ECE102 - VH - L4 1
 Electric Current: Motion of Charge
• Electric current is simply the rate of flow of charge, when
summed amongst all the contributing charge carriers in all
the various participating energy bands (in the energy band
concept of a crystalline semicon)
• The primary charge carrier in a material is an electron
(symbol 𝑛)
• For charge to flow, 2 conditions need to be met:
• Charges must exist in a band to participate in motion
• Vacant energy states must exist in a band for the
charges to occupy those states
ECE102 - VH - L4 2
Electrons and Holes
• In a pure (ie. intrinsic/undoped) semicon at 0 °K,
the valance band is fully filled with electrons and
the conduction band is totally empty
• Thus, neither of the above 2 conditions are met
• But at room temperature, some of the electrons
are thermally excited from the valence band to
the conduction band, leaving behind some empty
states in the valence band
ECE102 - VH - L4 3
 Electrons and Holes
• The vacant energy states in the
valence band are called holes (symbol
𝑝)
• They can be thought of as contributing
to current as the participating agents
in the valence band
• The real physical carrier is still the
electron, albeit with different dynamics
(viz. effective mass), primarily due to the
different band structures (E-k) of the
conduction and valence bands
ECE102 - VH - L4 4
 Electrons and Holes
• Effectively, you can deem:
• Electrons as being the participating charge carriers in the conduction
band
• Holes as being the participating charge carriers in the valence band
• This does not tantamount to double counting, because the
underlying physical considerations does indeed factor in vacant
energy states being available in both bands
• For purposes of current calculation, holes are deemed as
positive charges (same magnitude as electron charge), albeit
having a different effective mass from that of an electron

ECE102 - VH - L4 5
 Electrons and Holes – Intrinsic Semicon
• Rough physical visualization for Si
(silicon) shown here (though it is
inaccurate from a Quantum Mechanics
standpoint and for purposes of
calculating useful quantities)
• If one of the covalent bonds is broken
such that one of the electrons in the
bond becomes free to move about in
the lattice, a conduction electron is
created and a broken bond (hole) is
left behind
• The energy required to break the bond
is the band gap energy 𝐸𝑔
ECE102 - VH - L4 6
 BS-Effective-Mass of Carriers
• As stated earlier, band diagrams show
spatial variation of the band edges
(bottom of conduction band 𝐸𝑐 and
top of valence band 𝐸𝑣 )
• Figure on the right shows the band
structure superimposed on a band
diagram
• The x-axis is spatial location (say 𝑥)
• The y-axis is total energy 𝐸 (= 𝑃𝐸 + 𝐾𝐸)
• 𝐸𝑣 and 𝐸𝑐 levels denote PE (Potential
Energy)
• Carriers are distributed somewhere in
such an energy space
ECE102 - VH - L4 7
 BS-Effective-Mass of Carriers
• The slope of 𝐸𝑐 (or 𝐸𝑣 , since they are parallel in a homogeneous
material) wrt space denotes the potential gradient, which is a
measure of the electric field in the material
• The extent above 𝐸𝑐 is the KE (Kinetic Energy) of the electron
• The extent below 𝐸𝑣 is the KE of the hole
• In such diagrams, electron energies increase in the up direction
whereas hole energies increase in the down direction
• Whenever possible, carriers always tend to move to a point of
lowest allowed energy (stable state)
ECE102 - VH - L4 8
 BS-Effective-Mass of Carriers
• Thus, electrons higher up in the conduction band tend to
settle down to the 𝐸𝑐 level whereas holes deep down in the
valence band tend to float up like bubbles to the 𝐸𝑣 level
• In doing so, they give up their KE (as heat to the lattice, for
example) to reach those PE levels
• Similarly, in a potential gradient (ie. electric field 𝐸𝐹𝑖𝑒𝑙𝑑),
electrons at the bottom of 𝐸𝑐 will tend to roll downhill
along the downslope of 𝐸𝑐 whereas holes at the top of 𝐸𝑣
will tend to float uphill along the upslope of 𝐸𝑣
• This is the same as saying that electrons get attracted to the
positive electrode of a pair of electrodes that are creating the
𝐸𝐹𝑖𝑒𝑙𝑑
ECE102 - VH - L4 9
 BS-Effective-Mass of Carriers
• Consider the superimposed band diagram in slide 7 with an electron moving through
the sample from A to B under the influence an 𝐸𝐹𝑖𝑒𝑙𝑑
• The 𝐸𝐹𝑖𝑒𝑙𝑑 is given by the slope of either band edge (PE)
• As the electron moves from A to B in the band diagram (horizontally in the energy
space in the figure), it gains KE (the differential above band edge) while losing PE (𝐸𝑐
downslope)
• Correspondingly, in the (E, k) band structure, the electron starts at 𝑘 = 0, but moves
to a nonzero wavevector 𝑘𝐵 . The electron then loses KE as heat to the lattice by
scattering mechanisms and returns to the bottom of the band at B
• In reality, the electron may lose its KE in stages by a series of scattering events, as
shown by the colored dashed lines
• Note: If the 𝐸𝐹𝑖𝑒𝑙𝑑 between A and B were not constant, the slope of the band edges
would not be uniform but would vary at each point reflecting the magnitude and
direction of the local 𝐸𝐹𝑖𝑒𝑙𝑑
ECE102 - VH - L4 10
 BS-Effective-Mass of Carriers
• For isotropic materials (where band structures are simple
and identical in all k-directions), the band-structure-
effective-mass (BSEM) is defined as
2
ℏ
𝑚∗ = 2
𝑑 𝐸/𝑑𝑘 2
ℎ
where ℏ = and ℎ is the Planck’s constant
2𝜋
• It is defined thus so as to be able to semi-classically
approach carrier dynamics, ie. apply Newton’s Laws to the
𝐸 − 𝑘 description (that in turn has been derived quantum
mechanically) of a material
ECE102 - VH - L4 11
 BS-Effective-Mass of Carriers
• The 2nd derivative is related to the curvature of E-k
diagram (band structure) => BSEM is inversely
proportional to the curvature
• Figure shows the band structure of GaAs. Three 𝐸𝑐
minimas are seen, at 𝑘 values corresponding to 3
different high symmetry points
• The one at 𝑘 = 0 is called the Γ point (pronounced gamma)
• The one along [100] direction is called the 𝑋 point
• The one along [111] direction is called the 𝐿 point
• The 𝐸𝑐 edge at the Γ point (𝑘 = 0) has the sharpest (ie.
highest) curvature => Electrons in this particular
conduction band have the least effective mass,
compared to the 𝐿 and 𝑋 conduction band minimas

ECE102 - VH - L4 12
 Extrinsic Semiconductors
• Intrinsic semicons have very few carriers at room temperature
• E.g.: Si has about 1.5𝑒10 𝑐𝑚−3 electrons/holes (ie. 1.5x1010 𝑐𝑚−3 )
• Intrinsic semiconductors will always have an equal number of electrons and holes at
equilibrium, since they are thermally generated in a pure sample
• These can be doped by specific impurities to make it n-rich or p-
rich, in which case they are called extrinsic semiconductors
• Example:
• Silicon (Si) is a group-4 element (in the periodic table)
• Doping Si with a group 5 element like Arsenic (As), Phosphorus (P), etc
creates fully occupied energy levels in the band gap close to 𝐸𝑐 , barely
50 𝑚𝑒𝑉 (milli-electron Volt) away from 𝐸𝑐 . These are called donor levels 𝐸𝑑
• Doping Si with a group 3 element like Boron (B), Indium (In), etc creates fully
empty energy levels in the band gap close to 𝐸𝑣 , barely 50 𝑚𝑒𝑉 (milli-
electron Volt) away from 𝐸𝑣 . These are called acceptor levels 𝐸𝑎
ECE102 - VH - L4 13
 Extrinsic Semiconductors
• Exact mechanics of doping:
• N-type doping: Electrons at donor level
𝐸𝑑 are easily excited at room
temperature to almost fully jump to 𝐸𝑐 ,
thus making the conduction band
electron rich (n-rich)
• P-type doping: Electrons in 𝐸𝑣 are easily
excited at room temperature to almost
fully occupy all the vacant states in the
acceptor level 𝐸𝑎 , thus making the
valence band hole rich (p-rich)
ECE102 - VH - L4 14
 Extrinsic Semiconductors
• Exact mechanics of doping:
• Structurally, such doped impurities occupy lattice sites in the crystal
• An As atom (Gr V) in the Si lattice has four necessary valence electrons to complete the
covalent bonds with the neighboring Si atoms, plus one extra electron. This fifth electron
does not fit into the bonding structure of the lattice and is therefore loosely bound to
the As atom. Little thermal energy enables this extra electron to overcome its coulombic
binding to the impurity atom and be donated to the lattice as a whole. Thus it is free to
participate in current conduction. This process is a qualitative model of the excitation of
electrons out of a donor level and into the conduction band
• Similarly, the Gr III impurity B has only three
valence electrons to contribute to the covalent
bonding, thereby leaving one bond incomplete.
With a little thermal energy, this incomplete
bond can be transferred to other atoms as the
bonding electrons exchange positions. The
resulting electron "hopping" from an adjacent
bond into the incomplete bond at the B site is
akin to hole flow in the valence band
ECE102 - VH - L4 15
END OF LECTURE

ECE102 - VH - L4 16
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L5 0
 Agenda
• Carrier statistics
• Carrier concentrations at equilibrium

Note: All energies under discussion are electron energies (and are therefore negative) unless otherwise stated, and increase as
you go up in the energy axis in plots/graphs (meaning they remain a negative number albeit become smaller in magnitude)

ECE102 - VH - L5 1
•
Carrier Statistics
Given the large number of allowed energy states (≈ 10 19
per cm3) and the
large number of charge carriers (≈ 1010 per cm3) in any given sample size,
the occupancy of these energy states with electrons is a statistical
phenomenon, bringing in concepts like finding the most likely occupancy
pattern (MLOP) where:
• There is a band of closely-spaced energy levels each of which has some # of
available states (that depends on that energy level and is derived from Quantum
Mechanics, irrelevant for calculation of the MLOP),
• Each occupiable state can atmost hold 1 electron (Pauli Exclusion Principle),
• The total number of electrons (that need to be distributed amongst these
energies) is fixed, and
• The total energy of the electrons over all energies is specified (ie. fixed)
• Most likely occupancy pattern: What we mean by this is that if we repeat
the random distribution experiment (like billiard balls in a multi-storied
building) millions of times, then it turns out that one particular
distribution occurs the most often. That distribution is the MLOP
ECE102 - VH - L5 2
 Fermi-Dirac Statistics
• It is possible to show (beyond the scope of a UG course) that
under these considerations, the most likely occupancy pattern at
an energy 𝐸 is given by the Fermi-Dirac Statistics given by:
1
𝑓 𝐸 =
1 + 𝑒 𝐸−𝐸𝐹 Τ𝑘𝑇
where:
• 𝐸𝐹 ≜ Fermi energy level
• 𝑘 = Boltzmann constant = 1.38x10−23 𝐽/𝐾 (Joules per Kelvin)
• 𝑇 = Absolute temperature (= 𝑇𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒𝐼𝑛𝐶𝑒𝑙𝑐𝑖𝑢𝑠 + 273)
• Note: 𝑘𝑇 has units of energy (Joules, or often also stated in electron-volts 𝑒𝑉)
• 𝑓(𝐸) can be interpreted as the probability of an electron
occupying an energy level 𝑬 (ie. a probability distribution,
without regard to whether 𝐸 is an allowed energy level or not;
that is factored in later as a superseding condition)
ECE102 - VH - L5 3
 Fermi-Dirac (FD) Statistics
• Figure shows the 𝑓(𝐸) distribution at
various temperatures is shown.
Notable features:
• 𝐸𝐹 signifies the energy level at which
the occupation probability is 0.5 (other
than at 𝑇 = 0, read on)
• At 𝑇 = 0, 𝐸𝐹 signifies the energy level
below which all levels are occupied and
above which all are empty (Think why!)
• 1 − 𝑓 𝐸 denotes the probability than
an energy level 𝐸 is vacant, ie.
unoccupied (think of holes)
• 𝑓 𝐸𝐹 + Δ𝐸 = 1 − 𝑓 𝐸𝐹 − Δ𝐸
• That is: Probability of occupancy and
vacancy above and below 𝐸𝐹 respectively,
is symmetric, at all temperatures
• The tail of the distribution spreads out
at higher temperatures
ECE102 - VH - L5 4
Fermi-Dirac Stats & Allowed Energies
• 𝑓(𝐸) can be rotated 90° left and superimposed on a band diagram that shows the
band-gap (ie. disallowed energies), to make more sense of the occupancy statistics
• Intrinsic semicon => #electrons = #holes => 𝐸𝐹 (= 𝐸𝑖 ) is very close to middle of bandgap
• N-type doped semicon => #electrons ≫ #holes => 𝐸𝐹 is closer to 𝐸𝑐
• P-type doped semicon => #holes ≫ #electrons => 𝐸𝐹 is closer to 𝐸𝑣

ECE102 - VH - L5 5
 Density of States (DoS)
• From fundamental Quantum Mechanics as applied to a crystalline
semicon, it is possible to calculate (beyond the scope of a UG
course) the Density of States (DoS) 𝑁 𝐸 𝑑𝐸 which is the number
of allowed energy states per unit volume within an elemental
energy 𝒅𝑬 around an energy level 𝑬
• Then, the way to factor in this information (of allowed energy
states) onto the FD statistics when calculating the total number of
carriers (say electrons) at equilibrium, is by multiplying the two
and integrating over all allowed energies. For example:
∞
𝑛0 = න 𝑓 𝐸 𝑁 𝐸 𝑑𝐸
𝐸𝑐
• Because of the large DoS in a band, small changes in the profile of
𝑓(𝐸) can result in huge changes in 𝑛0
ECE102 - VH - L5 6
 Effective DoS 𝑁𝑐 , 𝑁𝑣
• This integral can be solved approximately to yield:
𝑛0 ≅ 𝑁𝑐 𝑓 𝐸𝑐
(𝑁𝑜𝑡𝑖𝑐𝑒, 𝒘𝒊𝒕𝒉𝒐𝒖𝒕 𝑡ℎ𝑒 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑙!)
• Equivalent to saying that all the allowed energy states exist
only at conduction band edge 𝐸𝑐 with 19
a DoS of 𝑁𝑐 . 𝑵𝒄 ≜
Effective DoS at 𝑬𝒄 (𝑁𝑐 ≅ 3.23x10 cm-3 for Si at 300𝐾)
𝐸𝑣
• Similarly, 𝑝0 = ‫׬‬−∞ 1 − 𝑓 𝐸 𝑁 𝐸 𝑑𝐸 ≅ 𝑁𝑣 1 − 𝑓 𝐸𝑣
• Equivalent to saying that all the vacant energy states exist
only at valence band edge 𝐸𝑣 with 19 a DoS of 𝑁𝑣 . 𝑵𝒗 ≜
Effective DoS at 𝑬𝒗 (𝑁𝑣 ≅ 1.83x10 cm-3 for Si at 300𝐾)
ECE102 - VH - L5 7
 Effective DoS 𝑁𝑐 , 𝑁𝑣
• It can be shown (beyond the scope of a UG course) that 𝑁𝑐 and
𝑁𝑣 are given by:
∗ 3/2 ∗ 3/2
2𝜋𝑚𝑛 𝑘𝑇 2𝜋𝑚𝑝 𝑘𝑇
𝑁𝑐 = 2 2
𝑎𝑛𝑑 𝑁𝑣 = 2
ℎ ℎ2
where 𝑚𝑛∗ and 𝑚𝑝∗ are called the density-of-states-effective-masses
(DoSEM) of electrons and holes respectively, not to be confused
with BSEM (though they are related, viz.: DoSEM 𝛼 Geometric Mean
(GM) of the BSEMs along various directions: transverse, longitudinal,
etc)
• The concept of DoS-effective-mass comes about by virtue of a
non-spherical band-structure 𝐸 − 𝑘 for a general semicon
ECE102 - VH - L5 8
Equilibrium Carrier Concentrations
• Using 𝑁𝑐 and 𝑁𝑣 and using the approximation that FD
stats can be approximated (as Maxwell-Boltzmann stats) when
𝐸 − 𝐸𝐹 > Few 𝑘𝑇, 𝑛0 and 𝑝0 can be derived as:
𝒏𝟎 = 𝑵𝒄 𝒆− 𝑬𝒄 −𝑬𝑭 /𝒌𝑻
𝒑𝟎 = 𝑵𝒗 𝒆− 𝑬𝑭 −𝑬𝒗 /𝒌𝑻
• 𝑛0 is the #electrons per unit volume at equilibrium
• 𝑝0 is the #holes per unit volume at equilibrium
• These formulae apply to extrinsic (ie. doped) as well
as intrinsic semicons, at any temperature T
ECE102 - VH - L5 9
 Equilibrium Carrier Concentrations
• Multiplying the two, we see that:
𝑛0 𝑝0 = 𝑁𝑐 𝑁𝑣 𝑒 − 𝐸𝑐 −𝐸𝑣 /𝑘𝑇 = 𝑁𝑐 𝑁𝑣 𝑒 −𝐸𝑔 /𝑘𝑇
• Thus, the product of electron and hole equilibrium
concentrations is independent of 𝐸𝐹 , which means it is
independent of the doping level! (Recall, a non-zero
doping implies an 𝐸𝐹 that is shifted away from its intrinsic
value 𝐸𝑖 , which is approximately mid-gap)
• Why isn’t 𝐸𝑖 exactly at mid-gap? This is because 𝑁𝑐 ≠ 𝑁𝑣 , and
this in turn is bcoz the DoS effective masses of electron and
hole are different, and this in turn is because the 𝐸 − 𝑘 band
structure shape of 𝐸𝑐 and 𝐸𝑣 (curvature, # of equi-energy
surfaces, etc) are different near 𝐸𝑐 bottom and 𝐸𝑣 top
ECE102 - VH - L5 10
 Equilibrium Carrier Concentrations
• Hence: 𝑛0 𝑝0 = 𝑛𝑖 𝑝𝑖 where 𝑛𝑖 and 𝑝𝑖 refer to the
electron and hole concentrations in an intrinsic semicon
• But in an intrinsic semicon, 𝑛𝑖 = 𝑝𝑖
∴ 𝒏𝟎 𝒑𝟎 = 𝒏𝟐𝒊 (often called Mass-Action Law)
𝑛𝑖 = 𝑛0 𝑝0
• Using the results from the previous slide, this implies:
𝒏𝒊 = 𝑵𝒄 𝑵𝒗 𝒆−𝑬𝒈 /𝟐𝒌𝑻
ECE102 - VH - L5 11
 Equilibrium Carrier Concentrations
• Reiterating the results from slide 9, we have:
𝑛0 = 𝑁𝑐 𝑒 − 𝐸𝑐−𝐸𝐹 /𝑘𝑇
𝑝0 = 𝑁𝑣 𝑒 − 𝐸𝐹−𝐸𝑣 /𝑘𝑇
• In this, noting that 𝐸𝐹 = 𝐸𝑖 in an intrinsic semicon, we have:
𝑛𝑖 = 𝑁𝑐 𝑒 − 𝐸𝑐−𝐸𝑖 /𝑘𝑇
𝑝𝑖 (= 𝑛𝑖 ) = 𝑁𝑣 𝑒 − 𝐸𝑖 −𝐸𝑣 /𝑘𝑇
• Eliminating 𝑁𝑐 and 𝑁𝑣 amongst the 2 sets of above
equations, we get a more usable form of these equations:
𝒏𝟎 = 𝒏𝒊 𝒆 𝑬𝑭 −𝑬𝒊 /𝒌𝑻
𝒑𝟎 = 𝒏𝒊 𝒆 𝑬𝒊 −𝑬𝑭 /𝒌𝑻
ECE102 - VH - L5 12
•
Majority Carrier Concentration
As the name implies, majority carriers are those that are dominant. Thus:
• In n-type semicon (ie. doped to be n-type), electrons are the majority carriers
and holes are the minority carriers
• In p-type semicon (ie. doped to be p-type), holes are the majority carriers and
electrons are the minority carriers
• Easy way to calculate the majority carrier concentrations in doped
semicon at room temperature. Recall from lecture L4 that:
• n-type doping involves creating donor energy levels very close to 𝐸𝑐 that
overwhelmingly donate almost all its electrons to 𝐸𝑐
• p-type doping involves creating acceptor energy levels very close to 𝐸𝑣 that
overwhelmingly fill/absorb/accept almost all its vacant states with electrons
from 𝐸𝑣 (thereby creating holes in 𝐸𝑣 )
• Above two observations imply that:
• In n-type semicon obtained by doping 𝑵𝒅 impurity concentration where 𝑵𝒅 ≫ 𝒏𝒊
(which is generally the case), the equilibrium electron concentration 𝒏𝟎 ≅ 𝑵𝒅
• In p-type semicon obtained by doping 𝑵𝒂 impurity concentration where 𝑵𝒂 ≫ 𝒏𝒊
(which is generally the case), the equilibrium hole concentration 𝒑𝟎 ≅ 𝑵𝒂
ECE102 - VH - L5 13
 Example
A Si sample is doped with 1017 Arsenic 𝑎𝑡𝑜𝑚𝑠/𝑐𝑚3 . What is the equilibrium hole
concentration 𝑝0 at 300°𝐾? Show position of 𝐸𝐹 wrt 𝐸𝑖 in a band-diagram. Solution:
• As is an n-type dopant, thus 𝑁𝑑 = 1017 𝑐𝑚−3
• Clearly 𝑁𝑑 ≫ 𝑛𝑖 (since we know that 𝑛𝑖 ≅
1.5x1010 at 300°𝐾)
• Hence, the Si sample is doped n-type and 𝒏𝟎 ≈
𝟏𝟎𝟏𝟕 𝒄𝒎−𝟑
• 𝑛0 𝑝0 = 𝑛𝑖2 at equilibrium ⇒ 𝑝0 = 𝑛𝑖2 Τ𝑛0
• Hence 𝑝0 = 2.25𝑒20Τ1𝑒17 ⇒ 𝒑𝟎 = 𝟐. 𝟐𝟓x𝟏𝟎𝟑 𝒄𝒎−𝟑
• Notice the drastic difference in orders of magnitude of 𝑛0 and 𝑝0 !
• 𝑘𝑇 works out to 26 𝑚𝑒𝑉 (milli-electronvolts) at 300°𝐾 ) (remember this!)
𝑛 1𝑒17
• From slide#12, we see that 𝑬𝑭 − 𝑬𝒊 = 𝑘𝑇 ∙ 𝑙𝑛 0 = 26 ∙ 𝑙𝑛 = 26 ∙
𝑛𝑖 1.5𝑒10
𝑙𝑛 6.67𝑒6 = 407 𝑚𝑒𝑉 = 𝟎. 𝟒𝟎𝟕 𝒆𝑽. This is shown in band-diagram in top-right
ECE102 - VH - L5 14
Temperature Dependence of 𝑛𝑖
• By virtue of the relation 𝒏𝒊 =
𝑵𝒄 𝑵𝒗 𝒆−𝑬𝒈/𝟐𝒌𝑻 , 𝑛𝑖 varies
strongly with temperature, most
dominantly due to the
exponential term
• Figure shows the variation
ECE102 - VH - L5 15
 Miscellaneous Points
• Compensated doping:
• If a semicon is doped by 𝑁𝑑 and 𝑁𝑎 (sequence immaterial) with
𝑁𝑑 > 𝑁𝑎 , then the net doping is n-type with concentration 𝑁𝑑 −
𝑁𝑎 (assuming 100% ionization of dopants, a good approximation)
• If a semicon is doped by 𝑁𝑑 and 𝑁𝑎 (sequence immaterial) with
𝑁𝑑 < 𝑁𝑎 , then the net doping is p-type with concentration 𝑁𝑎 −
𝑁𝑑 (assuming 100% ionization of dopants, a good approximation)
• Fermi level 𝐸𝐹 is spatially constant at equilibrium (ie. in
regions where there is no flow of either carrier type). In 1D,
this is like saying 𝑑𝐸𝐹 Τ𝑑𝑥 = 0. Or in other words, there is
neither any discontinuity nor any slope/gradient in 𝐸𝐹 at
equilibrium
• Nicely explained in §3.5 in Streetman text, refer to it
ECE102 - VH - L5 16
END OF LECTURE

ECE102 - VH - L5 17
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L6 0
 Agenda
• Excess carriers
• Recombination-Generation relations

ECE102 - VH - L6 1
Excess Carriers
• At thermal equilibrium at some temperature, there
are a certain number of carriers in a semicon, viz.:
• 𝑛𝑖 electrons (and holes) if intrinsic
• 𝑛0 electrons (and 𝑝0 holes) if extrinsic
• There are ways in which carriers more than these (ie.
excess carriers) can be created. These are often
useful for specific device behavior and dominate the
conduction processes in the semiconductor
• Excess carriers can be created by:
• Optical excitation (focus of this lecture)
• Electron bombardment
• Injected due to a current flow
ECE102 - VH - L6 2
•
Optical Excitation
Consider a photon of energy 𝐸
incident on a semicon, where 𝐸 =
ℎ𝜈 and 𝜈 is its frequency (visible or
invisible, depends on 𝜈)
• If 𝐸 > 𝐸𝑔 (𝐸𝑔 being the bandgap),
then the photon is absorbed by the
semicon, and in the process its
energy is transferred to the
semicon by generating a EHP
(electron-hole pair):
a) A valence band electron gains energy E from the photon and jumps up to (excited) the
conduction band (gaining KE+PE)
b) It loses its KE as heat to the lattice
c) It recombines with the hole of′ the EHP losing PE in the process, and
′
emits a photon (ie.
luminescence)
′
of frequency 𝜈 (visible or invisible, depends on 𝜈 ; which in turn is given
by ℎ𝜈 = 𝐸𝑔 )
ECE102 - VH - L6 3
 Optical Excitation
• Depending on the semicon sample’s size, the absolute number of
electrons in the valence band (VB) is finite.
• Similarly, depending on the intensity of light 𝐼 being shined, the
#photons impinging on the semicon sample per unit area per unit time is
finite
• If intensity of light 𝐼 is too high, even if ℎ𝜈 > 𝐸𝑔 , all photons may not be
absorbed by the sample if sample size is too small (ie. if absolute# of
electrons in VB is small)
• If such incomplete absorption happens, then remaining photons will
come out from other end of sample with a reduced intensity which
degrades exponentially, viz.: For a 1D sample of length 𝑙, we have:
𝐼𝑡 = 𝐼0 𝑒 −𝛼𝑙 where:
• 𝐼𝑡 is the transmitted intensity (ie. coming out at the other end of the sample), and
• 𝛼 is the absorption coefficient and depends on the semicon material and frequency 𝜈
ECE102 - VH - L6 4
𝐸𝑔 of Some Semicons Wrt Spectrum
Note:
• Visible light spectrum
consists of VIBGYOR colors.
The approximate
wavelengths at either end
of this spectrum is:
• Violet: 400 𝑛𝑚 = 0.4 𝜇𝑚
• Red: 780 𝑛𝑚 = 0.78 𝜇𝑚

ECE102 - VH - L6 5
 Example
• A 0.46 𝜇𝑚-thick sample of GaAs (𝐸𝑔 = 1.43 𝑒𝑉)
is illuminated with monochromatic light of ℎ𝜈 =
2𝑒𝑉. The absorption coefficient 𝛼 =
5x104 𝑐𝑚−1 . The power incident on the sample
is 10 𝑚𝑊. Find:
a) Total energy absorbed by the sample per second (𝐽/𝑠)
b) Rate of excess thermal energy given up by the electrons to
the lattice before recombination (𝐽/𝑠, ie. Watts)
c) Number of photons per second given off from recombination
events, assuming perfect quantum efficiency (QE)
ECE102 - VH - L6 6
 Solution – Part (a)
𝐼0 = 10 𝑚𝑊
𝑙 = 0.46 𝜇𝑚 = 0.46x10−4 𝑐𝑚
𝛼 = 5x104 𝑐𝑚−1
∴ 𝐼𝑡 = 𝐼0 𝑒 −𝛼𝑙
= 10 ∙ 𝑒 −5𝑒4 ∙ 0 .46𝑒−4 𝑚𝑊
= 10 ∙ 𝑒 −5∙ 0 .46 = 10 ∙ 𝑒 −2.3
⇒ 𝑰𝒕 = 𝟏 𝒎𝑾
• Incident light 10 𝑚𝑊,
transmitted light 1 𝑚𝑊
• Thus 9 𝑚𝑊 was absorbed by
the sample
ECE102 - VH - L6 7
 Solution – Part (b)
• The fraction of each photon energy which is converted
to heat is:
2 − 1.43
= 0.285
2
∴ 𝑃𝑜𝑤𝑒𝑟𝐿𝑜𝑠𝑡𝑇𝑜𝐿𝑎𝑡𝑡𝑖𝑐𝑒𝐴𝑠𝐻𝑒𝑎𝑡 = 0.285x9 = 2.57 𝑚𝑊

ECE102 - VH - L6 8
 Solution – Part (c)
• The #photons per second given off from recombination is from the
absorbed power (not transmitted power)
• Absorbed power was found in part (a) to be 9 𝑚𝑊
• For every 1 photon absorbed, emission of exactly 1 photon occurs (full
QE). In emission of this photon, whether the absorbed energy was part
lost as heat to lattice before EHP-recombination, or none is lost as heat
to lattice, in either case, recombination with exactly 1 hole happens
• This emission (luminescence) happens from the available 9 − 2.57 =
6.43 𝑚𝑊 with each photon having energy 𝐸𝑔 = 1.43 𝑒𝑉 = 1.43 ∙ 1.6 ×
−19 −19 6.43×10−3
10 𝐽 = 2.288 × 10 𝐽. Hence #photons per second = −19 =
2.288×10
𝟐. 𝟖𝟏 × 𝟏𝟎𝟏𝟔 photons per second
ECE102 - VH - L6 9
Direct & Indirect Recombination
• Recombination of excited EHPs can
be direct or indirect
• Direct: Recombine directly across 𝐸𝑔
• Indirect: Recombine by means of an
allowed energy state 𝐸𝑡 within 𝐸𝑔
(mid-gap trap), ie. trap-assisted
• “Direct” and “indirect”
recombination is not to be
confused with direct and indirect
bandgap materials. Here in the
context of recombination,
“indirect” means trap-assisted Indirect recombination

ECE102 - VH - L6 10
 Dopants for Creating Traps
• Energy levels of dopants in Si that
create mid-gap trap levels is shown
in the alongside figure
• The energies shown are in 𝑒𝑉 from
the nearest band edge (CB or VB)
• Donor levels (ie. become +ve when
ionized) are shown with a + sign
• Acceptor levels (ie. become -ve
when ionized) are shown with a −
sign
• Some impurities introduce multiple
levels in 𝐸𝑔 such as:
• Zn introduces a level (𝑍𝑛− ) located
0.31 eV above the VB and a second
level (𝑍𝑛= ) near the middle of the
gap. Each 𝑍𝑛 impurity atom can
accept two electrons from the
semicon, one in the lower level and
another in the upper level
ECE102 - VH - L6 11
Recombination-Generation Relations
• Generation (partly thermal, and partly by external means such as
optical excitation, etc) creates EHPs and carriers in bands and
thus increases their counts (e.g. electrons in conduction band
(CB) and holes in VB)
• Recombination decreases the counts of such carriers in their
respective bands
• Recombination and generation (RG) happen concurrently
• The net rate of change-of-carrier-concentration-due-to-RG in a
band = 𝐺𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 – 𝑅𝑒𝑐𝑜𝑚𝑏𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒. In case of
electrons
𝜕𝑛𝑅𝐺
in the conduction band, this net rate can be expressed
as ൗ𝜕𝑡
• Specifically suffixing with RG because as mentioned before, carrier
concentrations can change due to other phenomena too. In this lecture,
we are specifically talking about change due to RG
ECE102 - VH - L6 12
 R-G Relations
• It can be shown that in steady-state, the net rate of change of carrier concentration in a
band (eg. electrons in the CB), due to R-G ie. 𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑛 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 𝑔𝑛 – 𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑛 𝑟𝑒𝑐𝑜𝑚𝑏𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 𝑟𝑛 is:
𝑑𝑛𝑅𝐺 (𝑡)
= 𝛼𝑟 𝑛𝑖2 − 𝛼𝑟 𝑛 𝑡 𝑝(𝑡)
𝑑𝑡
where 𝛼𝑟 is a constant of proportionality (not to confuse with absorption coefficient 𝛼 in
optical excitation)
• This is a simplified statement of the Shockley-Read-Hall model (SRH). It is derived based
on capture and emission modeling from assumed single-level traps/RG-centers, and
following
𝑑𝑛𝑅𝐺
through with “steady-state analysis” where 𝑑𝑝𝑅𝐺
net rate of change of electrons
( ൗ𝑑𝑡) is equal to net rate of change of holes ( ൗ𝑑𝑡) so as to imply equal net
creation of electrons and holes
• This is as opposed to “equilibrium analysis” where a principle of detailed balance applies and net
electron generation rate = net hole generation rate = 0). Thus, equilibrium is a more stringent
condition than steady-state
• Even though that derivation was done for trap-assisted recombination (ie. indirect), the
expressions are similar except for the fact that the expressions for carrier lifetime are
more complex in the case of trap-assisted recombination, read on
ECE102 - VH - L6 13
 R-G Relations
• Subtle point: Generally steady-state is deemed a𝑑𝑛
state
where all time-derivatives are 0. Why then does 𝑅𝐺ൗ𝑑𝑡
have a non-zero expression?
• Answer: Here, it is assumed that there is some excitation
process that is constantly applied, by which EHPs are
constantly created some of whom recombine and the rest
move out elsewhere (eg. as current flow across the sample)
such that at a point in space in the sample, while the rate of
change of overall carrier concentration of a type (e.g.
electrons, 𝑑𝑛Τ𝑑𝑡) is indeed 0 consistent with the meaning of
steady-state, the rate of change of carrier concentration solely
𝑑𝑛𝑅𝐺
due to RG ൗ𝑑𝑡 is non-zero
• Explained nicely in the discussion around Fig. 5.6 in Pierret’s reference
text
ECE102 - VH - L6 14
END OF LECTURE

ECE102 - VH - L6 15
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L7 0
 Agenda
• Transient analysis of excess carriers produced
by a one-time optical excitation
• Steady-state analysis of excess carriers
produced by a constant, ongoing excitation
• Non-equilibrium and Quasi-Fermi levels
(IMREF)
• Photoconductive cells
ECE102 - VH - L7 1
R-G Transient Analysis
• Consider a p-type semicon sample that has been excited by a
short flash of light. Let us see what happens in the immediate
aftermath after stopping the light excitation
• The light flash would have created excess EHP at 𝑡 = 0
• Since EHPs occur in pairs, the initial (ie. at 𝑡 = 0) excess
electron and hole concentrations Δ𝑛 and Δ𝑝 are equal. As the
EHPs recombine in pairs, the instantaneous concentrations of
excess carriers Δ𝑛 𝑡 and Δ𝑝 𝑡 are also equal
• Let us use the SRH model to analyse this time variation (ie.
transient analysis) of electrons (ie. minority carriers)
• Note: Atleast for low-level generation, it is the excess minority
carriers (electrons in this case) that have a more drastic impact on
device behavior than excess majority carriers

ECE102 - VH - L7 2
 R-G Transient Analysis
• Since the SRH model has a non-zero expression for time rate of change
of carrier concentrations solely due to R-G, it can be applied to a
transient analysis of that component of carrier concentrations that are
solely due to R-G. The present example being analysed is a situation
where the only processes in action are R-G processes, hence, the model
can be deemed to apply to the overall carrier concentration
• Reiterating the SRH model but this time dropping the RG subscript due
to reasons stated above, we have:
𝒅𝒏(𝒕)
= 𝜶𝒓 𝒏𝟐𝒊 − 𝜶𝒓 𝒏 𝒕 𝒑(𝒕)
𝒅𝒕
• In the RHS of the above equation:
• First term is the thermal generation rate 𝒈𝒕𝒉_𝒏 𝑻 = 𝜶𝒓 𝒏𝟐𝒊 which always exists
happening in the background
• Second term is the recombination rate 𝒓𝒏 𝑻 = 𝜶𝒓 𝒏 𝒕 𝒑(𝒕)
ECE102 - VH - L7 3
 R-G Transient Analysis
• A similar equation exists for holes too, except that the 𝛼𝑟
coefficient may be different (because the hole capture and
emission dynamics may be different than for electrons in
general). It is especially different when indirect recombination
is the dominant recombination mechanism
• Let 𝑛0 and 𝑝0 be the equilibrium values of electron and hole
concentrations (ie. in the absence of excitation)
• Note: These would be 𝑛𝑖 and 𝑝𝑖 if the semicon is intrinsic. But here
we are considering a general case, viz. a doped semicon (ie. extrinsic)
• Then, it is clear that:
𝑛 𝑡 = 𝑛0 + Δ𝑛(𝑡)
𝑝 𝑡 = 𝑝0 + Δ𝑝 𝑡 = 𝑝0 + Δ𝑛(𝑡) (refer slide 2)
ECE102 - VH - L7 4
 R-G Transient Analysis
• As an example, let us analyze only for electrons (ie. minority carriers in this p-
type sample), by substituting above equations into the SRH model
𝑑 𝑑 Δ𝑛
• We have (after noting that 𝑛0 + Δ𝑛(𝑡) = since 𝑛0 is constant):
𝑑𝑡 𝑑𝑡
𝑑Δ𝑛(𝑡)
= 𝛼𝑟 𝑛𝑖2 − 𝛼𝑟 𝑛0 + Δ𝑛(𝑡) 𝑝0 + Δ𝑛(𝑡)
𝑑𝑡
• Noting that 𝑛0 𝑝0 = 𝑛𝑖2 (refer lecture L5):
𝑑Δ𝑛(𝑡)
= −𝛼𝑟 Δ𝑛(𝑡) 𝑛0 + 𝑝0 + Δ𝑛(𝑡)
𝑑𝑡
• In the square bracket:
• We have 𝑝0 ≫ 𝑛0 since it is a p-type sample, hence 𝑛0 can be neglected
• If we assume a low-level excitation (aka low-level injection) where excess minority
carriers are created but yet in much lesser quantity than equilibrium majority carriers
(ie. 𝑝0 ≫ Δ𝑛 𝑡 ), then we can also neglect Δ𝑛(𝑡)

ECE102 - VH - L7 5
 R-G Transient Analysis
• Thus:
𝒅𝚫𝒏(𝒕)
= −𝜶𝒓 𝒑𝟎 𝚫𝒏(𝒕)
𝒅𝒕
𝟏
• Defining 𝝉𝒏 = , we have the solution to this as:
𝜶𝒓 𝒑 𝟎
𝚫𝒏 𝒕 = 𝚫𝒏 ∙ 𝒆−𝒕Τ𝝉𝒏 where Δ𝑛 = Δ𝑛(𝑡 = 0)
• 𝜏𝑛 is called the recombination lifetime of the
electron. Specifically in this case, it is the minority
carrier lifetime
• Similarly for excess holes in an n-type sample:
𝚫𝒑 𝒕 = 𝚫𝒑 ∙ 𝒆−𝒕Τ𝝉𝒑
ECE102 - VH - L7 6
 R-G Transient Analysis
• Decay of excess electrons and holes
in a p-type GaAs sample by
recombination, for Δ𝑛 = Δ𝑝 = 0.1𝑝0 ,
with n0 negligible, and 𝜏𝑝 = 𝜏𝑛 =
10 𝑛𝑠
• Note: The exponential decay minority
carriers is linear because it this a
semilogarithmic graph
• As can be seen, majority carriers 𝑝(𝑡)
are only marginally changed
ECE102 - VH - L7 7
Steady Excitation
• In transient analysis in earlier slides, we saw how
excess carriers that have once been generated
(by optical excitation or any means), decay with
time after that excitation is removed
• We next analyse a situation where there is a
steady excitation where a non-equilibrium,
steady state is reached (because the EHP creation
by excitation is balanced by recombination)
ECE102 - VH - L7 8
•
Steady Excitation
If a steady light is shone on a semiconductor sample, an optical generation rate 𝑔 will 𝑜𝑝
be added to the background (always existing) thermal generation 𝑔𝑡ℎ , and the carrier
concentrations 𝑛 and 𝑝 will change to new steady-state values. The net generation rate
in this situation is:
𝑔 𝑇 = 𝑔𝑡ℎ + 𝑔𝑜𝑝 , where 𝑔𝑡ℎ = 𝛼𝑟 𝑛𝑖2
• The net recombination rate is: 𝑟 𝑇 = 𝛼𝑟 𝑛𝑝
• In steady state, 𝑔 𝑇 = 𝑟 𝑇 ⇒ 𝒈𝒐𝒑 + 𝜶𝒓 𝒏𝟐𝒊 = 𝜶𝒓 𝒏𝒑. Another way of looking at this
that is consistent with what we learned in the last two slides of L6, is as follows:
𝑑𝑛𝑅𝐺 (𝑡)
= 𝛼𝑟 𝑛𝑖2 − 𝛼𝑟 𝑛 𝑡 𝑝(𝑡)
𝑑𝑡
𝑑𝑛𝑜𝑝 (𝑡)
= 𝑔𝑜𝑝
𝑑𝑡
𝑑𝑛𝑜𝑝 (𝑡)
• Adding the above 2 equations and noting that
𝑑𝑛(𝑡)
𝑑𝑡
=
𝑑𝑛𝑅𝐺 (𝑡)
𝑑𝑡
+
𝑑𝑡
= 𝟎 in steady-state, we get
𝑔𝑜𝑝 + 𝛼𝑟 𝑛𝑖2 − 𝛼𝑟 𝑛𝑝 = 0 ⇒ 𝒈𝒐𝒑 + 𝜶𝒓 𝒏𝟐𝒊 = 𝜶𝒓 𝒏𝒑
ECE102 - VH - L7 9
Steady Excitation
• Proceeding as in the case of transient analysis and
again assuming low-level excitation, and noting that
𝑛0 𝑝0 = 𝑛𝑖2 (refer lecture L5), we get:
𝟏
𝚫𝒏 = 𝒈𝒐𝒑 𝝉𝒏 where 𝝉𝒏 =
𝜶𝒓 𝒏𝟎 +𝒑𝟎
𝟏
𝚫𝒑 = 𝒈𝒐𝒑 𝝉𝒑 where 𝝉𝒑 =
𝜶′𝒓 𝒏𝟎 +𝒑𝟎
• Note: In steady-state as long as the light is shining
constantly, Δ𝑛 and Δ𝑝 don’t change with time
ECE102 - VH - L7 10
 Quasi-Fermi Levels
• Equilibrium carrier statistics (L5) led us to expressions of the form:
𝒏𝟎 = 𝒏𝒊 𝒆 𝑬𝑭 −𝑬𝒊 /𝒌𝑻
𝒑𝟎 = 𝒏𝒊 𝒆 𝑬𝒊 −𝑬𝑭 /𝒌𝑻
• The steady excitation considered in this lecture is a non-
equilibrium state. In non-equilibrium, it is convenient to
define/introduce the concept of quasi-Fermi levels (also called
IMREF levels) 𝐸𝐹𝑛 and 𝐸𝐹𝑝 for electrons and holes separately,
such that the above equilibrium expressions still hold albeit with
those modified quasi-Fermi levels, viz.:
𝒏 = 𝒏𝒊 𝒆 𝑬𝑭𝒏 −𝑬𝒊 /𝒌𝑻
𝒑 = 𝒏𝒊 𝒆 𝑬𝒊 −𝑬𝑭𝒑 /𝒌𝑻
ECE102 - VH - L7 11
Quasi-Fermi Levels
• A given concentration of excess EHPs causes a large
shift in the minority carrier quasi-Fermi level
compared with that for the majority carriers. The
separation of the quasi-Fermi levels 𝐸𝐹𝑛 − 𝐸𝐹𝑝 is a
measure of the deviation from equilibrium (at
equilibrium, we have 𝐸𝐹𝑛 = 𝐸𝐹𝑝 = 𝐸𝐹 )
• Quasi-Fermi levels are very useful in visualizing
minority and majority carrier concentrations inside
devices where these quantities vary with position
ECE102 - VH - L7 12
 Example
• Consider an n-type Si sample with 𝑛0 = 1014 𝑐𝑚−3 . Suppose 1013 𝐸𝐻𝑃/𝑐𝑚3
are created optically every 𝜇𝑠 by constantly shining light on it with 𝜏𝑛 = 𝜏𝑝 =
2𝜇𝑠
• Thus 𝑔𝑜𝑝 = 1013 𝑐𝑚−3 /𝜇𝑠
• Therefore the steady state hole (and electron) concentration is:
Δ𝑝 = Δ𝑛 = 𝑔𝑜𝑝 𝜏𝑛 = 2𝑒13 𝑐𝑚−3
• Hence 𝑛 = 𝑛0 + Δ𝑛 = 1𝑒14 + 2𝑒13 = 1.2𝑒14 𝑐𝑚−3
𝑛𝑖2
• 𝑝0 = = 2.25𝑒6 𝑐𝑚−3 and 𝑝 = 𝑝0 + Δ𝑝 = 2.25𝑒6 + 2𝑒13 ≈ 2𝑒13 𝑐𝑚−3
𝑛0
• Thus we see that:
• Electron concentration (majority) has changed only marginally from 𝑛0 = 1𝑒14 𝑐𝑚 −3 to
1.2𝑒14 𝑐𝑚−3
• Hole concentration
−𝟑
(minority)
−𝟑
has changed drastically (by shining light) from 𝒑𝟎 =
𝟐. 𝟐𝟓𝒆𝟔 𝒄𝒎 to 𝟐𝒆𝟏𝟑 𝒄𝒎
ECE102 - VH - L7 13
 Example (contd.)
• The quasi-Fermi levels can be found using their defining relations:
𝐸 −𝐸 /𝑘𝑇
𝒏 1.2𝑒14
𝑛 = 𝑛𝑖 𝑒 𝐹𝑛 𝑖 ⇒ 𝑬𝑭𝒏 − 𝑬𝒊 = 𝒌𝑻 ∙ 𝒍𝒏 = 26 ∙ 𝑙𝑛
𝒏𝒊 1.5𝑒10
𝐸𝑖 −𝐸𝐹𝑝 /𝑘𝑇
𝒑 2.0𝑒13
𝑝 = 𝑛𝑖 𝑒 ⇒ 𝑬𝒊 − 𝑬𝑭𝒑 = 𝒌𝑻 ∙ 𝒍𝒏 = 26 ∙ 𝑙𝑛
𝒏𝒊 1.5𝑒10
∴ 𝐸𝐹𝑛 − 𝐸𝑖 = 0.233 𝑒𝑉
and 𝐸𝑖 − 𝐸𝐹𝑝 = 0.186 𝑒𝑉
• Thus, 𝐸𝐹𝑛 lies 0.233 𝑒𝑉 above 𝐸𝑖 and 𝐸𝐹𝑝
lies 0.186 eV below 𝐸𝑖
• Note that 𝑛𝑝 = 𝑛𝑖2 cannot be used in
non-equilibrium. That is, 𝑛𝑝 ≠ 𝑛𝑖2
ECE102 - VH - L7 14
 Photoconductive Cells
• Because semicons respond to constant excitation of light of certain
wavelengths (depending on 𝐸𝑔 ) by generating excess carriers, the conductivity
of these semicons also change upon such optical excitation. Hence useful as
photoconductive cells, where change in resistance is sensed upon optical
excitation. This fact will be better appreciated when we cover drift as a
transport mechanism in the lectures
• CdS (𝐸𝑔 = 2.42 𝑒𝑉) is commonly used as a photoconductor in the visible spectra, and
narrow-gap materials such as Ge (𝐸𝑔 = 0.67 𝑒𝑉) and InSb (𝐸𝑔 = 0.18 𝑒𝑉) are useful in
the infrared spectra
• For maximum photoconductive sensitivity, we want high mobilities and long
minority carrier lifetimes
• InSb has an electron mobility of about 105 𝑐𝑚2 Τ𝑉𝑠 and therefore is used as a sensitive
infrared detector in many applications
• The time response of a photoconductive cell is limited by the recombination
time 𝜏 and other factors, some of which can be optimized by proper choice of
material and device geometry, often by trading-off sensitivity
ECE102 - VH - L7 15
END OF LECTURE

ECE102 - VH - L7 16
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L8 0
 Agenda
• Drift current ≜ Carrier transport due
to an electric field

ECE102 - VH - L8 1
Electrons as Moving Particles
In vacuum In semiconductor

𝐹 = −𝑞 𝐸 = 𝑚0 𝑎 𝐹 = −𝑞 𝐸 = 𝑚𝑛∗ 𝑎
where mn∗ is the
conductivity effective
mass (read on)
ECE102 - VH - L8 2
 Conductivity Effective Mass, m*
Under the influence of an electric field (E-field), an electron or a hole is accelerated:
− q
a= electrons
mn*
q
a= *
 holes
mp
From the Drude Model (beyond scope), it can be shown that the Conductivity effective-mass (CEM) is
equal to the Harmonic Mean (HM) of the BSEMs along various directions: transverse, longitudinal, etc

Electron and hole conductivity effective masses

Si Ge GaAs
mn*/mo 0.26 0.12 0.068 mo = 9.110-31 kg
mp*/mo 0.39 0.30 0.50
ECE102 - VH - L8 3
 Carrier Scattering
• Mobile electrons and atoms in the Si lattice are
always in random thermal motion.
• Electrons make frequent collisions with the vibrating atoms
“lattice scattering” or “phonon scattering” – increases with increasing
temperature

• Other scattering mechanisms:

• Deflection by ionized impurity atoms
• Deflection due to Coulombic force between carriers
“carrier-carrier scattering” – only significant at high carrier concentrations

• The net current in any direction is zero, if no EField is

applied.
ECE102 - VH - L8 4
Thermal Velocity, vth
3 1 * 2
Average electron kinetic energy KE = kT = mn vth
2 2

3k T 3  0.026 eV  (1.6  10 −19 J/eV)

vth = =
mn*
0.26  9.1 10 −31 kg
= 2.3  10 5 m/s = 2.3  10 7 cm/s

ECE102 - VH - L8 5
 Carrier Drift
• When an EField (eg. due to an externally applied voltage) exists within a
semicon, mobile charge-carriers will be accelerated by the electrostatic
force: 2
3 1

4 electron
5

E
• Electrons drift in the direction opposite to the EField → net current
• Because of scattering, electrons in a semicon don’t accelerate uniformly.
However, they can be viewed as quasi-classical particles moving at a
constant average drift velocity vdn
ECE102 - VH - L8 6
Electron Momentum
• With every collision, the electron loses
momentum m* v n dn

• Between collisions, the electron gains momentum

–qEtmn

tmn ≡ average time between electron scattering events

Conservation of momentum → |mn*vdn | = | qEtmn|

ECE102 - VH - L8 7
 Carrier Mobility, m
For electrons: 𝑣𝑑𝑛 = 𝑞𝐸𝜏𝑚𝑛ൗ𝑚𝑛∗ ⇒ 𝒗𝒅𝒏 = 𝝁𝒏 𝑬
mn  [qtmn / mn*] is the electron
mobility
𝑞𝐸𝜏
𝑚𝑝
Similarly, for holes: 𝑣𝑑𝑝 = ൗ𝑚𝑝∗ ⇒ 𝒗𝒅𝒑 = 𝝁𝒑 𝑬
mp  [qtmp / mp*] is the hole
mobility
Electron and hole mobilities for intrinsic semiconductors @ 300K
Si Ge GaAs InAs
mn (cm2/Vs) 1400 3900 8500 30,000
mp (cm2/Vs) 470 1900 400 500
ECE102 - VH - L8 8
 Example: Drift Velocity Calculation
• Hole drift velocity in an intrinsic Si sample for E = 103 𝑉/𝑐𝑚:
𝑐𝑚2
𝑣𝑑𝑝 = 𝜇𝑝 𝐸 = 470 ∙ 1000 𝑉ൗ𝑐𝑚 = 4.7𝑒5 𝑐𝑚Τ𝑠
𝑉𝑠
• Thus, drift velocity is much lesser than random thermal velocity 𝑣𝑡ℎ
• Average hole scattering time:
𝑞𝜏𝑚𝑝 𝜇𝑝 𝑚𝑝∗ 𝜇𝑝 ∙ 0.39𝑚0
𝜇𝑝 = ∗ ⇒ 𝜏𝑚𝑝 = =
𝑚𝑝 𝑞 𝑞
2
470 × 10−4 𝑚 ൗ𝑉𝑠 ∙ 0.39 ∙ 9.11 × 10−31 𝑘𝑔 −19 𝑠
= = 1044 × 10
1.6 × 10−16 𝐶𝑜𝑢𝑙𝑜𝑚𝑏
= 0.1044 𝑓𝑠 (𝑓𝑒𝑚𝑡𝑜 𝑠𝑒𝑐𝑜𝑛𝑑𝑠)
ECE102 - VH - L8 9
 Mechanisms of Carrier Scattering
Dominant scattering mechanisms:
1. Phonon scattering (lattice scattering)
2. Impurity (dopant) ion scattering
3. Carrier-carrier scattering (deflection due to Coulombic
force between carriers) – only significant at high carrier
concentrations
Phonon scattering limited mobility decreases with increasing T:
1 1
m phonon  t phonon    T −3 / 2
phonon density  carrier th ermal velocity T  T 1/ 2

m = qt / m Take at face value vth  T

ECE102 - VH - L8 10
Impurity Ion Scattering

There is less change in the electron’s direction if the electron

travels by the ion at a higher speed.
Ion scattering limited mobility increases with increasing T:
3
vth T 3/ 2
mimpurity   Take at face value
N A + ND N A + ND
ECE102 - VH - L8 11
 Matthiessen's Rule
• The probability that a carrier will be scattered by mechanism 𝑖 within a time
period 𝑑𝑡 is 𝑑𝑡Τ𝜏𝑖

𝜏𝑖 ≡ mean time between scattering events due to mechanism 𝑖

→ Probability that a carrier will be scattered by any

mechanism within a time period 𝑑𝑡 is σ𝑖 𝑑𝑡Τ𝜏𝑖

1 1 1 1 1 1
= +  = +
t t phonon t impurity m m phonon mimpurity

ECE102 - VH - L8 12
Mobility Dependence on Doping
Carrier mobilities in Si at 300K

ECE102 - VH - L8 13
 Mobility Dependence on
Temperature
1 1 1
= +
m m phonon mimpurity
• Top 3 traces : When doping is low,
phonon scattering dominates,
which we know degrades mobility
with increasing temperature
• Lower 2 traces: When doping is
high, then ionized impurity
scattering also makes its presence
felt, which we know degrades
mobility with decreasing
temperature
ECE102 - VH - L8 14
 Hole Drift Current Density, Jp,drift
Consider a p-type sample of cross-
sectional area 𝐴, across which a
uniform electric field is applied, and due
to which holes are drifting through it

vdp Dt A = volume from which all holes cross plane in time Dt

p vdp Dt A = number of holes crossing plane in time Dt
q p vdp Dt A = hole charge crossing plane in time Dt
q p vdp A = hole charge crossing plane per unit time = hole current
→ Hole drift current per unit area Jp,drift = q p vdp
ECE102 - VH - L8 15
Conductivity and Resistivity
• In a semiconductor, both electrons and holes conduct current:
J p ,drift = qpm p E J n ,drift = −qn(− m n E )
J drift = J p ,drift + J n ,drift = qpm p E + qnm n E
J drift = (qpm p + qnm n ) E  E

• The conductivity of a semiconductor is   qpm p + qnm n

– Unit: mho/cm
1
• The resistivity of a semiconductor is  
– Unit: ohm-cm 

ECE102 - VH - L8 16
Resistivity Dependence on Doping
For n-type material:
1

qnm n

For p-type material:

1
p- 
type qpm p
n-
Note: This plot (for Si)
type
does not apply to
compensated material
(doped with both
acceptors and donors).

ECE102 - VH - L8 17
Electrical Resistance
V
I _
+
W
t
uniformly doped semiconductor

V L
Resistance R  = [Unit: ohms]
I Wt
where  is the resistivity
ECE102 - VH - L8 18
Example: Resistance Calculation
What is the resistivity of a Si sample doped with 1016/cm3
Boron?
Answer:
1 1
= 
qnm n + qpm p qpm p

= (1.6 10 −19 16
)(10 )(450 ) −1
= 1.4  − cm

ECE102 - VH - L8 19
 Example: Dopant Compensation
Consider the same Si sample doped with 1016/cm3 Boron,
and additionally doped with 1017/cm3 Arsenic. What is its
resistivity?
1 1
Answer: 𝜌 = ≈
𝑛𝑞𝜇𝑛 +𝑝𝑞𝜇𝑝 𝑛𝑞𝜇𝑛
= 1.6𝑒 − 19 9𝑒16 750 −1 = 0.093 Ω − 𝑐𝑚
Points to note:
1. Since the doping concentrations are nearby, we choose to calculate 𝑛 as 𝑛 =
𝑁𝑑 − 𝑁𝑎 = 1𝑒17 − 1𝑒16 = 9𝑒16
a) Since 𝑛0 ≫ 𝑛𝑖 , it is clear that 𝑝0 = 𝑛𝑖2 Τ𝑛0 ≪ 𝑛𝑖 ≪ 𝑛0 , and therefore the 𝑝𝑞𝜇𝑝
term can be neglected
2. The value of 𝜇𝑛 = 750 𝑐𝑚2 /𝑉𝑠 comes from the plot on slide 13, which we use to
get the mobility at total dopant concentration 𝑁𝑑 + 𝑁𝑎 = 1𝑒17 + 1𝑒16 = 1.1𝑒17
ECE102 - VH - L8 20
Example: T Dependence of 
Consider a Si sample doped with 1017cm-3 As. How will its
resistivity change when the temperature is increased from T=300K
to T=400K?
Answer:
The temperature dependent factor in  (and therefore ) is mn.
From the mobility vs. temperature curve for 1017 cm-3, we find that
mn decreases from 770 at 300K to 400 at 400K.

770
Thus,  increases by = 1.93 times
400

ECE102 - VH - L8 21
END OF LECTURE

ECE102 - VH - L8 22
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L9 0
 Agenda
• Diffusion current ≜ Carrier transport
due to a concentration gradient

ECE102 - VH - L9 1
Diffusion
Particles diffuse from regions of higher concentration to regions
of lower concentration region, due to random thermal motion.

ECE102 - VH - L9 2
 Diffusion
• Spatial variation (gradient) in 𝑛 and 𝑝 results in
a motion of carriers from regions of high carrier
concentration to regions of low carrier
concentration. This type of motion is called
diffusion and is an important charge transport
process in semiconductors
• Diffusion current can flow even if there is no
externally applied electric field
ECE102 - VH - L9 3
 Diffusion of a Pulse of Electrons
• Imagine a pulse of high
electron concentration
created somehow at 𝑥 =
0 at time 𝑡 = 0
• Picture shows the spatial
distribution of carriers at
various instants of time
𝑡1 , 𝑡2 , 𝑡3 , etc
• The peak flattens out
over time due to
electrons diffusing out
ECE102 - VH - L9 4
 Diffusion
• From fundamental considerations, it can be shown that
the carrier flux (#carriers flowing per unit cross-section
area per unit time) at a point is proportional to the
spatial gradient of the carrier concentration (ie. of the
volume-density) at that point
• Diffusion current density 𝐽 (ie. current per unit cross-
sectional area) due to a carrier type (e.g. electrons) is
simply equal to this carrier-flux multiplied by charge,
and keeping in mind the direction of diffusion
ECE102 - VH - L9 5
 Diffusion
• Thus, for one-dimensional (1D) diffusion transport, we have:
𝑑𝑛(𝑥)
𝐽𝑛𝑑𝑖𝑓𝑓 (𝑥) = +𝑞𝐷𝑛
𝑑𝑥
𝑑𝑝(𝑥)
𝐽𝑝𝑑𝑖𝑓𝑓 (𝑥) = −𝑞𝐷𝑝
𝑑𝑥
• 𝐷𝑛 (𝐷𝑝 ) is called the electron (hole) diffusion coefficient or
diffusivity (units being 𝑐𝑚2 /𝑠)
• Think why the signs are as above?
• Hint: Electrons and holes move together in a carrier gradient, but the
resulting currents are in opposite directions because of the opposite
charge of electrons and holes.
ECE102 - VH - L9 6
Total Current: Drift + Diffusion
• If an externally applied electric field 𝐸 as well as a
concentration gradient (of electrons and holes) is there,
then there is:
• Drift currents due to:
• Electrons and
• Holes
• Diffusion currents due to:
• Electrons and
• Holes
• Thus, 4 components
ECE102 - VH - L9 7
 Total Current: Drift + Diffusion
• Total 1D current density due to electrons is:
𝒅𝒏(𝒙)
𝑱𝒏 𝒙 = 𝐽𝑛 𝑑𝑟𝑖𝑓𝑡 + 𝐽𝑛(𝑑𝑖𝑓𝑓) = 𝒏𝒒𝝁𝒏 𝑬(𝒙) + 𝒒𝑫𝒏
𝒅𝒙
𝒅𝒑(𝒙)
𝑱𝒑 𝒙 = 𝐽𝑝 𝑑𝑟𝑖𝑓𝑡 + 𝐽𝑝(𝑑𝑖𝑓𝑓) = 𝒑𝒒𝝁𝒑 𝑬 𝒙 − 𝒒𝑫𝒑
𝒅𝒙
• The total current density is:
𝑱 𝒙 = 𝑱𝒏 𝒙 + 𝑱𝑝 (𝒙)
ECE102 - VH - L9 8
Important Take-away
• Drift current depends on carrier concentration
• Thus, drift current is mostly due to majority carriers
• Diffusion current depends on gradient of carrier
concentration
• Hence, minority carriers can contribute significantly
to the total diffusion current, exceeding the total drift
current in some devices in specific locations (e.g. in
diodes close to the junction, as we’ll see later)

ECE102 - VH - L9 9
 Side-note
• When a uniform EField 𝐸 is along
the +𝑥 direction, the electron
energy bands are tilted as shown.
Why?
• Explanation:
• Electric field 𝐸 = − 𝑑𝑉Τ𝑑𝑥
• But Potential 𝑉 = − 𝐸𝑖ൗ𝑞 (because energy bands
traditionally show electron energies)
• Why’d we choose 𝐸𝑖 ? Why not 𝐸𝑐 or 𝐸𝑣 ? Think!
1 𝑑𝐸𝑖
• Hence, Electric field 𝐸 =
𝑞
ൗ𝑑𝑥
𝑑𝐸𝑖
• Thus, +ve EField ⇒ +ve ൗ𝑑𝑥 ⇒ Electron energy
bands increase as you go along +𝑥
ECE102 - VH - L9 10
 Side-note
• Hence, when a uniform EField 𝐸 is
along the +𝑥 direction, the electron
energy bands are tilted as shown
alongside
• Cue to remember: In an electron
energy band diagrams:
• Electrons are like balls, they roll downhill
• Holes are like balloons, they float uphill

ECE102 - VH - L9 11
 Einstein Relation
• At equilibrium, there is no current
• Principle of detailed balance (applicable at equilibrium)
implies that it’s not just that the sum total of 4 components
is 0, but rather sum total of 2 components are individually 0
(for electrons and holes individually). That is:
𝐽𝑛 𝑥 = 0
𝐽𝑝 𝑥 = 0
• Considering first for electrons, we have:
𝑑𝑛 𝑥
𝑛𝑞𝜇𝑛 𝐸 𝑥 + 𝑞𝐷𝑛 =0
ECE102 - VH - L9
𝑑𝑥 12
 Einstein Relation
𝐸𝐹 𝑥 −𝐸𝑖 𝑥
• But we know that 𝑛 𝑥 = 𝑛𝑖 𝑒 ⇒ 𝑘𝑇
𝒅𝒏 𝑛𝑖 𝐸𝐹 −𝐸𝑖 𝑑𝐸𝐹 𝑑𝐸𝑖 𝒏 𝒅𝑬𝑭 𝒅𝑬𝒊
= 𝑒 𝑘𝑇 − = −
𝒅𝒙 𝑘𝑇 𝑑𝑥 𝑑𝑥 𝒌𝑻 𝒅𝒙 𝒅𝒙
• Thus:
𝑛 𝑑𝐸𝐹 𝑑𝐸𝑖
𝑛𝑞𝜇𝑛 𝐸 𝑥 + 𝑞𝐷𝑛 − =0
𝑘𝑇 𝑑𝑥 𝑑𝑥
𝐷𝑛 𝑑𝐸𝐹 𝑑𝐸𝑖
∴ 𝜇𝑛 𝐸 𝑥 + − =0
𝑘𝑇 𝑑𝑥 𝑑𝑥
ECE102 - VH - L9 13
 Einstein Relation
𝑑𝐸𝐹
• But we know that = 0 in equilibrium (see lecture L5)
𝑑𝑥
𝑑𝐸𝑖
• We also know that = 𝑞𝐸 where 𝐸 is the EField (slide
𝑑𝑥
10)
𝐷𝑛 𝑫𝒏 𝒌𝑻
∴ 𝜇𝑛 𝐸 + 0 − 𝑞𝐸 = 0 ⇒ =
𝑘𝑇 𝝁𝒏 𝒒
𝑫𝒑 𝒌𝑻
• It can similarly be shown for holes that =
𝝁𝒑 𝒒
• These are called the Einstein relations
ECE102 - VH - L9 14
END OF LECTURE

ECE102 - VH - L9 15
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L10 0
 Agenda
• Continuity Equation: An equation
about the conservation of #particles!

ECE102 - VH - L10 1
 Continuity Equation
• So far we have discussed the following physical processes:
• Generation & recombination (RG)
• Drift current
• Diffusion current
• Let us now analyze a 1D
situation when all of these
processes are taking place
simultaneously. Consider a
differential length Δ𝑥 of a
semiconductor sample with
cross-sectional area 𝐴
ECE102 - VH - L10 2
Continuity Equation
• Consider the elemental cuboidal volume (Δ𝑉 =
𝐴Δ𝑥) of the slab (shown shaded) formed by
length Δ𝑥 and cross-sectional area 𝐴. Without
any loss of generality, let us first analyze for holes
• The net increase in hole concentration per unit
time 𝜕𝑝ൗ𝜕𝑡 within the slab is simply the difference
between the hole flux per unit volume entering
versus leaving the slab, plus the net hole
generation rate within the slab due to RG
ECE102 - VH - L10 3
Continuity Equation
• Here, the hole flux per unit volume at a point being
the time rate of #holes flowing per unit volume at
that point
• If 𝐽𝑝 𝑥 is the hole current density (ie. current per
unit cross-sectional area) at 𝑥, then obviously
𝐽𝑝 𝑥 Τ𝑞 is the #holes crossing point 𝑥 per unit cross-
sectional area, and therefore 𝐽𝑝 𝑥 Τ 𝑞Δ𝑥 is the
#holes crossing point 𝑥 per unit volume
𝐽𝑝 𝑥
• Thus, Hole flux per unit volume at 𝑥 is
𝑞Δ𝑥
ECE102 - VH - L10 4
 Continuity Equation
• Thus, we have:
𝜕𝑝(𝑥) 𝐽𝑝 𝑥 𝐽𝑝 𝑥 + Δ𝑥
= − + 𝑁𝑒𝑡𝐻𝑜𝑙𝑒𝐺𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑅𝑎𝑡𝑒𝐷𝑢𝑒𝑇𝑜𝑅𝐺
𝜕𝑡 𝑞Δ𝑥 𝑞Δ𝑥
• Now, in the absence of any other excitation mechanisms (optical, etc), the
𝑁𝑒𝑡𝐻𝑜𝑙𝑒𝐺𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑅𝑎𝑡𝑒𝐷𝑢𝑒𝑇𝑜𝑅𝐺 is simply 𝑔𝑡ℎ − 𝑟𝑃 where: 𝑔𝑡ℎ is the thermal
generation rate of EHP = 𝛼𝑟 𝑛𝑖2 (always present in the background) and 𝑟𝑃 = Hole
recombination rate = 𝛼𝑟 𝑛𝑝
• Thus, 𝑁𝑒𝑡𝐻𝑜𝑙𝑒𝐺𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑅𝑎𝑡𝑒𝐷𝑢𝑒𝑇𝑜𝑅𝐺 = 𝛼𝑟 𝑛𝑖2 − 𝛼𝑟 𝑛𝑝
• Under low-level generation conditions, we have already seen that this can be
approximated as − 𝛿𝑝Τ𝜏𝑝 where 𝛿𝑝 is the excess hole concentration (refer Slide 10 in
lecture L7). Thus, we have:
𝝏𝒑(𝒙) 𝐽𝑝 𝑥 𝐽𝑝 𝑥+Δ𝑥 𝜕𝑝 𝑱𝒑 𝒙 −𝑱𝒑 𝒙+𝜟𝒙 𝜹𝒑
= − −𝜏 = −𝝉
𝝏𝒕 𝑞Δ𝑥 𝑞Δ𝑥 𝑝 𝒒𝜟𝒙 𝒑

ECE102 - VH - L10 5
 Continuity Equation
• In the limit as 𝛥𝑥 approaches 0, this becomes
𝝏𝒑 𝒙, 𝒕 𝟏 𝝏𝑱𝒑 𝒙, 𝒕 𝜹𝒑 𝒙, 𝒕
=− −
𝝏𝒕 𝒒 𝝏𝒙 𝝉𝒑
• This is the hole continuity equation, a very important equation
• Notice that the numerator in the last term in the RHS is 𝛿𝑝 (ie. excess hole
concentration). Do not misread it as 𝜕𝑝!
• Similarly, if we go through the motions for electrons, we get the
electron continuity equation as:
𝝏𝒏(𝒙, 𝒕) 𝟏 𝝏𝑱𝒏 (𝒙, 𝒕) 𝜹𝒏 𝒙, 𝒕
= −
𝝏𝒕 𝒒 𝝏𝒙 𝝉𝒏
• Note: Both 𝑛 𝑥, 𝑡 and 𝑝 𝑥, 𝑡 (as also 𝛿𝑛 and 𝛿𝑝) are functions of space and
time in general, hence the partial derivatives
ECE102 - VH - L10 6
Continuity Eqn & Diffusion-only
• A special case of interest for applying the continuity
equation is to a case when there is predominantly
diffusion currents, and negligible drift current. This is a
surprisingly common case encountered (e.g. PN junction
diodes). Here, we have:
𝜕𝑝(𝑥)
𝐽𝑝 𝑥 ≈ 𝐽𝑝 𝑑𝑖𝑓𝑓 𝑥 = −𝑞𝐷𝑝
𝜕𝑥
𝜕𝑝(𝑥,𝑡) 𝜕𝑝0 (𝑥,𝑡)
𝜕 𝛿𝑝(𝑥,𝑡)
• But 𝑝 = 𝑝0 + 𝛿𝑝 ⇒ = +
𝜕𝑡 𝜕𝑡 𝜕𝑡
𝜕𝑝 (𝑥,𝑡)
• 𝑝0 being an equilibrium concentration, clearly 0 =0
𝜕𝑡
𝝏𝒑 𝒙,𝒕 𝝏 𝜹𝒑 𝒙,𝒕 𝝏𝒑 𝒙,𝒕 𝝏 𝜹𝒑 𝒙,𝒕
• Thus, = . Similarly, =
𝝏𝒕 𝝏𝒕 𝝏𝒙 𝝏𝒙
ECE102 - VH - L10 7
 Continuity Eqn & Diffusion-only
• Substituting these into the hole continuity equation, we have:
𝝏 𝜹𝒑 𝒙, 𝒕 𝝏𝟐 𝜹𝒑 𝒙, 𝒕 𝜹𝒑 𝒙, 𝒕
= 𝑫𝒑 𝟐
−
𝝏𝒕 𝝏𝒙 𝝉𝒑
• It is a 2nd order partial differential equation (PDE) in 𝛿𝑝 𝑥, 𝑡
• Similarly, for electrons, we get:
𝝏 𝜹𝒏 𝒙, 𝒕 𝝏𝟐 𝜹𝒏 𝒙, 𝒕 𝜹𝒏 𝒙, 𝒕
= 𝑫𝒏 −
𝝏𝒕 𝝏𝒙𝟐 𝝉𝒏
• These equations are called diffusion equations for holes and electrons
respectively. They are useful in solving transient problems of diffusion
with RG. For e.g., a pulse of electrons in a semiconductor (slide 4 in L9)
spreads out by diffusion and disappears by recombination
ECE102 - VH - L10 8
Excess Carriers Diffusion - Steady State
• In lecture L7, we briefly looked at 𝐸𝐹 − 𝐸𝑖 (position of 𝐸𝐹
wrt 𝐸𝑖 ) in a case when there is steady excess carriers
generated. There we didn’t bother about how these excess
carriers out-diffuse
• We now look at that aspect (out-diffusion of excess
carriers). For this, we need to solve the diffusion equations
• We neglect drift, and just consider diffusion here
• Consider an n-type sample, let us model the hole out-
diffusion (minority carrier out-diffusion)
• Clearly, for minority carriers, drift can be ignored
ECE102 - VH - L10 9
Excess Carriers Diffusion - Steady State
• We restate the hole diffusion equation below, with, the time
derivatives zeroed out (since steady state)
𝜕 2 𝛿𝑝 𝑥, 𝑡 𝛿𝑝 𝑥, 𝑡 𝜕 𝛿𝑝 𝑥, 𝑡
𝐷𝑝 2 − = =0
𝜕𝑥 𝜏𝑝 𝜕𝑡
𝜕 2 𝛿𝑝 𝑥, 𝑡 𝛿𝑝 𝑥, 𝑡 𝝏𝟐 𝜹𝒑 𝜹𝒑 𝜹𝒑
∴ 𝐷𝑝 2 = ⇒ 𝟐 = = 𝟐
𝜕𝑥 𝜏𝑝 𝝏𝒙 𝑫𝒑 𝝉𝒑 𝑳𝒑
• where:
• we have dropped showing the space/time dependence for brevity
• A new term 𝐿𝑝 is introduced by defining it as 𝐿2𝑝 = 𝐷𝑝 𝜏𝑝 . It is called the hole
diffusion length
• Similar analogous terms exist for electrons too
ECE102 - VH - L10 10
Excess Carriers Diffusion - Steady State
• Given the boundary conditions that the excess carriers
𝛿𝑝 𝑥 = 0 = Δ𝑝 and 𝛿𝑝 𝑥 = ∞ = 0, the solution to this can
easily be solved as:
𝜹𝒑 𝒙 = 𝚫𝒑 ∙ 𝒆−𝒙Τ𝑳𝒑
• A classic use-case of this analysis is an n-type semiconductor
sample where excess holes are somehow injected at 𝑥 = 0, and
the steady state hole injection maintains a constant excess hole
concentration at the injection point 𝛿𝑝(𝑥 = 0) = Δ𝑝. The
injected holes diffuse along the bar, recombining with a
characteristic lifetime 𝜏𝑝 . In steady state we expect the
distribution of excess holes to decay to zero for large values of 𝑥
(ie. far from the point of injection), because of the
recombination
ECE102 - VH - L10 11
Excess Carriers Diffusion - Steady State
• The figure alongside shows
this carrier distribution in
steady-state
• Consistent with the hole
concentration gradient, there
is a hole diffusion current
along +x direction
• The injected excess holes die
out exponentially in 𝑥 due to
recombination, and the
diffusion length 𝐿𝑝 represents
the distance at which the
excess hole distribution is
reduced to 1/𝑒 of its value at
the point of injection
ECE102 - VH - L10 12
 Example
• In a very long p-type Si bar with cross-sectional
area = 0.5 𝑐𝑚2 and 𝑁𝑎 = 1𝑒17 𝑐𝑚−3 , we inject
holes such that the steady state excess hole
concentration is 5𝑒16 𝑐𝑚−3 at 𝑥 = 0.
Neglecting drift currents, and assuming that
𝜇𝑝 = 500 𝑐𝑚2 /𝑉𝑠 and 𝜏𝑝 = 0.1 𝑛𝑠 …
• What is the steady state separation between 𝐹𝑝 and
𝐸𝑖 at 𝑥 = 100 𝑛𝑚?
• What is the hole current there?
ECE102 - VH - L10 13
 Solution
• Here, even though it is a majority carrier that we are analyzing,
we ignore drift because of specifically being asked to do so. So we
can apply our earlier analysis for excess hole concentration:
−𝑥 Τ𝐿𝑝
𝛿𝑝 𝑥 = Δ𝑝 ∙ 𝑒
• From the Einstein relation we have:
𝑘𝑇
𝐷𝑝 = 𝜇𝑝 = 0.026𝑉 ∙ 500 𝑐𝑚2 Τ𝑉𝑠 = 13 𝑐𝑚2 /𝑠
𝑞
• Hence 𝐿𝑝 = 𝐷𝑝 𝜏𝑝 = 360 𝑛𝑚
• 𝑝 𝑥 = 𝑝0 + Δ𝑝 ∙ 𝑒 −𝑥Τ𝐿𝑝 ⇒ 𝑝 100 𝑛𝑚 = 1𝑒17 + 5𝑒16 ∙
𝑒 −100Τ360 = 1.379𝑒17 𝑐𝑚−3
• Given that 𝑝 = 𝑝𝑖 ∙ 𝑒 𝐸𝑖−𝐹𝑝 Τ𝑘𝑇 , we have 𝐸𝑖 − 𝐹𝑝 = 𝑘𝑇 ∙ 𝑙𝑛 𝑝Τ𝑝𝑖
ECE102 - VH - L10 14
 Solution
• Hence 𝐸𝑖 − 𝐹𝑝 = 26 𝑚𝑒𝑉 ∙ 𝑙𝑛 1.379𝑒17Τ1.5𝑒10 =
0.417 𝑒𝑉
• As expected, being a majority carrier, it hasn’t changed much
𝜕𝑝 𝑥
• Hole current 𝐼(𝑥) = 𝐽𝑝(𝑑𝑖𝑓𝑓) 𝑥 𝐴 = −𝑞𝐷𝑝 𝐴 =
𝜕𝑥
Δ𝑝 −𝑥 Τ𝐿𝑝
𝑞𝐷𝑝 𝐴 ∙ 𝑒
𝐿𝑝
Δ𝑝 −100Τ360
• Hence, 𝐼 100 𝑛𝑚 = 𝑞𝐷𝑝 𝐴 ∙ 𝑒 = 1.6𝑒 − 19 ∙
𝐿𝑝
1
13 ∙ 0.5 ∙ 5𝑒16 ∙ ( ) ∙ 0.757 = 1093 𝐴
360𝑒−7
ECE102 - VH - L10 15
 Gradient in Quasi-Fermi Levels
• It can be shown very easily (using the Einstein relation) and the
drift+diffusion current expression for each carrier individually and
noting that the band bending in the band diagram implies an
electric field (see lecture L9.Slide10), that the total current due to
one carrier can be written as a gradient of the quasi-Fermi level of
that carrier. For example, for electrons, we can show that:
𝑑𝐹𝑛
𝐽𝑛 𝑥 = 𝜇𝑛 𝑛 𝑥
𝑑𝑥
• Thus, electron (hole) drift + diffusion total current is described by
the spatial variation of the electron (hole) quasi-Fermi level
ECE102 - VH - L10 16
END OF LECTURE

ECE102 - VH - L10 17
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L11 0
Agenda
• P-N junctions
• Equilibrium electrostatic analysis of
P-N junctions

ECE102 - VH - L11 1
 P-N Junctions
• Most semicon devices contain junctions between p-type and n-
type material. These p-n junctions are fundamental to the
operation of devices like rectifier, voltage-dependent capacitor,
photocell, light-emitting diode (LED), etc in electronic circuits
• Current-vs-voltage (I-V) characteristics

ECE102 - VH - L11 2
 P-N Junctions
• Figure shows the cross-section of a pn-junction, as seen in a
discrete diode. A p-region is diffused (or implanted) into an
n-type wafer. The surface is passivated with SiO2 and ohmic
contacts are made to the front and back of the wafer

ECE102 - VH - L11 3
 Equilibrium Electrostatics
• The purpose of this analysis is to get expressions for key
metrics within the device, in terms of controllable
quantities like doping concentrations and dimensions
• To ease the analysis, we consider a step junction, which has
uniform P doping on one side of a sharp junction and
uniform N doping on the other side
• This approximation suits epitaxial junctions well (but not diffused
or implanted junctions, which form graded junctions, ie. 𝑁𝑑 – 𝑁𝑎
varies over a significant distance on either side of the junction)
• Consider a thought experiment of P and N regions brought
together suddenly (with perfect crystalline orientation and
continuity across the junction)
ECE102 - VH - L11 4
 Built-in Potential at Equilibrium
• Before they are joined, the N
region has a large density of
electrons and few holes, and
converse for the P region. Upon
joining the two regions, holes
diffuse from the P side into the
N side, and electrons diffuse
from N to P. The resulting
diffusion leaves behind
uncompensated ions (ie.
charged), that soon results in
an opposing electric field that
prevents any further diffusion
ECE102 - VH - L11 5
 Built-in Potential at Equilibrium
• The resulting EField is directed from the positive
charge toward the negative charge. Thus EField is
in the direction opposite to that of diffusion
current for each type of carrier
• Since there can be no net buildup of electrons or
holes on either side of the junction, the drift and
diffusion currents must cancel for each type of
carrier:
𝐽𝑛 = 0 and 𝐽𝑝 = 0. That is:
𝐽𝑛(𝑑𝑟𝑖𝑓𝑡) + 𝐽𝑛(𝑑𝑖𝑓𝑓) = 0 and 𝐽𝑝(𝑑𝑟𝑖𝑓𝑡) + 𝐽𝑝(𝑑𝑖𝑓𝑓) = 0
ECE102 - VH - L11 6
 Built-in Potential at Equilibrium
• The EField builds up to the point where the net
current is zero at equilibrium. The region W
about the junction where EField appears, is
called the space-charge region (also known
depletion region or the transition region)
• Note: Space charge exists within W (due to
immobile ions but there are almost no mobile
carriers, ie. region W is depleted of carriers) while
neutrality is maintained outside this region
• As a result, an equilibrium potential difference
𝑉0 known as the built-in potential (also called
the contact potential) develops across W
ECE102 - VH - L11 7
 Built-in Potential at Equilibrium
• 𝐸𝑣 and 𝐸𝑐 are higher on the p side of the
junction than on the n side by the amount
𝑞𝑉0
• The separation of the bands at equilibrium
is just that required to make the 𝐸𝐹 level
constant throughout the device (recall the
lemma about the Fermi level 𝐸𝐹 having 0
spatial variation at equilibrium)
• Note, here 𝐸𝐹𝑝 and 𝐸𝐹𝑛 are not quasi-Fermi
levels. They are just Fermi levels on the p-side
and the n-side respectively
ECE102 - VH - L11 8
 Built-in Potential at Equilibrium
• Let us consider the equilibrium current
equation for holes, to start with:
𝐽𝑝(𝑑𝑟𝑖𝑓𝑡) + 𝐽𝑝(𝑑𝑖𝑓𝑓) = 0
𝑑𝑝 𝑥
∴ 𝑝𝑞𝜇𝑝 𝐸 𝑥 − 𝑞𝐷𝑝 =0
𝑑𝑥
𝐷𝑝 𝑑𝑝 𝑥
∴ 𝐸 𝑥 𝑑𝑥 =
𝜇𝑝 𝑝
𝑘𝑇 𝑑𝑝 𝑥
∴𝐸 𝑥 𝑑𝑥 = (using the Einstein relation)
𝑞 𝑝
𝑘𝑇 𝑑𝑝 𝑥 𝑑𝑉
∴ −𝑑𝑉 = (since 𝐸 = − )
𝑞 𝑝 𝑑𝑥
ECE102 - VH - L11 9
 Built-in Potential
𝑘𝑇 𝑑𝑝 𝑥
at Equilibrium
−𝑑𝑉 =
𝑞 𝑝
𝑉𝑛 𝑘𝑇 𝑝𝑛 𝑑𝑝 𝑥
∴ − ‫𝑉𝑑 𝑉׬‬ = ‫׬‬ (where 𝑝𝑝 & 𝑝𝑛 are the
𝑝 𝑞 𝑝𝑝 𝑝
hole densities on p- and n-sides)
𝑘𝑇 𝑝𝑛
∴ 𝑉𝑝 − 𝑉𝑛 = 𝑙𝑛
𝑞 𝑝𝑝
𝑘𝑇 𝑝𝑛 𝑘𝑇 𝑝𝑝
𝑖𝑒. −𝑉0 = 𝑙𝑛 ⇒ 𝑉0 = 𝑙𝑛
𝑞 𝑝𝑝 𝑞 𝑝𝑛
𝑘𝑇 𝑁𝑎 𝒌𝑻 𝑵𝒂 𝑵𝒅
∴ 𝑉0 = 𝑙𝑛 ⇒ 𝑽𝟎 = 𝒍𝒏
𝑞 2
𝑛𝑖 Τ𝑁𝑑 𝒒 𝒏𝟐𝒊
ECE102 - VH - L11 10
 Built-in Potential at Equilibrium
• Another useful form is derived as follows:
𝑘𝑇 𝑝𝑝 𝑝𝑝
𝑉0 = 𝑙𝑛 ⇒ = 𝑒 𝑞𝑉0 Τ𝑘𝑇
𝑞 𝑝𝑛 𝑝𝑛
But 𝑝𝑝 𝑛𝑝 = 𝑛𝑖2 = 𝑝𝑛 𝑛𝑛 in equilibrium
𝑝𝑝 𝑛𝑛
∴ =
𝑝𝑛 𝑛𝑝
𝒑𝒑 𝒏𝒏
∴ = = 𝒆𝒒𝑽𝟎 Τ𝒌𝑻
𝒑𝒏 𝒏𝒑
• In other words: For each carrier type, the ratio of
its density as a majority carrier to that as a
minority carrier is equal to 𝑒 𝑞𝑉0 Τ𝑘𝑇
ECE102 - VH - L11 11
 Depletion Approximation
• The figure alongside shows the spatial variation
of the charge density (immobile, of course) and
the Efield in the depletion region (DR). We now
derive these using the following steps:
• Use the depletion approximation regarding the
charge density 𝜌(𝑥) in the DR
• Using this 𝜌(𝑥), solve the Poisson equation to find
𝐸 𝑥 (using BC that EField is 0 outside the DR)
• Integrate 𝐸 𝑥 to obtain 𝑉 𝑥 , noting the already
calculated value for built-in potential 𝑉0
• For 𝐸(𝑥) to be continuous at 𝑥 = 0, enforce that
𝑁𝑎 𝑥𝑝0 = 𝑁𝑑 𝑥𝑛0 , and solve for 𝑥𝑝0 , 𝑥𝑛0
ECE102 - VH - L11 12
 Depletion Approximation
• Depletion Approximation (DA) is the
approximation that the space-charge region
is completely devoid of mobile carriers
𝒅𝟐 𝑽 𝝆
• Poisson’s equation in 1D: = −
𝒅𝒙𝟐 𝝐
where 𝜌 is the net charge density and 𝜖 is
the permittivity
𝑑𝑉
𝑑 𝑑 −𝐸 𝑑𝐸 𝜌 𝒅𝑬 𝝆
𝑑𝑥
= =− =− ⇒ =
𝑑𝑥 𝑑𝑥 𝑑𝑥 𝜖 𝒅𝒙 𝝐
ECE102 - VH - L11 13
 Depletion Approximation
• Assume origin of x-axis to be at the
metallurgical junction, and that 𝑥𝑛0 and
– 𝑥𝑝0 are the coordinates of the DR edges
on the n- and p-sides
• Now, charge density in the DR (assuming
the DA and full ionization) is simply:
𝜌 𝑥 = 𝑞𝑁𝑑 𝑜𝑛 𝑛 − 𝑠𝑖𝑑𝑒 (𝑓𝑜𝑟 0 ≤ 𝑥 ≤ 𝑥𝑛0 )
𝜌 𝑥 = −𝑞𝑁𝑎 𝑜𝑛 𝑝 − 𝑠𝑖𝑑𝑒 (𝑓𝑜𝑟 − 𝑥𝑝0 ≤ 𝑥 ≤ 0)
ECE102 - VH - L11 14
 Depletion Approximation
• Given the different charge densities on the p- and n-sides, we
solve the Poisson equation separately for the two sides:
𝑑𝐸 𝜌 𝑑𝐸 −𝑞𝑁𝑎
• P-side: 𝑑𝑥 = ⇒ =
𝜖 𝑑𝑥 𝜖
𝑑𝐸 𝜌 𝑑𝐸 𝑞𝑁𝑑
• N-side: = 𝜖 ⇒ 𝑑𝑥 = 𝜖
𝑑𝑥
• Considering first the P-side, we have
𝐸0 0
𝑑𝐸 −𝑞𝑁𝑎 𝑞𝑁𝑎
= ⇒ න 𝑑𝐸 = − න 𝑑𝑥
𝑑𝑥 𝜖 𝜖
0 −𝑥𝑝0
𝑞𝑁𝑎 𝑥𝑝0
∴ 𝐸0 = − where 𝐸0 is the EField at the junction (peak EField)
𝜖
𝑞𝑁𝑑 𝑥𝑛0
• Proceeding similarly for the N-side, we’ll get 𝐸0 = − 𝜖
𝒒𝑵𝒂 𝒙𝒑𝟎 𝒒𝑵𝒅 𝒙𝒏𝟎
• Thus: 𝑬𝟎 = − =−
𝝐 𝝐
ECE102 - VH - L11 15
 Depletion Approximation
• Thus: 𝑵𝒂 𝒙𝒑𝟎 = 𝑵𝒅 𝒙𝒏𝟎
• This is also consistent with implications of continuity of
the EField (obtained from Gauss Law) at the junction
• Now, for the P-side, we have
𝐸0 0
𝑑𝐸 −𝑞𝑁𝑎 𝑞𝑁𝑎
= ⇒ න 𝑑𝐸 = − න 𝑑𝑥
𝑑𝑥 𝜖 𝜖
𝐸 𝑥 𝑥
𝑞𝑁𝑎 𝑥 𝑞𝑁𝑎 𝑥
∴ 𝐸0 − 𝐸 𝑥 = 𝐸 𝑥 = ⇒ = 𝐸0 −
𝑞𝑁𝑎 𝑥𝑝0 𝑞𝑁 𝑥 𝜖
𝒒𝑵 𝜖
− − 𝑎 ⇒ 𝑬 𝒙 = − 𝒂 𝒙 + 𝒙𝒑𝟎 on the
𝜖 𝜖 𝝐
P-side
ECE102 - VH - L11 16
 Depletion Approximation
• We have on the p-side:
𝑞𝑁𝑎
𝐸 𝑥 =− 𝑥 + 𝑥𝑝0 on the P-side
𝜖
• But 𝐸 𝑥 = −𝑑𝑉 Τ𝑑𝑥
𝑑𝑉 𝑞𝑁𝑎
∴ = 𝑥+𝑥
𝑑𝑥 𝜖 𝑉𝑚𝑗 𝑝0 0
𝑞𝑁𝑎 𝑞𝑁𝑎
∴ 𝑑𝑉 = 𝑥 + 𝑥𝑝0 𝑑𝑥 ⇒ න 𝑑𝑉 = න 𝑥 + 𝑥𝑝0 𝑑𝑥
𝜖 𝜖
𝑉𝑝 −𝑥𝑝0

where 𝑉𝑚𝑗 is the potential at the metallurgical junction

𝑞𝑁𝑎 𝑥 2 0 𝑞𝑁𝑎 2
𝑥𝑝0 ൘ − 𝑥𝑝0 2
∴ 𝑉𝑚𝑗 − 𝑉𝑝 = ൗ2 + 𝑥𝑥𝑝0 = 0− 2
𝜖 −𝑥𝑝0 𝜖
2
𝑞𝑁𝑎 𝑥𝑝0
∴ 𝑉𝑚𝑗 − 𝑉𝑝 =
2𝜖
2
𝑞𝑁𝑑 𝑥𝑛0
• A similar analysis on the N-side will give: 𝑉𝑛 − 𝑉𝑚𝑗 =
2𝜖

ECE102 - VH - L11 17
 Depletion Approximation
• Repeating the equations for ease of readability:
2
𝑞𝑁𝑎 𝑥𝑝0
𝑉𝑚𝑗 − 𝑉𝑝 =
2𝜖 2
𝑞𝑁𝑑 𝑥𝑛0
𝑉𝑛 − 𝑉𝑚𝑗 =
2𝜖
• Adding the above 2 equations eliminates 𝑉𝑚𝑗 and
we get:
𝑞 2 2
𝑉𝑛 − 𝑉𝑝 = 𝑁𝑎 𝑥𝑝0 + 𝑁𝑑 𝑥𝑛0
2𝜖
• But 𝑉𝑛 − 𝑉𝑝 = 𝑉0
ECE102 - VH - L11 18
•
Depletion
Thus we have:
Approximation
𝑞 2 2
𝑉0 = 𝑁𝑎 𝑥𝑝0 + 𝑁𝑑 𝑥𝑛0
2𝜖
• But we also have: 𝒙𝒏𝟎 + 𝒙𝒑𝟎 = 𝑾, the width of the DR
and 𝑁𝑎 𝑥𝑝0 = 𝑁𝑑 𝑥𝑛0
• The above 2 equations can be solved for 𝑥𝑛0 and 𝑥𝑝0 to give:
𝑾𝑵𝒂 𝑾𝑵𝒅
𝒙𝒏𝟎 = and 𝒙𝒑𝟎 =
𝑵𝒂 +𝑵𝒅 𝑵𝒂 +𝑵𝒅
• These two can be substituted in the first equation to yield:
𝑞𝑊 2 𝑁𝑑 𝑁𝑎2 + 𝑁𝑎 𝑁𝑑2 𝑞𝑊 2 𝑁𝑎 𝑁𝑑 𝟐𝝐𝑽𝟎 𝑵𝒂 + 𝑵𝒅
𝑉0 = = ⇒𝑾=
2𝜖 𝑁𝑎 + 𝑁𝑑 2 2𝜖 𝑁𝑎 + 𝑁𝑑 𝒒𝑵𝒂 𝑵𝒅
𝟐𝝐𝑽𝟎
• For a p+n junction (𝑁𝑎 ≫ 𝑁𝑑 ) this approximates to 𝑾 = 𝒒𝑵𝒅

ECE102 - VH - L11 19
Key Features
• The DR extends farther into the side with the lighter doping.
For example, if 𝑁𝑎 ≪ 𝑁𝑑 , then 𝑥𝑝0 is larger than 𝑥𝑛0 . This
agrees with the qualitative argument that a deeper penetration
is necessary in lightly doped material to balance the same
amount of space charge as in the heavily doped material where
a shorter penetration is needed
• The DR width W varies as the square root of the potential
across the region
• As we will see later (when we analyze non-equilibrium cases,
such as a forward-biased diode), an applied voltage can
increase or decrease the potential across the DR by aiding or
opposing the equilibrium EField. In other words, an applied
voltage will increase or decrease the width of the DR as well
ECE102 - VH - L11 20
Solved Example 5-1 from Text

ECE102 - VH - L11 21
Solved Example 5-2 from Text

ECE102 - VH - L11 22
END OF LECTURE

ECE102 - VH - L11 23
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L12 0
Agenda
• P-N junction under forward bias

ECE102 - VH - L12 1
 P-N Junction under a Bias Voltage

The term 𝐼0 or 𝐼(𝑔𝑒𝑛) is called the

reverse-saturation current (also
called the generation current (because
the small reverse bias current is due to drift
current due to EHP’s generated thermally in the
depletion region (DR), and also in the QNR
(quasi-neutral regions, read on) up to a diffusion
length from the edge of either DR. Thus it is
highly temperature-dependant)
ECE102 - VH - L12 2
P-N Junction under Bias
• So far we have seen that a PN junction in equilibrium
(ie. 0 bias) has a built-in potential 𝑉0 , that gets set up
due to the built in EField in the depletion region (DR)
to oppose the diffusion of carriers across the 2
regions (holes from P to N and electrons from N to P)
• Now we look at what happens when a bias voltage is
applied across the P-N junction
• Next 2-3 slides will look at it qualitatively
• After that we model it analytically
ECE102 - VH - L12 3
 P-N Junction under Bias
• Bias across a P-N junction affects the DR width
and EField, electrostatic potential, and energy
band diagram, shown alongside for:
a) Equilibrium (ie. no bias)
b) Forward bias
c) Reverse bias
• Notice the quasi-Fermi levels, they separate out
in the biased (ie. non-equilibrium) cases of (b)
and (c)
• The EFields vary linearly with position (as we saw
in the equilibrium case in the last class),
assuming that doping densities in the DRs are
constant on either side of the metallurgical
junction (MJ)
ECE102 - VH - L12 4
 P-N Junction under Bias
• Let us focus on forward bias first. Under
forward bias, the potential barrier height
reduces from the built-in potential value 𝑉0
• The dark-shaded regions (ie. on either side
of the DR) are called quasi-neutral regions
(QNRs)
• In the QNRs close to the DR edges (up to a
few diffusion lengths 𝐿), 𝐽(𝑑𝑖𝑓𝑓) significantly
dominates over 𝐽(𝑑𝑟𝑖𝑓𝑡)
• Under forward bias, the potential barrier
height reduces from the built-in potential
value 𝑉0
ECE102 - VH - L12 5
•
P-N Junction under Bias
As we saw in L05, there is a distribution of energies for
electrons in the N-side conduction band, and some
electrons in the high-energy “tail” of the distribution
have enough energy to diffuse from N to P at
equilibrium in spite of the barrier
• With forward bias, however, the barrier is lowered (to
𝑉0 – 𝑉𝑓 ), and many more electrons in the N-side
conduction band have sufficient energy to diffuse from
N to P over the smaller barrier, resulting in a spurt of
electrons into the p-side of the DR-edge and a spurt of
holes into the n-side of the DR-edge (minority carrier
injection)
• These minority carriers then out-diffuse further into the
QNR (e.g. electrons out-diffuse into p-side QNR) and
recombine with the majority holes. The out-diffusion
results in 𝐽𝑑𝑖𝑓𝑓 . This qualitatively explains the strong
current conduction under forward bias
ECE102 - VH - L12 6
 Modeling Forward Bias
• Recall from the previous lecture that for each carrier
type, the ratio of its density as a majority carrier in a QNR
to that as a minority carrier in a QNR is equal to 𝑒 𝑞𝑉0Τ𝑘𝑇
in equilibrium, where 𝑉0 is the built-in potential
• It can be shown that when a bias 𝑉 is applied across the
diode (neglecting the ohmic drop across the QNR), this
ratio continues to show the same trend albeit depending
on 𝑒 𝑞 𝑉0−𝑉 Τ𝑘𝑇 (called the quasi-equilibrium assumption), with
the additional caveat that now the minority density in the
above relation is to be deemed at the QNR/DR edge.
• Minor detail: The validity of the quasi-equilibrium assumption is greater
when/since:
• When 𝑉 is small enough (as is usually the case: 𝑉 ≈ 0.7𝑉) and
• Since typical dimensions and dopings are such that for each carrier type independently, 𝐽𝑑𝑖𝑓𝑓 and 𝐽𝑑𝑟𝑖𝑓𝑡 are
much larger at equilibrium (albeit equal & opposite and so cancelling out), compared to their values under
bias (where they don’t cancel out). Refer Muller/Kamins book §5.3 for details
ECE102 - VH - L12 7
Modeling Forward Bias
• Applying this relation for say, holes, at the DR edges
– 𝑥𝑝0 (where it is a majority carrier) and 𝑥𝑛0 (where it
is a minority carrier), we have:
𝑝 – 𝑥𝑝0 𝑉0 −𝑉 Τ𝑘𝑇
= 𝑒𝑞
𝑝 𝑥𝑛0
𝑝
• Under equilibrium, this same relation is 𝑝𝑝 = 𝑒 𝑞𝑉0Τ𝑘𝑇
𝑛
where 𝑝𝑝 and 𝑝𝑛 are the hole densities in the P and N
sides in equilibrium (and hence they don’t vary
spatially, hence the spatial dependence is dropped)
• The majority carrier density doesn’t change much with
bias, in other words: 𝑝 – 𝑥𝑝0 ≈ 𝑝𝑝 . Hence, we get:
𝒑 𝒙𝒏𝟎
= 𝒆𝒒𝑽Τ𝒌𝑻
𝒑𝒏
ECE102 - VH - L12 8
Modeling Forward Bias
• A similar relation can be got for electrons too
• In other words, at the DR/QNR edges, the minority
carrier density under bias increases by a factor of 𝑒 𝑞𝑉Τ𝑘𝑇
compared to its equilibrium value at that respective edge
• Hence, excess holes on the DR edge on the N-side is:
Δ𝑝 = 𝑝 𝑥𝑛0 − 𝑝𝑛 = 𝑝𝑛 𝑒 𝑞𝑉Τ𝑘𝑇 − 1
• Similarly, excess electrons on the DR edge on the P-side
is:
Δ𝑛 = 𝑛 −𝑥𝑝0 − 𝑛𝑝 = 𝑛𝑝 𝑒 𝑞𝑉Τ𝑘𝑇 − 1
• This steady excess minority carrier density is always
maintained at the DR/QNR edges as long as a constant
bias 𝑉 is applied. This is called minority carrier injection
ECE102 - VH - L12 9
END OF LECTURE

ECE102 - VH - L12 10
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L13 0
Agenda
• P-N junction under bias
• Forward bias
• Reverse bias

ECE102 - VH - L13 1
 P-N Junction under a Bias Voltage

The term 𝐼0 or 𝐼(𝑔𝑒𝑛) is called the

reverse-saturation current (also
called the generation current (because
the small reverse bias current is due to drift
current due to EHP’s generated thermally in the
depletion region (DR), and also in the QNR
(quasi-neutral regions, read on) up to a diffusion
length from the edge of either DR. Thus it is
highly temperature-dependant)
ECE102 - VH - L13 2
Modeling Forward Bias (contd.)
• At the DR/QNR edges, the minority carrier density under
bias increases by a factor of 𝑒 𝑞𝑉Τ𝑘𝑇 compared to its
equilibrium value at that respective edge
• Hence, excess holes on the DR edge on the N-side is:
Δ𝑝 = 𝑝 𝑥𝑛0 − 𝑝𝑛 = 𝑝𝑛 𝑒 𝑞𝑉Τ𝑘𝑇 − 1
• Similarly, excess electrons on the DR edge on the P-side
is:
Δ𝑛 = 𝑛 −𝑥𝑝0 − 𝑛𝑝 = 𝑛𝑝 𝑒 𝑞𝑉Τ𝑘𝑇 − 1
• This steady excess minority carrier density is always
maintained at the DR/QNR edges as long as a constant
bias 𝑉 is applied. This is called minority carrier injection

ECE102 - VH - L13 3
Modeling Forward
• From our earlier analysis (L10) of
Bias
solving the diffusion equation in a case
where there is a steady density of
excess carriers at a point, we saw that
we get a steady-state spatial
distribution as the carriers out-diffuse
• Here, holes are injected in the N-region
at 𝑥𝑛0 and electrons are injected in the
P-region at −𝑥𝑝0
• Holes (for eg), diffuse into the N-type
QNR and recombine with electrons,
where 𝐽(𝑑𝑟𝑖𝑓𝑡) starts to dominate deep
in QNR far from the DR edges
• Spatial variation of the minority carrier
density is shown in the top figure
ECE102 - VH - L13 4
Modeling Forward Bias
• We now use 2 new coordinate
systems, with axes labelled 𝑥𝑝 and 𝑥𝑛
in the P- and N-QNRs (instead of 𝑥)
• The results of lecture L10 tell us that
excess electron density in the P-QNR
is:
𝛿𝑛 𝑥𝑝 = Δ𝑛𝑝 𝑒 −𝑥𝑝Τ𝐿𝑛 = 𝑛𝑝 𝑒 𝑞𝑉Τ𝑘𝑇 − 1 𝑒 −𝑥𝑝Τ𝐿𝑛

• Excess hole density in the N-QNR is:

𝛿𝑝 𝑥𝑛 = Δ𝑝𝑛 𝑒 −𝑥𝑛Τ𝐿𝑝 = 𝑝𝑛 𝑒 𝑞𝑉Τ𝑘𝑇 − 1 𝑒 −𝑥𝑛Τ𝐿𝑝

• Absorb concepts & make sure you

can interpret each variable (why 𝑥𝑛
and not 𝑥𝑝 , why 𝐿𝑛 and not 𝐿𝑝 , etc)!
ECE102 - VH - L13 5
Modeling Forward
• Recognizing that diffusion current
Bias
𝐽(𝑑𝑖𝑓𝑓) is dominant in the QNR near the
DR-edges, we can calculate the
currents from the spatial gradient (ie.
slope) of the excess minority carrier
distributions at the two respective DR
edges (where the respective carrier is a
minority carrier), viz.:
• 𝐽𝑝(𝑑𝑖𝑓𝑓) hole diffusion current injected into the N-side at
𝑥𝑛0 (ie. 𝑥𝑛 = 0 in the new coordinate system)
• 𝐽𝑛(𝑑𝑖𝑓𝑓) electron diffusion current injected into the P-
side at −𝑥𝑝0 (ie. 𝑥𝑝 = 0 in the new coordinate system)

• Adding the two currents gives the total

current under the assumption that no
recombination takes place within the
DR (called the Shockley ideal diode
approximation)
ECE102 - VH - L13 6
Modeling Forward Bias
• Following this recipe, we get:

𝛿𝑝 𝑥𝑛 = Δ𝑝𝑛 𝑒 −𝑥𝑛Τ𝐿𝑝 = 𝑝𝑛 𝑒 𝑞𝑉Τ𝑘𝑇 − 1 𝑒 −𝑥𝑛Τ𝐿𝑝

𝑑𝑝 𝑥 𝑑 𝑝𝑛 + 𝛿𝑝 𝑥𝑛 𝑑 𝛿𝑝 𝑥𝑛
𝐽𝑝 𝑑𝑖𝑓𝑓 𝑥 = −𝑞𝐷𝑝 = −𝑞𝐷𝑝 = −𝑞𝐷𝑝
𝑑𝑥 𝑑𝑥 𝑑𝑥
𝑞𝐷𝑝 𝑝𝑛 𝑞𝑉Τ𝑘𝑇
∴ 𝐽𝑝 𝑑𝑖𝑓𝑓 𝑥 = 𝑒 − 1 𝑒 −𝑥𝑛Τ𝐿𝑝
𝐿𝑝
𝑞𝐷𝑝 𝑝𝑛
∴ 𝐽𝑝 𝑑𝑖𝑓𝑓 𝑥𝑛 = 0 = 𝑒 𝑞𝑉Τ𝑘𝑇 − 1 = Minority hole (ie. on the n-side) diffusion current density at DR edge
𝐿𝑝

• Similarly:
𝑞𝐷𝑛 𝑛𝑝
𝐽𝑛 𝑑𝑖𝑓𝑓 𝑥𝑝 = 0 = 𝐿𝑛
𝑒 𝑞𝑉Τ𝑘𝑇 − 1 = Minority electron (ie. on the p-side) diffusion current density at DR edge

ECE102 - VH - L13 7
Modeling Forward Bias
• Thus the total current is:
𝐷𝑝 𝐷𝑛
𝐼 = 𝐴 ∙ 𝐽𝑝 𝑑𝑖𝑓𝑓 𝑥𝑛 = 0 + 𝐽𝑛 𝑑𝑖𝑓𝑓 𝑥𝑝 = 0 = 𝑞𝐴 𝑝𝑛 + 𝑛𝑝 𝑒 𝑞𝑉Τ𝑘𝑇 − 1
𝐿𝑝 𝐿𝑛

• This can be simply

written as:
𝑰 = 𝑰𝟎 𝒆𝒒𝑽Τ𝒌𝑻 − 𝟏
𝑫𝒑 𝑫𝒏
where 𝑰𝟎 = 𝒒𝑨 𝒑𝒏 + 𝒏𝒑
𝑳𝒑 𝑳𝒏
is the reverse-saturation current

• The above is the

diode equation, and
holds true for
forward as well as
reverse bias
ECE102 - VH - L13 8
Modeling Forward Bias 𝑄𝑝 𝑄𝑛
• It can be easily shown that an alternate way of writing this is: 𝐼 = +
𝜏𝑝 𝜏𝑛

where 𝑄𝑝 and 𝑄𝑛 is the excess hole minority charge (ie. in the n-side QNR) and the
excess electron minority charge (ie. in the p-side QNR)
• The proof of this can be
seen in the far-right inset
figure also (Fig. (b))
• This can be interpreted as
the diffusion current being
due to the stored minority
carrier charge being
injected every 𝜏 seconds to
replenish them as and
when they vanish due to
recombination
ECE102 - VH - L13 9
 Forward Bias: Closing Remarks
• Since the root cause physical effects takes place across the DR,
we focus our analysis to the DR and DR edges, in calculating all
metrics
• Diffusion currents are dominant at the DR edges
• Drift current dominate far from the junctions (which replenish
carriers as and when the out-diffusing carriers recombine with
the opposite carrier type in the QNR)
• IMP: The relation for width 𝑊 of the DR with bias 𝑉 is:
𝟐𝝐 𝑽𝟎 −𝑽 𝑵𝒂 +𝑵𝒅
𝑾= (same as in equilibrium except replace 𝑉0 by 𝑉0 − 𝑉)
𝒒𝑵𝒂 𝑵𝒅
• Thus, it narrows in forward bias and widens in reverse bias!
ECE102 - VH - L13 10
 Reverse Bias
• The same equation for current as forward bias holds
true in reverse bias also
• Minority-carrier-extraction: In reverse bias, the
minority carriers are sucked-out at the DR-edges (quite
contrary to minority-carrier-injection in forward bias).
The excess minority carrier goes negative, so that total
minority carrier density is 0
• For instance, Δ𝑛𝑝 = 𝑛𝑝 𝑒 𝑞𝑉Τ𝑘𝑇 − 1 ≈ −𝑛𝑝 since 𝑉 ≪
0, even as little as 5 times the thermal voltage 𝑘𝑇Τ𝑞
ECE102 - VH - L13 11
 Reverse Bias
• Band-diagram is shown here. As
can be seen, the barrier height
increases in reverse bias, and
hence greatly reducing carriers
diffusing across the junction
• As can be seen in the excess
minority carrier spatial profile, it
goes negative, so that total
minority carrier is 0 at the DR
edge (= 𝑛𝑝 + Δ𝑛𝑝 = 𝑛𝑝 − 𝑛𝑝 =
0)
ECE102 - VH - L13 12
Solved Example

ECE102 - VH - L13 13
 Solution
• This solved example is very
important, it illustrates the point
that the terms 𝐷/𝐿 in the reverse
saturation current 𝐼0 need to be
calculated for minority carriers (ie.
for 𝐷𝑝 /𝐿𝑝 , calculate it on the n-
side. This value differs on the n-
and p-sides, because the same
carrier-type has different
mobilities on the n- and p-sides!)
ECE102 - VH - L13 14
END OF LECTURE

ECE102 - VH - L13 15
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L14 0
Agenda
• P-N junction: Turn-off transient

ECE102 - VH - L14 1
 Transient Behavior
• As we have seen
in L12/13, there is
stored charge in
the QNR due to
minority carrier
injection
• When switching a
diode from ON to
OFF (ie. forward
to reverse bias), it
takes some time
to empty out this
charge, this limits
the switching
speed
ECE102 - VH - L14 2
 Transient Behavior
• To analyze the transient behavior, the time variation of the continuity
equation needs to be solved for (lecture L10)
• Integrating the continuity equation for a carrier throughout the QNR side
where that carrier is a minority carrier, and noting that the minority carrier
current far from the DR-edge is 0 while close to the DR-edge is the complete
current contributed by that carrier-type (viz. 𝐽𝑑𝑖𝑓𝑓 , since 𝐽𝑑𝑟𝑖𝑓𝑡 ≈ 0 at DR-
edge), it can be shown that:
𝑄𝑐 𝑑𝑄𝑐 (𝑡)
𝑖𝑐 𝑡 = + where the suffix 𝒄 is 𝒏 and 𝒑 in turn
𝜏𝑐 𝑑𝑡
• This is the Charge Control Model (CCM):
𝒊 𝒕 = 𝒊𝒑 𝒕 + 𝒊𝒏 (𝒕)
𝑸𝒑 𝒅𝑸𝒑 (𝒕)
𝒊𝒑 𝒕 = + where 𝑄𝑝 and 𝜏𝑝 are on the n-side QNR
𝝉𝒑 𝒅𝒕
𝑸𝒏 𝒅𝑸𝒏 (𝒕)
𝒊𝒏 𝒕 = + where 𝑄𝑛 and 𝜏𝑛 are on the p-side QNR
𝝉𝒏 𝒅𝒕
ECE102 - VH - L14 3
 Transient Behavior
• For example, for holes these equations can be interpreted
as the hole current injected across the junction being
determined by two charge storage effects in the n-QNR:
𝑸𝒑
1. The usual recombination term in which the excess carrier
𝝉𝒑
distribution is replaced every 𝝉𝒑 seconds, and
𝒅𝑸𝒑 (𝒕)
2. A charge buildup (or reduction) term , which allows for the
𝒅𝒕
fact that the distribution of excess carriers can be increasing or
decreasing as a time-dependent problem
• Note: In steady-state (as we studied in L12/L13), we had
omitted the time derivative term, as is to be expected
ECE102 - VH - L14 4
 Current-forced Turn-off Transient
• Consider a p+-n diode that is forward biased
using a current source 𝐼. As discussed, there
will be a forward voltage drop 𝑉 across the
diode, that is suitably built to support that
current using the diode I-V relation
• Suppose the current-source is suddenly
switched off at 𝑡 = 0. Contrary to a naïve
expectation, the voltage 𝑉 won’t die off
instantly. It takes some time, this is called the
step turn-off transient
ECE102 - VH - L14 5
 Current-forced Turn-off Transient
• A p+-n diode is a junction where the p-side is much heavier doped than
the n-side. Hence, it is a good approximation to consider only the excess
holes injected in the n-QNR, for total current & charge computation
• Since the excess holes in the n-QNR must die out by recombination with
the matching excess electron population, some time is required for
𝑄𝑝 𝑡 to reach zero
• We can solve it easily by solving
the CCM for holes, at 𝑡 > 0:
𝑄𝑝 𝑑𝑄𝑝 (𝑡)
+ = 0 where 𝑄𝑝 0 = 𝐼𝜏𝑝
𝜏𝑝 𝑑𝑡

• The solution is:

𝑄𝑝 𝑡 = 𝐼𝜏𝑝 𝑒 −𝑡Τ𝜏𝑝
ECE102 - VH - L14 6
 Current-forced Turn-off Transient
• An important implication of this is that even though the
current is suddenly stopped, the voltage across the
junction has to persist until 𝑄𝑝 disappears, in order to
support that excess charge that is responsible for 𝑄𝑝
• The excess hole concentration is related to junction
voltage by the below equation (refer L12.Slide9)
Δ𝑝𝑛 (𝑡) = 𝑝 𝑥𝑛0 , 𝑡 − 𝑝𝑛 = 𝑝𝑛 𝑒 𝑞𝑣 𝑡 Τ𝑘𝑇 − 1
ECE102 - VH - L14 7
 Current-forced Turn-off Transient
• We further make a quasi-steady-state approximation
where we assume 𝛿𝑝 𝑥 as an exponential even though it
isn’t exactly (because of 𝑆𝑙𝑜𝑝𝑒 = 0 due to 𝑖(𝑡 = 0+ ) =
0):
𝛿𝑝 𝑥𝑛 , 𝑡 = Δ𝑝𝑛 𝑡 𝑒 −𝑥𝑛 Τ𝐿𝑝 in the p-QNR
• Integrating the above along the QNR, we can find
𝑄𝑝 (𝑣 𝑡 )
• Then, we can equate this with the 𝑄𝑝 expression on slide
6 to solve for 𝑣 𝑡
ECE102 - VH - L14 8
 Current-forced Turn-off Transient
• We get:
𝒌𝑻 𝑰𝝉𝒑
𝒗 𝒕 = 𝐥𝐧 𝒆−𝒕Τ𝝉𝒑 + 𝟏
𝒒 𝒒𝑨𝑳𝒑 𝒑𝒏
• A plot of this will show that the voltage can be deemed
pinned at a constant value of the forward voltage drop
(approximately 0.7 𝑉) for about 2-3 minority carrier
lifetimes (about 2𝜏𝑝 to 3𝜏𝑝 ), and thereafter slowly dies
down to 0
• This shows that the voltage across a p-n junction cannot be
changed instantaneously when it is switched off by a current
source, because that stored charge takes time to vanish
ECE102 - VH - L14 9
Current-forced Turn-off Transient
GNUPlot code to generate this plot:
reset session

vt=0.026
I=1.06e-3
tau_p=10e-6
q=1.6e-19
A_cm2=1e-4
Lp_cm=1.08e-2
p_n_per_cc=2.25e5

v(x)=vt*log(1+(I*tau_p/(q*A_cm2*Lp_cm*p_n_per_cc))*exp(-x/tau_p))

set xlabel "Time (us)"

set ylabel "Voltage (V)"
set xrange [0:400]
set nokey
set grid

set label "Using the same parameters for holes \nin n-side as in
the solved example in L13" at 200,0.65

plot v(x*1e-6) with lines

ECE102 - VH - L14 10
 Voltage-forced Turn-off Transient
• Consider a p+-n diode that is forward biased using a voltage source 𝐸 (through a
resistor, as usual) such that 𝐸 ≫ forward voltage drop of a diode (approximately
0.7𝑉). There will be a forward current 𝐼𝑓 through the diode, consistent with the
diode I-V relation, albeit 𝐼𝑓 will be predominantly governed by the resistance 𝑅,
viz.: 𝐼𝑓 ≈ 𝐸 − 0.7 Τ𝑅 ≈ 𝐸 Τ𝑅
• Suppose the voltage-source is suddenly
reversed at 𝑡 = 0 to −𝐸, ie. reverse bias.
Contrary to a naïve expectation, the
current through the diode won’t die off
instantly. It takes some time, this is called
the reverse recovery transient
• The associated delay is called the storage
delay time 𝑡𝑠𝑑 , that we analyze here
ECE102 - VH - L14 11
 Voltage-forced Turn-off Transient
• The underlying cause for this phenomenon is
similar, viz. it takes time to flush out the
stored charge due to the excess minority
carriers (in this case, predominantly excess
holes in the n-QNR)
• The spatial profile of the
excess holes in the n-QNR is
shown alongside, at various
snapshots of time
• We next discuss this
behavior in detail
ECE102 - VH - L14 12
 Voltage-forced Turn-off Transient
• When 𝐸 is reversed (𝑡 = 0+ ), the voltage across the diode cannot
change instantly, due to the charge storage arising from excess
minority carriers. Hence by KVL, the −𝐸 voltage falls almost
entirely across the resistor 𝑅. Hence the current initially reverses
to 𝑖 𝑡 = 𝐼𝑟 ≈ −𝐸/𝑅. This reverse current slowly reduces/sucks-
out the excess stored charge
• As the stored minority charge reduces in the QNR especially close
to the DR-edge, we can find the junction voltage again from the
voltage-dependence relation for excess minority carriers Δ𝑝𝑛
reiterated on slide 7. Nevertheless, as long as Δ𝑝𝑛 > 0, the
junction voltage 𝑣 𝑡 > 0 and small (≈ 0.7𝑉) and getting even
smaller. Therefore 𝑖 𝑡 continues to be 𝑖 𝑡 = 𝐼𝑟 ≈ −𝐸/𝑅, until
Δ𝑝𝑛 goes to 0. Thereafter, when Δ𝑝𝑛 < 0, the junction exhibits a
negative voltage (consistent with reverse bias)
ECE102 - VH - L14 13
•
Voltage-forced Turn-off Transient
Since the reverse-bias voltage of a junction
can be very large, KVL dictates that 𝐸 now
divides between 𝑅 and the junction, thereby
reducing the magnitude of the reverse
current. As time proceeds, the magnitude of
the reverse current becomes smaller as more
of −𝐸 appears across the reverse biased
junction, until finally the only current is the
small reverse saturation current 𝐼0
• The time 𝒕𝒔𝒅 required for the stored charge
(and therefore the junction voltage) to
become zero is called the storage delay time
ECE102 - VH - L14 14
Voltage-forced Turn-off Transient
• As before, we can solve this using the CCM for holes, at 𝑡 > 0:
𝑄𝑝 𝑑𝑄𝑝 (𝑡)
𝜏𝑝
+ 𝑑𝑡 = −𝐼𝑟 where 𝑄𝑝 0 = 𝐼𝑓 𝜏𝑝
𝑑𝑄𝑝 (𝑡) 𝑄𝑝 𝑑𝑄𝑝 (𝑡)
∴ = − 𝐼𝑟 + ⇒ = −𝑑𝑡
𝑑𝑡 𝜏𝑝 𝑄𝑝
𝐼𝑟 + 𝜏
𝑝
𝑄𝑝 𝑡
𝑄𝑝
𝐼𝑟 +
𝑑𝑄𝑝 (𝑡) 𝜏𝑝
∴ න = − න 𝑑𝑡 ⇒ 𝜏𝑝 ln = −𝑡
𝑄𝑝 𝐼𝑟 + 𝐼𝑓
𝐼𝑓 𝜏𝑝 𝐼𝑟 + 0
𝜏𝑝
𝑄𝑝
∴ 𝐼𝑟 + = 𝐼𝑟 + 𝐼𝑓 𝑒 −𝑡Τ𝜏𝑝 ⇒ 𝑄𝑝 𝑡 = 𝜏𝑝 𝐼𝑟 + 𝐼𝑓 𝑒 −𝑡Τ𝜏𝑝 − 𝐼𝑟
𝜏𝑝
−𝑡𝑠𝑑 𝐼𝑟 𝑡𝑠𝑑 𝐼𝑟 +𝐼𝑓
∴ 𝑄𝑝 𝑡𝑠𝑑 = 0 ⇒ 𝐼𝑟 + 𝐼𝑓 𝑒 −𝑡𝑠𝑑Τ𝜏𝑝 − 𝐼𝑟 = 0 ⇒ = ln ⇒ = ln
𝜏𝑝 𝐼𝑟 +𝐼𝑓 𝜏𝑝 𝐼𝑟
𝑰𝒇
∴ 𝒕𝒔𝒅 = 𝝉𝒑 𝐥𝐧 𝟏 +
𝑰𝒓
ECE102 - VH - L14 15
 Voltage-forced Turn-off Transient

Effects of storage delay time on switching

signal: (a) switching voltage; (b) diode current
ECE102 - VH - L14 16
Summary of Both Transient Analyses
• Similarities between current-forced & voltage-forced
transients:
• In both cases there are excess minority carriers to be removed
when switching from ON to OFF
• Thus, it is like capacitive action (called diffusion capacitance, as
we’ll discuss later), where we know as a thumb-rule that voltage
across a capacitor cannot change suddenly (because changing
charge takes time). Hence the voltage takes some time to settle to
its final value
• In both cases, the peak voltage across the junction is of the order
of the forward diode drop, which is of the order of 0.2V − 0.7𝑉
depending on the kind of diode
ECE102 - VH - L14 17
Summary of Both Transient Analyses
• Differences between the analysis-approach of the
current-forced & voltage-forced transient problem:
• In the current-forced case, the total current term in the CCM
is made 0, consistent with the situation’s demand
• In the voltage-forced case, the situation doesn’t demand the
current to be 0. But to be consistent with potential-
dependent carrier statistics (slide 7), a voltage 𝑣 𝑡 has to be
present to support the excess carrier profile (as was the case
in the current-forced case too). Since this voltage is small
compared to the external bias 𝐸, the current is largely
determined by the resistor and is 𝐼𝑟 ≈ −𝐸/𝑅, which is set as
the total current term in the CCM
ECE102 - VH - L14 18
 Optimizing Turn-off Transients
• Some common ways to mitigate such problems of
stored charge when making fast switching diodes:
1. Design the diode with the lighter doped side (e.g. n-
QNR if it’s a p+-n diode) being very short in length. If
shorter than a minority carrier diffusion length (e.g. 𝐿𝑝
in this case), very little charge is stored. This is called a
short-base diode
2. Intentionally dope with recombination centers (mid-gap
trap levels), such as gold (Au) atoms in Si, to increase
the recombination rate in the QNR (though this
degrades 𝐼0 , so it’s a trade-off)
ECE102 - VH - L14 19
END OF LECTURE

ECE102 - VH - L14 20
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L15 0
Agenda
• P-N junction: Capacitances
• P-N junction: Reverse Breakdown

ECE102 - VH - L15 1
 Capacitances
• Capacitive effects are due to charge storage
• Depletion Capacitance
• Due to charge stored in the depletion region (DR)
• More dominant in RB, since DR has greater charge in RB
• Diffusion Capacitance
• Due to charge stored in excess minority carrier profile in
the quasi-neutral regions (QNRs)
• Relevant in FB
• Negligible in long-base diodes (length of QNR longer than a
few diffusion lengths), so we skip this in this UG-level
course where our focus is mainly long-base diodes
ECE102 - VH - L15 2
 Depletion Capacitance 𝐶𝑗
𝑑𝑄𝑑
• Fundamental definition: 𝐶𝑗 ≜ where:
𝑑𝑉𝑗
• 𝑄𝑑 is the charge in the DR
• 𝑉𝑗 is the voltage across the DR, thus 𝑉𝑗 = 𝑉0 − 𝑉 where 𝑉0 is the
built-in potential and 𝑉 is the externally applied bias across the
diode terminals (Note: 𝑽 < 𝟎 in RB)
• Minor note: Such a definition (as opposed to 𝐶 = 𝑄/𝑉) is
called a differential capacitance
• The choice of whether to use 𝐶 = 𝑄/𝑉 or 𝐶 = 𝑑𝑄/𝑑𝑉 depends
on the intended use. If the intended use is to develop a small
signal model (as is the case here), then 𝑑𝑄/𝑑𝑉 is more apt
ECE102 - VH - L15 3
 Depletion Capacitance 𝐶𝑗
• Proceeding thus from first principles, it
can easily be shown that for an abrupt
junction:
𝟏/𝟐
𝝐𝑨 𝒒 𝑵 𝒂 𝑵𝒅
𝑪𝒋 = = 𝝐𝑨
𝑾 𝟐𝝐 𝑽𝟎 − 𝑽 𝑵𝒂 + 𝑵𝒅
• For a p+n abrupt junction, this
simplifies to:
𝟏/𝟐
𝑨 𝟐𝒒𝝐𝑵𝒅
𝑪𝒋 =
𝟐 𝑽𝟎 − 𝑽
• The figure alongside shows the typical
variation of depletion capacitance
with bias
• Voltage-dependent capacitance (called
varactor diode: variable capacitor diode)
ECE102 - VH - L15 4
 Reverse Breakdown
• Forward bias (FB) characteristics: 𝑉 > 0
• Reverse bias (RB) characteristics: 𝑉 < 0
• We have already seen reverse saturation
current 𝐼0 in RB
• At sufficiently high RB 𝑉𝑏𝑟 , reverse
breakdown occurs:
• At this critical voltage the
reverse current through
the diode increases
sharply, and relatively large
currents can flow with little
further increase in voltage
ECE102 - VH - L15 5
 Reverse Breakdown
• Just as in the FB case, in the RB case too the
resistance 𝑅 is chosen to limit the reverse
current to a safe level, to prevent damage due
to overheating if the device’s maximum power
rating is exceeded
• The exact mechanics of reverse breakdown are:
• Zener breakdown
• Avalanche breakdown
ECE102 - VH - L15 6
 RB: Zener Effect
• Exact mechanics (field-
ionization):
• In sufficiently high RB (viz. 𝑉𝑏𝑟 ),
bands bend a lot so as to align
VB levels alongside CB levels
• This, in conjunction with small
depletion width 𝑊 (due to high doping), results in quantum mechanical
tunnelling (BTBT: band-to-band-tunnelling) of electrons from VB on p-side to
CB on n-side, which gives rise to large reverse currents
• Device engineering characteristics for this to happen:
• Junction should be highly doped (both sides)
• Doping profile switchover from p- to n-side is abrupt (ie. not gradual)
ECE102 - VH - L15 7
•
RB: Avalanche Breakdown
Exact mechanics (impact-ionization):
• At sufficiently high built-in electric field 𝐸 in the DR, an electron
somehow entering from the p-side gains high enough KE (kinetic
energy) to cause an ionizing collision with the lattice, resulting
in an EHP creation. The original electron and the generated electron are
both swept to the n-side of the junction, and the generated hole is
swept to the p-side. These electrons & holes can in turn gain enough KE
to continue to encounter ionizing collisions with the lattice, resulting in
secondary, tertiary, etc EHP generation; and the process continues
• Hence the large reverse currents
• Device engineering characteristics for
this to happen:
• Junction should be lightly doped (atleast one
side)
• Doping profile switchover from p- to n-side
should be gradual
ECE102 - VH - L15 8
 Reverse Breakdown
• In general, the critical reverse
voltage 𝑉𝑏𝑟 for breakdown
increases with 𝐸𝑔 , since more
energy is required for an ionizing
collision (whether field- or
impact-)
• Peak electric field 𝐸0 within DR
increases with increased doping
on the more lightly doped side of
the junction. Therefore, 𝑉𝑏𝑟
decreases as the doping increases
ECE102 - VH - L15 9
 Reverse Breakdown
• The more common breakdown mechanism is avalanche.
By precisely varying the doping we can fabricate diodes
with specific breakdown voltages ranging from less than
1 𝑉 to several hundred volts. Diodes engineered
precisely for a specific 𝑉𝑏𝑟 are called breakdown diodes
• Diodes which breakdown due to Zener effect have a
lower 𝑉𝑏𝑟 than those which breakdown due to
avalanche effect (though in the market, all such diodes
are called Zener diodes, irrespective of their breakdown
mechanism ☺)
ECE102 - VH - L15 10
END OF LECTURE

ECE102 - VH - L15 11
ECE102: Semiconductor
Devices
Dr. Venkatnarayan Hariharan
Dept of Electrical Engg., Shiv Nadar Univ., Delhi NCR

ECE102 - VH - L16 0
Agenda
• Analysing circuits involving P-N
junction diodes
• Device and circuit applications of
diodes

Source: “Electronic Principles” book by Malvino & Bates

ECE102 - VH - L16 1
 Diode Symbol

• (a) Schematic symbol

• (b) Diode case styles
• (c) Forward bias
ECE102 - VH - L16 2
Schematic Symbols for Other Variants

• All of the above are P-N junctions, each differently

engineered (with suitable material or sequence of materials,
suitable dopings, etc for that particular purpose)
• The left side of the symbol in the figures above (ie. triangle-
base) is the anode and the triangle-tip is the cathode
ECE102 - VH - L16 3
 Diode I-V Characteristics
• For circuit
analysis,
deem as FB
only if
voltage is
greater than
the knee-
voltage
ECE102 - VH - L16 4
 Quiz: Check whether FB or RB
• Assume what
intuitively
makes sense
and proceed
with analysis
• If contradiction
inferred, then
alter the
assumption
ECE102 - VH - L16 5
Trick for DC hand-analysis
• Once you determine it is FB, assume that the
voltage across the diode is 𝑉 = 0.7 𝑉

ECE102 - VH - L16 6
Trick for DC hand-analysis
• Use usual circuit analysis methods, such as
converting to Thevenin equivalent, etc

ECE102 - VH - L16 7
•
Device Applications
Varactor diodes:
−1/2
• As seen, abrupt junctions have 𝐶𝑗 ∝ 𝑉𝑟 when 𝑉𝑟 = 𝑉 ≫ 𝑉0 . With good
−𝑛
junction engineering, 𝐶𝑗 ∝ 𝑉𝑟 can be obtained (𝑛 > 0), with a desirable
goal to have 𝑛 as high as possible for maximum range of 𝐶𝑗 with 𝑉𝑟 (Why?)
• Abrupt junctions: 𝑛 = 1/2 (possible if epitaxially grown)
• Linearly graded junctions: 𝑛 = 1/3 (likely by a sequence of epitaxial layers)
• Hyperabrupt junctions: 𝑛 = 2 is possible (possible by ion implantation)
• Used in the tuning stage of a radio receiver (replacing bulky variable plate
capacitors), harmonic generation, microwave frequency multiplication,
active filters, etc
• Zener diodes:
• Because of steep I-V characteristics
in RB at 𝑉𝑏𝑟 , Zener diodes wired up
in RB can be used to build voltage
regulators, as shown alongside
ECE102 - VH - L16 8
 Device Applications
• LEDs (Light Emitting Diodes). Mechanics:
• When carriers are injected across a FB junction, the current (arising
out of diffusion of injected minority carriers) is usually accounted for
by recombination in the depletion region (a non-ideality) and in the
QNRs near the junction. In a semiconductor with an indirect band
gap, such as Si or Ge, the recombination releases heat to the lattice,
not light. But in a direct bandgap material like GaAs, radiative
recombination occurs and considerable light is given off from the
junction under forward bias.
• The frequency (color) of the photon is governed by the bandgap 𝐸𝑔 of the
semiconductor as given by the Planck relation, 𝒉𝝂 = 𝑬𝒈 , which, in
appropriate units, can be expressed as 𝑬𝒈 (𝒆𝑽) = 𝟏. 𝟐𝟒/𝝀 (𝝁𝒎)
ECE102 - VH - L16 9
 Device Applications
• LEDs (very energy efficient):
• The most suited materials for LEDs are compound semiconductors, like
𝐺𝑎𝐴𝑠1−𝑥 𝑃𝑥 where 𝑥 is the mole fraction of P (phosphorus). As 𝑥 is varied, it
turns out that the bandgap 𝐸𝑔 varies, and it remains a direct bandgap
material till about 𝑥 = 0.45. Indeed, 𝑥 = 0.4 is used for red LEDs
• Red, green, yellow, etc LEDs were discovered long back. But the material
engineering (ie. fab & doping) for blue LEDs was discovered recently. RGB
being primary colors, this opened up the door to white light generation,
and therefrom high-brightness (HB) white LEDs, used in homes, cars, etc.
LEDs are also used in TV panel backlighting, etc
• Infrared LEDs are used in fiber-optic communications, TV remote
controls, optoelectronic isolators (comprising a LED and
photodiode pair in a package), etc
ECE102 - VH - L16 10
Circuit Applns: Full-wave Rectifier

ECE102 - VH - L16 11
 Bridge Rectifier
• (a) Bridge
rectifier
circuit
• (b) Equivalent
circuit for
positive half-
cycle
• (c) Equivalent
circuit for
negative half-
cycle
• (d ) Full-wave
output
ECE102 - VH - L16 12
Bridge Rectifier & Capacitor Filter

ECE102 - VH - L16 13
Positive Clipper

ECE102 - VH - L16 14
 Positive Clamper
• (a) Ideal positive
clamper
• (b) At the
positive peak
• (c) Beyond the
positive peak
• (d ) Realistic
Clamper (ie. non-
ideal)
ECE102 - VH - L16 15
Voltage Multipliers

• Voltage multipliers
• (a) Doubler
• (b) Tripler
• (c) Quadrupler

ECE102 - VH - L16 16
END OF LECTURE

ECE102 - VH - L16 17