Optical Sensors and Actuators: The Eye

Chapter 4
Optical sensors and actuators
The eye
The human eye, like that of other vertebrates, is a marvelous, complex sensor
allowing us to perceive the world around us in minute detail and true colors.
In fact, the eye is akin to a video camera. It consists of a system of lenses (the
cornea and crystalline lens), an aperture (iris and pupil), an image plane
(retina), and a lens cover (eye lids). In humans and animals of prey, the eyes
point forward to create binocular vision with excellent depth perception.
Many prey animals have side-facing eyes to increase their field of view, but
the vision is monocular and lacks perception of depth. The eyelids, in addi-
tion to protecting the eye, also keep it clean and moist by distributing tears as
well as lubricants (the conjunctiva) and protect it from dust and foreign
objects in conjunction with the eyelashes. The front dome of the eye is made
of the cornea, a clear, fixed lens. This is a unique organ, as it has no blood
vessels and is nourished by tears and the fluid inside the eye sphere. Behind it
is the iris, which controls the amount of light that enters the eye. On the
periphery of the iris, there is a series of slits that allow fluid to pass out from
the eye sphere. This passes nutrients to the front of the eye and relieves the
pressure in the eye (when this is not perfectly regulated one has glaucoma, a
condition that can affect the retina and eventually can cause blindness).
Behind is the crystalline lens, an adjustable lens, controlled by the ciliary
muscle that allows the eye to focus on objects as close as about 10 cm and as
far as infinity. When the ciliary muscle loses some function, the ability of the
lens to focus is impaired, leading to the need for corrective action (glasses or
surgery). The lens itself can cloud over time (cataracts), a condition that
requires replacement of the lens. At the back of the eye lies the optical sensor
proper—the retina. It is made of two types of cells: cone cells that perceive
color and rod or cylindrical cells that are responsible for low-light (night)
vision. The cone cells are divided into three types, sensitive to red, green, and
blue light, with a total of about 6 million cells, most of them in the center of
the retina (the macula). Rod cells are distributed mostly on the peripheral
parts of the retina and are responsible for low-light vision. They do not
148 Sensors, actuators, and their interfaces
perceive color but are as much as 500 times more sensitive than cone cells.
There are also many more rod cells than cone cells—as many as 120 million
of them. The retina is connected to the visual cortex in the brain through the
optical nerve. Although the lens of the eye is adjustable, the size of the
optical ball also plays a role in vision. Individuals with larger eyeballs are
nearsighted, those with smaller eyeballs are farsighted.
The sensitivity of the human eye ranges from roughly 106 cd/m2 (dark
night, rod-dominated vision, essentially monochromatic) to about 106 cd/m2
(bright sunlight, cone-dominated vision, full color). This is a vast dynamic
range (120 dB). The spectral sensitivity of the eye is divided into four par-
tially overlapping zones. Blue cones are sensitive between about 370 nm and
530 nm, with peak sensitivity at 437 nm; green cones between 450 nm and
640 nm, with peak sensitivity at 533 nm; and red cones between 480 nm
and 700 nm, with peak sensitivity at 564 nm. Rods are sensitive between
about 400 nm and 650 nm, with peak at 498 nm. This peak is in the blue–
green range. For this reason, low-light vision tends to be dark green.
It should also be noted that the human-type eye, a structure shared by
many animals, is not the only type of eye. There are some ten different
structures ranging from simple light-sensitive cells that allow the organism to
detect light but not to create images, to compound eyes made of thousands of
simple, individual “eyes” particularly suited to detect motion but can only
create “pixilated” images.
4.1 Introduction
Optics is the science of light and light is an electromagnetic radiation that manifests
itself either as an electromagnetic wave or as photons (particles with quanta of
energy). Before continuing, it is worth mentioning that the term light refers spe-
cifically to the visible spectrum of electromagnetic radiation as perceived by the
human eye (see Figure 4.1), but because both below and above this spectrum the
behavior of radiation is similar, the term light is normally extended to include a
Visible light
Microwaves Infrared Ultraviolet X-rays
700 nm (red) 400 nm (violet)

1m 1 mm (430 THz) (750 THz) 1 nm 1 pm Wavelength
(300 Hz) (300 GHz) (300 PHz) (300 EHz) (frequency)
Figure 4.1 Spectrum of infrared, visible, and ultraviolet radiation

Optical sensors and actuators 149
much wider spectrum that includes infrared (IR) radiation (below the frequency of
visible light or “below red”) and ultraviolet (UV) radiation (above the visible range
or “above violet”). Even the nomenclature has been modified and we sometimes
say IR light or UV light. These terms are incorrect but are in widespread use. The
range that is properly called light is defined by the response of the human eye
between 430 THz and 750 THz (1 THz ¼ 1012 Hz). In characterizing light, it is
more common to use wavelength, defined as the distance in meters the light wave
propagates in one cycle or l ¼ c/f, where c is the speed of light and f is its fre-
quency. The range of wavelengths in the visible light region is between 700 nm
(deep red) and 400 nm (violet). However, the ranges of IR and UV radiation are not
as well defined and, as can be seen in Figure 4.1, the lower range of IR radiation
overlaps the higher range of microwave radiation (sometimes, this upper range is
called millimeter wave radiation), whereas the upper reaches of UV radiation reach
into the X-ray spectrum. For the purpose of this discussion, the IR range is between
1 mm and 700 nm and the UV range is between 400 nm and 1 nm. What unifies this
wide range for the purpose of this chapter is the fact that the principles of sensing
are similar and based on essentially the same effects. It should also be pointed out
that the term radiation here means electromagnetic radiation, distinct from nuclear
or radioactive radiation.
Optical sensors are those sensors that detect electromagnetic radiation in what
is generally understood as the broad optical range—from far IR to UV. The sensing
methods may rely on direct methods of transduction from light to electrical quan-
tities such as in photovoltaic (PV) or photoconducting sensors or indirect methods
such as conversion first into temperature variation and then into electrical quan-
tities such as in passive IR (PIR) sensors and bolometers.
There is a third method of sensing related to optics—sensors based on light
propagation and its effects (reflection, transmission and refraction), which will not
be discussed here because the optical aspect is usually not the sensing mechanism,
but rather an intermediate transduction mechanism. Nevertheless, the physics will
be mentioned briefly for completeness.
4.2 Optical units

The units used in optics seem to be more obscure than most. Thus, it is useful to
discuss these at this point. First, the SI units provide for a measure of luminous
intensity, the candela (cd) (see Section 1.6.1). The candela is defined as the
luminous intensity, in a given direction, of a source that emits monochromatic
radiation of frequency 540 1012 Hz and that has a radiation intensity of 1/683 W/sr.
In short, the candela is a measure of radiation intensity.
Other units are often used. The lumen (luminous flux) is a candela steradian
(cdsr) and is a measure of power. The lux (a measure of illumination) is a candela
steradian per square meter (cd sr/m2) and is therefore a power density. These are
summarized in Table 4.1.
Table 4.1 Optical quantities and their units
Quantity Name Unit Derived unit Comments

Luminous intensity Candela [cd] [W/sr] Power radiated per
steradian
Luminous flux Lumen [cd sr] [W] Power radiated
Illuminance Lux [cd sr/m2] [W/m2] Power density
Luminance Candela per [cd/m]2 [W/sr m2] Density of luminous
meter square intensity
Example 4.1: Conversion of optical units

A point source emits uniformly in all directions in space (e.g., the sun may be
considered a point source when viewed from Earth). Given a total power radiated
of 100 W, calculate the source’s luminous intensity and illuminance at a distance of
10 m from the source.
Solution: Since there are 4p solid unit angles in a sphere, the luminous intensity is
100
luminous intensity ¼ ¼ 7:958 W=sr:
4p
Since a candela is 1/683 W/sr, the luminous intensity

100
luminous intensity ¼ 683 ¼ 7:958 683 ¼ 5;435:14 cd:
4p
Although the units of illuminance are cd sr/m2, it is best to start from the
power radiated. That power is spread over the sphere of radius R ¼ 10 m, so we get
100 7:958
Illuminance ¼ ¼ ¼ 0:0796 lux:
4pR2 102
It should be noted that whereas the luminous intensity is a fixed value that only
depends on the source, illuminance depends as well on the distance from the source.
Note: A uniformly radiating source is called an isotropic source.
4.3 Materials
The sensors/actuators discussed in Chapter 3 and those that will follow take
advantage of many physical principles. But, in addition, they take advantage of
specific material properties, either of elements, alloys, or in other forms available,
including synthetic and naturally occurring salts, oxides, and others. As we will
discuss some of these, especially in conjunction with semiconducting materials, it
is perhaps useful to bear in mind the periodic table (see the inside back cover).
Many of the properties of materials are not specific to a single element, instead
belong to a group (often, a column in the table of elements), and one can expect that
if an element in a specific column is used for a given purpose, other elements from
the same column may have similar properties and be equally useful. For example, if
potassium (alkali column I) is useful in the production of cathodes for photoelectric
cells, then lithium, sodium, rubidium, and cesium should also be useful. But, there
are clear limits. Hydrogen and francium, which are also in the same column, are not
useful. The first because it is a gas, the second because it is radioactive. Similarly,
if gallium–arsenide (GaAs) makes a useful semiconductor, so should indium–
antimonide (InSb), and so on. We already saw some of these principles in dis-
cussing thermocouples. The elements in the VIII column—nickel, palladium, and
platinum—are used for the various types of thermocouples together with elements
from the IB and IIB columns. We shall refer to the periodic table often, but will
also refer to many simple or complex compounds with specific properties that have
been found to be useful in sensors and actuators. Here, we will be concerned pri-
marily with semiconductors, but other materials will become important in sub-
sequent chapters.
4.4 Effects of optical radiation

4.4.1 Thermal effects
The interaction of light (radiation) with matter results in absorption of energy in
two distinct ways. One is thermal and is usually viewed as absorption of electro-
magnetic energy. The other is a quantum effect. The thermal effect is based on
electromagnetic energy absorbed by the medium and converted into heat through
the increased motion of atoms. This heat is sensed and translated into a measure of
the incident radiation. Here, we will not go beyond the understanding that by
raising the temperature of a material, its electrons gain kinetic energy and may be
released given sufficient energy and, of course, that this interaction can be used for
sensing.
4.4.2 Quantum effects

4.4.2.1 The photoelectric effect
The second effect is a quantum effect and is governed by photons, the particle-like
manifestation of radiation. In this representation of light, and in general radiation,
energy travels in bundles (photons) whose energy is given by Planck’s equation:
e ¼ hf ½eV or J; (4.1)
34 15
where h ¼ 6.6262 10 Js or 4.1357 10 eV s, which is Planck’s constant,
and f is the frequency in hertz. e is the photon energy and is clearly frequency
dependent. The higher the frequency (the shorter the wavelength), the higher the
photon energy. In the quantum mode, energy is imparted to materials by elastic
collision of photons and electrons. The electrons acquire energy and this energy
allows the electron to release itself from the surface of the material by overcoming
the work function of the material. Any excess energy imparts kinetic energy to the
electron. This theory was first postulated by Albert Einstein in his photon theory,
which he used to explain the photoelectric effect in 1905 (and for which he received
the Nobel Prize). This is expressed as
hf ¼ e0 þ k ½eV; (4.2)
where e0 is the work function and is the energy required for an electron to leave the
surface of the material (see Table 4.2). The work function is a given constant for
each material. k ¼ mv2/2 represents the maximum kinetic energy the electron may
have outside the material. That is, the maximum velocity electrons can have outside
pffiffiffiffiffiffiffiffiffiffiffi
the material is v ¼ 2k=m, where m is the mass of the electron.
A photon with energy higher than the work function will, in principle, release
an electron and impart a kinetic energy according to (4.2). But does, in fact, each
photon release an electron? That depends on the quantum efficiency of the process.
Quantum efficiency is the ratio of the number of electrons released (Ne) to number
of photons absorbed (Nph):
Ne
h¼ : (4.3)
Nph
Typical values are around 10%–20%. This simply means that not all photons
release electrons.
Clearly, for electrons to be released, the photon energy must be higher than the
work function of the material. Since this energy depends on frequency alone, the
frequency at which the photon energy equals the work function is called a cutoff
frequency. Below it, quantum effects do not exist (except for tunneling effects) and
only thermal effects are observed. Above it, thermal and quantum effects are pre-
sent. For this reason, low-frequency radiation (IR in particular) can only give rise to
thermal effects, whereas at high frequencies (UV radiation and above) the quantum
effect dominates.
Table 4.2 Work functions for selected materials
Material Work Material Work

function [eV] function [eV]
Aluminum 3.38 Nickel 4.96
Bismuth 4.17 Platinum 5.56
Cadmium 4.0 Potassium 1.6
Cobalt 4.21 Silicon 4.2
Copper 4.46 Silver 4.44
Germanium 4.5 Tungsten 4.38
Gold 4.46 Zinc 3.78
Iron 4.4
Note: 1 eV ¼ 1.602 1019 J.
This then describes the photoelectric effect, which is the basis for a number of
sensing methods, as we shall discuss next. In all of these methods, surface electrons
are released.
Example 4.2: Longest wavelength for photoelectric emission

Consider a photoelectric device intended for light detection.
(a) Assuming it is made of a potassium-coated surface, what is the longest
wavelength that the device can detect?
(b) What is the kinetic energy of an emitted electron under red light radiation at a
wavelength of 620 nm?
Solution:
(a) The photon energy is given in (4.2). With the photon energy equal to the work
function we have
e0
hf ¼ e0 ! f ¼ ½Hz:
h
Since photons travel at the speed of light, the frequency may be written as
c
f ¼ ½Hz;
l
where c is the speed of light and l is the wavelength. The longest wavelength
detectable is
ch 3 108 4:1357 1015

l¼ ¼ ¼ 7:7544 107 m:
e0 1:6
This is 775.44 nm. From Figure 4.1, this is in the very near IR region.
(b) At 620 nm, the frequency is c/l and from (4.2) the kinetic energy is
3 108
k ¼ hf e0 ¼ 4:1357 1015 1:6 ¼ 0:4 ½eV:
620 109
This kinetic energy is rather low because the red light is close to the longest
wavelength to which the photoelectric device responds.
4.4.2.2 Quantum effects: the photoconducting effect

Many modern sensors are based on quantum effects in the solid state, and parti-
cularly in semiconductors. Although some electrons may still leave the surface
based on the photoelectric effect, when a semiconductor material is subjected to
photons they can also transfer energy to electrons in the bulk of the material. If
this energy is sufficiently high, these electrons become mobile, resulting in an
increase in the conductivity of the material and, as a result, in an increase in
current through the material. This current or its effects become a measure of the
radiation intensity (visible light, UV radiation, and, to a lesser extent, IR radia-

tion) that strikes the material. The model for this effect is shown in Figure 4.2(a).
Electrons are normally in the valence band—they are bound to lattice sites within
the crystal (i.e., bound to the atoms that make up the crystal) and have specific
densities and momentum. Valent electrons are those that are bound to an indi-
vidual atom. Covalent electrons are also bound, but are shared between two
neighboring atoms in the crystal. An electron can only move into the conduction
band if its energy is more than the energy gap (bandgap energy, Wbg) specific to
the material and if the momentum of the site in the conducting band is the same
as the momentum of the electron in the valence band (law of conservation of
momentum). This energy may be supplied thermally, but here we are interested
in energy absorbed from photons. If the radiation is of sufficiently high frequency
(sufficiently energetic photons), valence or covalence electrons may be released
from their sites and moved across the bandgap into the conduction band
(Figure 4.2(a)).
There are two possible mechanisms for this transition to occur. In direct
bandgap materials, the momentum in the top of the valence band and in the
bottom of the conduction band are the same and an electron can transit without
the need for a change in momentum, provided it acquires sufficient energy from
the photon interaction. In indirect bandgap materials, the electron must interact
with the crystal lattice to either gain or lose momentum before it can occupy a
site in the conduction band. This process is characterized by a lattice vibration
called a phonon and is a less efficient process than that in direct bandgap mate-
rials. When in the conduction band, electrons are mobile and free to move as a
current. When electrons leave their sites, they leave behind a “hole,” which is
simply a positive charge carrier. This hole may be taken by a neighboring elec-
tron with little additional energy (unlike the original electron released by the
photon; see Figure 4.2(b)) and therefore, in semiconductors, the current is due to
the net concentrations of electrons and holes. The release of electrons is mani-
fested as a change in the concentration of electrons in the conduction band and
etop
− Conduction band
Band gap −
Valence band
ebottom + +
(a) hf >Wbg (b) hf < Wbg
Figure 4.2 A model of the photoconductive effect. (a) The photon energy is
sufficiently high to move an electron across the bandgap, leaving
behind a hole. (b) The photon energy is too low, resulting in
recombination of the electron and the hole
of holes in the valence band. The conductivity of the medium is due to the con-
centrations of both carriers and their mobilities:
s ¼ eðme n þ mp pÞ ½S=m; (4.4)
where s is conductivity, me and mp are the mobilities (in m2/(V s) or, often,
cm2/(V s)) of electrons and holes, respectively, and n and p are the concentrations
(carriers/m3 or carriers/cm3) of electrons and holes. This change in conductivity or
the resulting change in current is then the basic measure of the radiation intensity in
photoconducting sensors.
The effect just described is called the photoconducting effect and is most
common in semiconductors because the bandgaps are relatively small. It exists in
insulators as well; but there, the bandgaps are very high and therefore it is difficult
to release electrons except at very high energies. In conductors, the valence and
conduction bands overlap (there is no bandgap). Most electrons are free to move,
indicating that photons will have minimal or no effect on the conductivity of the
medium. Therefore, semiconductors are the obvious choice for sensors based on the
photoconducting effect, whereas conductors will most often be used in sensors
based on the photoelectric effect.
From Table 4.3, it is clear that some semiconductors are better suited for low-
frequency radiation whereas others are better at high-frequency radiation. The
lower the bandgap, the more effective the semiconductor will be at detection at
Table 4.3 Bandgap energies, longest wavelength, and working temperatures for
selected semiconductors
Material Band gap Longest wavelength Working

[eV] lmax [mm] temperature [K]
ZnS 3.6 0.35 300
CdS 2.41 0.52 300
CdSe 1.8 0.69 300
CdTe 1.5 0.83 300
Si 1.2 1.2 300
GaAs 1.42 0.874 300
Ge 0.67 1.8 300
PbS 0.37 3.35
InAs 0.35 3.5 77
PbTe 0.3 4.13
PbSe 0.27 4.58
InSb 0.18 6.5 77
GeCu 30 18
Hg/CdTe 8–14 77
Pb/SnTe 8–14 77
InP 1.35 0.95 300
GaP 2.26 0.55 300
Note: Properties of semiconductors vary with doping and other impurities. The values shown should be
viewed as representative only.
low frequencies (long wavelengths, hence lower photon energies). The longest
wavelength specified for the material is called the maximum useful wavelength,
above which the effect is negligible. For example, InSb (indium antimony) has a
maximum wavelength of 5.5 mm, making it useful in the near IR range. Its bandgap
is very low, which also makes it very sensitive. However, that also means that
electrons can be easily released by thermal sources and, in fact, the material may be
totally useless for sensing at room temperatures (300 K), because at that tempera-
ture most electrons will be in the conduction band and these available conduction
carriers serve as a thermal background noise for the photon-generated carriers. For
this reason, it is often necessary to cool these long-wavelength sensors to make
them useful by reducing the thermal noise. The third column in Table 4.3 shows the
(highest) working temperature of the material.
4.4.2.3 Spectral sensitivity

Each semiconducting material has a range of the spectrum in which it is sensitive,
given as a function of frequency or wavelength. The upper range (longest wave-
length or minimum energy) is defined by the bandgap (about 1,200 nm in
Figure 4.3). Above the bandgap, the response of a material (i.e., the concentration
of conduction electrons in the conduction band due to photon interaction) increases
steadily to a maximum and then decreases, as shown schematically in Figure 4.3.
The reason for the increase and decrease in response is that electron density and
momentum are highest in the middle of the valence band and taper off to zero at its
boundaries. Because of the law of conservation of momentum, an electron can only
transit into the conduction band to a site of like momentum, and the probability of
this first increases with an increase in energy (decrease in wavelength); but after
most of the electrons in the middle of the valence band have been displaced, the
probability of electron transiting to sites of ever-increasing momentum decreases
until, at an electron energy equal to the difference between the top of the conduc-
tion band and the bottom of the valence band (etop–ebottom in Figure 4.2), this
probability goes to zero. In Figure 4.3, that occurs at approximately 650 nm.
4.4.2.4 Tunneling effect

Another important quantum effect is the tunneling effect in semiconductor devi-
ces. A simple explanation of this curious effect is that although carriers may not
Normalized intensity [%]
100
80
60
40
20
0
600 700 800 900 1,000 1,100 1,200
Wavelength [nm]
Figure 4.3 Spectral sensitivity of a semiconductor

have sufficient energy to go “over” the gap, they can tunnel “through” the
gap. Although this explanation is shaky at best, the tunneling effect is real, is a
direct consequence of quantum mechanics, and is fully predicted by the
Schrödinger equation. The tunneling effect explains behavior on the microscopic
level that cannot be explained through classical physics but which, nevertheless,
manifests itself on the macroscopic level. Semiconductor devices based on this
effect, particularly tunnel diodes, are common, and the effect is used extensively in
optical sensors.
4.5 Quantum-based optical sensors

Optical sensors are divided into two broad classes: quantum sensors (or detectors)
and thermal sensors (or detectors). (Optical sensors are most often called detec-
tors.) A quantum optical sensor is any sensor based on the quantum effects
described above and include photoelectric and photoconductive sensors as well as
photodiodes and phototransistors (variations of the photoconductive sensor).
Thermal optical sensors are mostly encountered in the IR region (and particularly in
the far IR region) and come in many variations, including PIR sensors, active far
infrared (AFIR) sensor, bolometers, and others, as we shall see shortly.
4.5.1 Photoconducting sensors

Photoconducting sensors, or as they are sometimes called, photoresistive sensors
or photoresistive cells, are possibly the simplest optical sensors. They are made
from a semiconducting material connected to two conducting electrodes and are
exposed to light through a transparent window. A schematic view of the sensor is
shown in Figure 4.4(a). Materials used for these sensors are cadmium sulfide
(CdS), cadmium selenide (CdSe), lead sulfide (PbS), indium antimonide (InSb),
and others, depending on the range of wavelengths for which the sensor is designed
to operate and other requisite properties of the sensor. Of these, CdS is the most
common material.
Light
m
inu de
lumectro
A el
m
inu de d
lumectro
CdS A el
L
V w
I
(a) (b)
Figure 4.4 Structure of a photoconductive sensor. (a) Simple electrodes.

(b) Schematic of a sensor showing the connections
In terms of construction, the electrodes are typically set on top of the photo-
conductive layer, which in turn is placed on top of a substrate layer. The electrodes
may be very simple (Figure 4.4(a)) or may resemble a meandering or comblike
shape (Figure 4.4(b)), depending on requirements. In either case, the area exposed
between the electrodes is the sensitive area. Figure 4.5 shows a few sensors of
various sizes and construction. The photoconductor is an active sensor that must be
connected to a source. The current through or the voltage on the sensors is taken as
the output, but what changes with light intensity is the conductivity of the semi-
conductor and hence its resistance.
The conductivity of the device, given in (4.4), results from the charge of
electrons e, the mobilities of electrons and holes (me and mp), and the concentrations
of electrons n and holes p from whatever source. In the absence of light, the
material exhibits what is called dark conductivity, which in turn results in a dark
current. Depending on the construction and materials, the resistance of the device
may be very high (a few megaohms) or may be in the range of a few kilo-ohms.
When the sensor is illuminated, its conductivity changes (the conductivity increases
and hence the resistance decreases) depending on the change in carrier concentra-
tions (excess carrier concentrations).
This change in conductivity is
Ds ¼ eðme Dn þ mp DpÞ ½S=m; (4.5)
where Dn and Dp are the excess carrier concentrations generated by the radiation
(light). The carriers are generated by the radiation at a certain generation rate (the
number of electrons or holes per second per unit volume), but they also recombine
at a set recombination rate. The generation and recombination rates depend on a
variety of properties, including the absorption coefficient of the material, dimensions,
incident power density (of the radiation) wavelength, and the carrier lifetime (the
lifetime of carriers is the time it takes for excess carriers to decay—recombine). Both
generation and recombination exist simultaneously, and under a given illumination a
Figure 4.5 Examples of photoconductive sensors. The sensor on the far right has
simple electrodes. The others have comblike electrodes
steady state is obtained when these are equal. Under this condition, the change in
conductivity may be written as
Ds ¼ egðmn tn þ mp tp Þ ½S=m; (4.6)
where tn and tp are the lifetimes of electrons and holes, respectively, and g is the
rate of carrier generation (number of carriers generated per second per unit
volume). These properties are material dependent and are generally known,
although they are temperature as well as concentration dependent. Although carrier
generation is in pairs, if the preexisting carrier density of one type dominates, the
excess carrier density of the second type will be negligible with respect to the
density of the dominant carrier. If electrons dominate, the material is said to be an
n-type semiconductor, whereas if the dominant carriers are holes, the semi-
conductor is said to be p-type. In each of these, the concentration of the opposite
type is negligible and the change in conductivity is due to the dominant carrier.
An important property of a photoresistor is its sensitivity to radiation (some-
times called its efficiency). Sensitivity, also called gain, is given as
V
G¼ ðm tn þ mp tp Þ ½V=V; (4.7)
L2 n
where L is the length of the sensor (distance between the electrodes) and V is the
voltage across the sensor. Note that the units in (4.7) are volt per volt, hence this is
a dimensionless quantity. Sensitivity gives the ratio of carriers generated per pho-
ton of the input radiation. To increase sensitivity, one should select materials with
high carrier lifetimes, but one must also keep the length of the photoresistor as
small as possible. The latter is typically achieved through the meander construction
shown in Figure 4.4(b) (see also Figure 4.5). The meander shape ensures the dis-
tance between two electrodes is reduced for a given exposure area. It also reduces
the resistance of the sensor that, referring to Figure 4.4(a), is given by
L
R¼ ½W; (4.8)
swd
where wd is the cross-sectional area of the device and s is its conductivity.
The excess carrier density depends on the power absorbed by the photo-
conductor. Given a radiation power density P [W/m2] incident on the top surface of
the photoconductor in Figure 4.4(a) and assuming that a fraction T of this power
penetrates into the photoconductor (the rest is reflected off the surface), the power
entering the device is PTS ¼ PTwL [W]. This is, by definition, the energy per unit
time absorbed in the device. Since the photons have an energy hf, we can write the
total number of excess carrier pairs released per unit time as

PTwL carriers
DN ¼ h ; (4.9)
hf s
where h is the quantum efficiency of the material (a known, given property,
dependent on the material used). The latter indicates how efficient the material is at
converting photon energy into carriers and clearly indicates that not all photons
participate in the process. Assuming carrier generation is uniform throughout the
volume of the photoconductor (an assumption only valid for thin photoconductors),
we can calculate the rate at which carriers are generated per unit volume per
second:

PTwL PT carriers
Dn=s ¼ h ¼h : (4.10)
hfwLd hfd m3 s
As mentioned before, the recombination rate influences the net excess carrier
density. The carrier density (concentration) is then obtained by multiplying the rate
of generation by the lifetime of the carriers, t:

PT t carriers
Dn ¼ h : (4.11)
hfd m3
Both majority and minority excess carriers generated by light have the same
densities.
Some of the terms in (4.11) are not necessarily constant with concentrations
and some may only be estimates as well as being temperature dependent. However,
the equation shows the link between light intensity and excess carrier concentration
and hence the dependence of conductivity on light intensity.
Other parameters to consider are the response time of the sensor, its dark
resistance (which depends on doping), the range of resistance for the span of the
sensor, and the spectral response of the sensor (i.e., the portion of the spectrum in
which the sensor is usable). These properties depend on the semiconductor used as
well as on the manufacturing processes used to produce the sensor.
Noise in photoconducting sensors is another important factor. Much of the
noise is thermally induced and becomes worse at longer wavelengths. Hence, many
IR sensors must be cooled for proper operation. Another source of noise is the
fluctuations in the rates of generation and recombination of the carriers. This noise
is particularly important at shorter wavelengths.
From a sensor production point of view, photoresistive sensors are made either
as a single crystal semiconductor, by deposition of the material on a substrate, or by
sintering (essentially, an amorphous semiconductor made of compressed, powdered
material sintered at high temperatures to form the photoconductive layer). Usually,
sensors made by deposition are the least expensive, whereas single crystal sensors
are the most expensive, but with better properties. A particular method may be
chosen based on requirements. For example, large surface area sensors may need to
be made by sintering because large single crystals are both difficult to make and
more expensive.
Example 4.3: Properties of a photoresistor

Properties of many semiconductors are determined experimentally primarily
because of the variability of constituents of the semiconductors and the difficulties
of obtaining reliable data. Nevertheless, especially for CdS sensors, some reliable
data are available. To see what the properties of a photoresistor are, consider a
simple CdS structure as in Figure 4.4(a) of length 4 mm, width 1 mm, and thickness
0.1 mm. The mobility of electrons in CdS is approximately 210 cm2/(Vs) and that
of holes is 20 cm2/(Vs). The dark concentration of carriers is approximately
1016 carriers/cm3 (for both electrons and holes). At a light density of 1 W/m2 the
carrier density increases by 11%:
(a) Calculate the conductivity of the material and the resistance of the sensor
under dark conditions and under the given illumination.
(b) Assuming a rate of carrier generation due to light of 1015carriers/s/cm3,
estimate the sensitivity of the sensor.
Solution:
(a) The conductivity is calculated directly from (4.4):
s ¼ eðme n þ mp pÞ ¼ 1:602 1019 ð210 1016 þ 20 1016 Þ

¼ 0:36846 ½S=cm:
Because of the units of mobility and carrier density, the result is in siemens
per centimeter. By multiplying this by 100 (1 m ¼ 100 cm), we get
s ¼ 36.85 S/m.
(b) Under light conditions, the carrier density increases by a factor of 1.11 and we
get:
s ¼ eðme n þ mp pÞ
¼ 1:602 1019 ð210 1016 1:11 þ 20 1016 1:11Þ
¼ 0:409 ½S=cm:
The conductivity under light conditions is s ¼ 40.9 S/m.
The resistance is found from (4.9):
L 0:004
R¼ ¼ ¼ 1;085:5 ½W:
sWH 36:85 0:001 0:0001
L 0:004
R¼ ¼ ¼ 978:0 ½W:
sWH 40:9 0:001 0:0001
Note that the resistance is directly proportional to the increase in carrier
density but the increase in carrier density is not linear with illumination. For
this reason, the resistance decreases rather quickly initially but then levels off,
since at high illumination levels there are fewer and fewer carriers available to
be released into the conduction band.
(c) The sensitivity of the sensor could be calculated from (4.7) directly if we had
information on the lifetimes of electrons and holes. In their absence, we write
from (4.5) and (4.6):
ðme Dn þ mp DpÞ
eðme Dn þ mp DpÞ ¼ egðmn tn þ mp tpÞ ! ðmn tn þ mp tp Þ ¼ :
g
Thus, we can rewrite (4.7) as

V me Dn þ mp Dp
G¼ 2
L g

V 210 104 1:0 1016 þ 20 104 1:0 1016
¼
ð0:004Þ2 1015
¼ 14;375 V=V:
Note that we have converted the units of mobility to m2/(Vs) but the units
of carrier density and rate of carrier generation do not need to be converted, as
they appear in the numerator and denominator. This gives a sensitivity of
14,375 V/V, that is, for every 1 V potential difference between the electrodes,
a photon generates 14,375 carriers. This is a very large sensitivity, typical of
CdS sensors.
The most common materials for inexpensive sensors are CdS and CdSe. These
offer high sensitivities (on the order of 103–104), but at a reduced response time,
typically about 50 ms. Construction is by deposition and electrodes are then
deposited to create the typical comblike shape seen in Figures 4.4(b) and 4.5, which
provides a short distance between the electrodes and a large sensing area. CdS and
CdSe can also be sintered. The spectral response of these sensors covers the visible
range, although CdS tends to respond better at shorter wavelengths (violet) while
CdSe responds better at longer wavelengths (red). Materials can be combined to
tailor specific responses. The use of PbS, which is typically deposited as a thin film,
shifts the response into the IR region (1,000–3,500 nm) and improves response to
less than 200 ms, but as is typical of IR sensors, at an increase in thermal noise and
hence the need for cooling. Examples of single crystal sensors are those made from
InSb. A sensor of this type can operate down to about 7,000 nm and can have a
response time of less than 50 ns but must be cooled to operate at the longer
wavelengths, typically to 77 K (by liquid nitrogen). For specialized application in
the IR region, and especially in the far IR, mercury cadmium telluride (HgCdTe),
and germanium boronide (GeB) materials may be used. These, especially GeB, can
extend operation down to about 0.1 mm if cooled to 4 K (by liquid helium).
In general, cooling of a sensor made of any material extends its spectral
response into longer wavelengths, but often slows its response. On the other hand, it
increases sensitivity and reduces thermal noise. Many of the far IR applications
are military or space applications. These specialized sensors must be made of single
crystals and must be housed in a package that is compatible with the low-temperature
requirements.
4.5.2 Photodiodes
If the junction of a semiconducting diode is exposed to light radiation, the gen-
eration of excess carriers due to photons adds to the existing charges in the con-
duction band exactly in the same fashion as for a pure semiconductor. The diode
itself may be forward biased (Figure 4.6(a)), reverse biased (Figure 4.6(b)), or
unbiased (Figure 4.6(c)). Figure 4.6(d) shows the current–voltage (I–V) character-
istics of the diode. Of the three configurations in Figure 4.6, the forward-biased
mode is not useful as a photosensor because in this mode the normal current (not
due to photons) is large in comparison to the current generated by photons. In the
reverse-biased mode, the diode carries a minute current (i.e., a “dark” current) and
the increase in current due to photons is large in comparison. In this mode, the
diode operates in a manner similar to the photoconducting sensor and is therefore
called the photoconducting mode of the diode. If the diode is not biased it operates
as a sensor (or an actuator) in the photovoltaic (PV) mode (Figure 4.6(c)).
The equivalent circuit of a diode in the photoconductive mode (Figure 4.6(b))
is shown in Figure 4.7(a). In addition to the current that would exist in the ideal
diode (Id), there is also a leakage current (I0) defined by the “dark” resistance R0
and a current through the capacitance (Ic) of the junction. The series resistance Rs is
due to conductors connecting the diode. The photons release electrons from the
valence band either on the p or n side of the junction. These electrons and the
resulting holes flow toward the respective polarities (electrons toward the positive
pole, holes toward the negative pole) generating a current, which in the absence of
a bias current in the diode constitutes the only current (the diode is reverse biased).
In practice, there will be a small leakage current, shown in the equivalent circuit as
[mA] Forward current

300
200
R [V] 100
R −40 −30 −20 −10 1 2 [V]
V + V +
Reverse −5 Forward
voltage −10 voltage
−20
−30
(a) (b) (c) (d) [μA] Reverse current
Figure 4.6 The semiconducting (p–n) junction. (a) Forward biased. (b) Reverse
biased. (c) Unbiased. (d) The I–V characteristics of the junction
I
Rs IL
Id I0 Ic
It RL
Vd
V
R0
C
I01
+ V I02
I03
(a) (b)
Figure 4.7 A photodiode connected in the photoconductive mode (reverse biased).

(a) Equivalent circuit. (b) I–V characteristics
I0. The attraction of electrons by the positive pole will tend to accelerate them and
in the process they can collide with other electrons and release them across the
bandgap, especially if the reverse voltage across the diode is high. This is called an
avalanche effect and results in multiplication of the carriers available. Sensors that
operate in this mode are called photomultiplier sensors.
In any diode, the current in the forward-biased mode is
Id ¼ I0 ðeeVd =nkT 1Þ ½A (4.12)

where I0 is the leakage (dark) current, e (in the exponent) is the charge of the
electron, Vd is the voltage across the junction, also called the voltage barrier or
built-in potential, k is Boltzmann’s constant (k ¼ 1.3806488 1023 kg m2/s2/K),
T is the absolute temperature [K], and n is an efficiency constant, also called an
ideality factor,varying in value between 1 and 2. In ideal diodes and in many
practical cases, it is equal to 1 (but see Section 3.4, where it was equal to 2). This
relation clearly indicates the dependence of the diode current on temperature as
well as on the bias voltage. However, in the reverse mode, only the current I0 may
flow and for most practical purposes it is very small, often negligible.
The current produced by photons is given as
hPAe
Ip ¼ ½A; (4.13)
hf
where P is the radiation (light) power density [W/m2], f is the frequency [Hz], and h
is Planck’s constant. All other terms are constants associated with the diode or the
semiconductor used. h is the quantum absorption efficiency, A is the exposed area
of the diode (hPA is the power absorbed by the junction). The total current avail-
able external to the diode is (using n ¼ 1)
hPAe
IL ¼ Id Ip ¼ I0 ðeeVd =kT 1Þ ½A (4.14)
hf
This is the current measured for a photodiode sensor under forward or back-
ward biasing conditions (depending on the voltage Vd across the junction). Under
reverse-bias conditions, the first term may be neglected altogether since I0 is small
(on the order of 10 nA) and Vd is negative. As a first approximation, especially at
low temperatures, one obtains a simple relation for the photodiode:
hPAe
IL ½A: (4.15)
hf
When measured, this current gives a direct reading of the power absorbed by
the diode, but as can be seen, it is not constant since the relation depends on
frequency and the power absorbed itself is frequency dependent except for mono-
chromatic radiation. As the input power increases, the characteristic curve of the
diode changes, as shown in Figure 4.7(b), resulting in an increase in the reverse
current, as expected.
Anode Anode Anode
SiO2 SiO2 SiO2 Thin gold anode
p p p
Intrinsic i n n
n layer
n n+
(a) Cathode (b) Cathode (c) Cathode (d) Cathode
Figure 4.8 Various structures of photodiodes. (a) Common planar structure.

(b) PIN diode. (c) pnnþ structure. (d) Schottky diode
Any diode can serve as a photodiode, provided that the n region, p region, or
p–n junction are exposed to radiation. However, specific changes in materials and
construction have been made to common diodes to improve one or more of their
photoconducting properties (usually, the dark resistance and response time). Taking
as an example the planar diffusion type of diode shown in Figure 4.8, it consists of
p and n layers and two contacts. The region immediately below the p layer is the so-
called depletion region, which is characterized by an almost total absence of car-
riers. This is essentially a regular diode. To increase dark resistance (lower dark
current), the p layer may be covered with a thin layer of silicon dioxide (SiO2)
(Figure 4.8(a)). The addition of an intrinsic layer of the semiconductor between the
p and n layers produces the so-called PIN photodiode, which, because of the high
resistance of the intrinsic layer, has lower dark current and lower junction capaci-
tance and hence better time response (Figure 4.8(b)). The exact opposite is true in
the pnnþ construction in which a thin, highly conductive layer is placed at the
bottom of the diode. This reduces the resistance of the diode and improves low-
wavelength sensitivity (Figure 4.8(c)). Another way of altering the response of a
diode is through the use of a Schottky junction. In this diode, the junction is formed
by use of a thin layer of sputtered conducting material (gold) on an n layer (the
Schottky junction is a metal–semiconductor junction) (Figure 4.8(d)). This pro-
duces a diode with a very thin outer layer (metal) above the n layer, improving its
long wavelength (IR) response. As mentioned before, a diode with high reverse
bias may operate in avalanche mode, increasing the current and providing a gain or
amplification (photomultiplier diode). The main requirement needed to obtain
avalanche is the establishment of a high reverse electric field across the junction
(on the order of 107 V/m or higher) to provide sufficient acceleration of electrons.
In addition, low noise is essential. Avalanche photodiodes are available for high-
sensitivity, low-light applications.
Photodiodes are available in various packages, including surface mount,
plastic, and small can packages. Figure 4.9(a) shows one type of diode used as a
detector for reflected laser light in CD players. They are also available in linear
arrays of photodiodes, such as in Figure 4.9(b), which shows a 512 element linear
array used as the sensor for a scanner. They are available for IR as well as for the
visible range and some extend the range into the UV and even the X-ray range.
Many photodiodes have a simple lens to increase the power density at the
junction.
(a) (b)
Figure 4.9 (a) A photodiode used as a sensor in a CD player shown installed in its
holder. (b) A photodiode linear array (512 photodiodes) in a single
integrated circuit used as the sensor element in a scanner. The top
cover is glass and light is allowed in through the transparent slit
+12 V
Laser Optical fiber
16 km
R 1 MΩ V0
Figure 4.10 An optical fiber communication link with the source (laser) and a
photodiode used as a detector
Example 4.4: Photodiode as a detector for fiber-optic communication

A digital communication link uses a red laser operating at 800 nm with an output of
10 mW. The optical link is 16 km long and is made of an optical fiber with an
attenuation of 2.4 dB/km. At the receiving end of the link, a photodiode detects the
pulses and the output is measured across a 1 MW resistor, as shown in Figure 4.10.
Assuming a series of pulses is transmitted by the laser and there are no losses on
either end, that is, all power produced by the laser enters the optical fiber and all
power at the diode is absorbed by the diode, calculate the amplitude of the received
pulses. Assume that the dark current (leakage current) of the diode is 10 nA and the
system operates at 25 C. How does the amplitude change if the temperature rises
to 50 C?
Solution: We calculate the diode current using (4.12) followed by the photon
current using (4.13), or we can calculate the total current using (4.14). However, to
do so, we need the radiation power density P. That is calculated using the incident
power density and the attenuation along the optical fiber as follows:
The laser produces 10 mW; but to calculate the power entering the diode, we
need to take into account the losses. To do so, we first calculate the input power, P,
in decibels:
P ¼ 10 103 W ! P ¼ 10 log10 ð10 103 Þ ¼ 20 dB:

The total attenuation along the line is 2.4 16 ¼ 38.4 dB. Therefore, the
power in decibels at the end of the line is
P ¼ 20 38:4 ¼ 58:4 dB:
Now, we convert this back to power. To do so, we write
10 log10 P0 ¼ 58:4 ! log10 P0 ¼ 5:84 ! P0 ¼ 106:84

¼ 1:445 106 W:
Now, we can use (4.14), but we first note that the term hPA is the total power
received by the diode. That, according to the problem statement, equals P0. Thus,
we write
P0 e 19 23
IL ¼ I0 ðeeVd =kT 1Þ ¼ 10 109 ðe1:6110 ð12Þ=1:380648810 298 1Þ
hf
1:445 106 1:61 1019
¼ 10 109 936:3 109 A:
6:6262 1034 3:75 1014
Clearly, the current due to temperature is negligible. Now, we can calculate the
output voltage across the resistor. When the laser beam is off, the current through
the photodiode is 10 nA and the output voltage is
V0 ¼ 10 109 1 106 ¼ 10 mV:

When the light beam is on, the current increases to 946.3 nA and the output
voltage is
V0 ¼ 946:3 109 1 106 ¼ 946:3 mV:

That is, the pulses will result in a voltage that changes from 10 mV for level
“0” to 0.946 V for level “1” of the pulse. This may not be sufficient for interfacing
and may require amplification.
At 50 C, the solution is essentially the same since the component of the
current that depends on temperature is negligible in this case. However, this is not a
general conclusion.
4.5.3 Photovoltaic diodes

Photodiodes may also operate in photovoltaic mode as shown in Figure 4.11(a). In
this mode, the diode is viewed as a generator and requires no biasing. The best-
known structure for photovoltaic diodes is the solar cell, which is a photodiode with
Rs IL
Id I0 Ic
It
Vd
RL
C
R0
(a) (b)
Figure 4.11 (a) Photodiode connected in the photovoltaic mode—equivalent

circuit. The diode is unbiased. (b) A particular type of photodiode,
the photovoltaic cell or solar cell. Two types are shown. The one on
the right is used in a calculator
a particularly large exposed area. All photodiodes can operate in this mode; but as a
rule, the larger the surface area, the larger the junction capacitance. This capacitance is
the main reason for the reduced time response of PV cells. In most other respects,
photodiodes operating in PV mode have the same properties as photodiodes operating
in photoconducting mode. There are differences as well. For example, the avalanche
effect cannot exist in this mode since there is no bias. Figure 4.11(a) shows the
equivalent circuit for a PV cell and Figure 4.11(b) shows two small PV (solar) cells.
Although typically used in PV arrays for solar power generation as well as
smaller arrays used to power small appliances (such as calculators), the PV diode
also makes an exceedingly simple light sensor that needs little more than a volt-
meter to measure light power density or light intensity.
Although the photovoltaic diode operates without a bias, under normal operation a
voltage develops across the junction and the total current is described by (4.14), where
the first term is the normal diode current and the second is the photocurrent. There are
two important properties in the operation of the diode that must be addressed. The first
is the short-circuit current. If the diode is short-circuited, the voltage across the diode is
zero and the only current that may exist is the photocurrent. Thus,
hPAe
Isc ¼ Ip ¼ ½A: (4.16)
hf
The second term is the open circuit voltage, characterized by the fact that the
normal diode current equals the photocurrent, that is, when the load current in
(4.14) is zero. The open circuit voltage, Voc, can be evaluated from this balance:
hPAe
I0 ðeeVoc =nkT 1Þ ¼ : (4.17)
hf
Clearly, Voc is equal to the built-in potential or potential barrier, which depends
on the material, doping, and, through carrier concentration, on temperature. The
efficiency term h in (4.16) and (4.17) is an overall efficiency of the cell and is a
product of the quantum absorption coefficient and the conversion coefficient of the
cell. The efficiency factor n is assumed to be equal to 1 unless otherwise indicated
based on the properties of particular photodiodes.
Example 4.5: Properties of a solar cell at low light

To establish the transfer function of a solar cell, one must measure the input power or
power density and the output voltage or power supplied by the cell. In particular, at low
light, the conversion efficiency of the cell is low and the transfer function is very much
dependent on the load (RL in Figure 4.11(a)). A solar cell, 11 cm 14 cm in area, is
connected to a 1 kW load, exposed to light from an artificial source, and the output
voltage is measured. The voltage versus input power density is shown in Figure 4.12(a).
The curve is characteristic, indicating that the cell tends toward a saturation output at
some (higher) power density. This particular solar cell only supplies a few milliwatts
because of the low illuminance level.
One of the important parameters of solar cells is its conversion efficiency.
Defined as the output power divided by the input power, it is usually given in
percentages. As with all other characteristics, it depends on load and operating
point (input power). The efficiency of the cell used here is calculated as follows:
Pout V 2 =RL
eff ¼ 100% ¼ 100%;
Pin Pd S
where Pd is the input power density [mW/cm2], S is the area of the cell [cm2] (in
this case 11 cm 14 cm ¼ 154 cm2), RL is the load [kW], and V is the output
voltage [V]. The efficiency is plotted in Figure 4.12(b). The efficiency reaches a
maximum of 7.78% at 0.174 mW/cm2 and declines beyond that, indicating the
same tendency toward saturation. It should be noted again that the maximum
efficiency point depends on the illuminance level and the load. Good solar cells
will have efficiencies between 15% and 30%.
7 8.0
6
Output voltage [V]
Efficiency [%]
5 6.0
4
4.0
3
2 2.0
1
0 0
0 50 100 150 200 250 300 0 0.5 1.0 1.5 2.0
(a) Input power [mW] (b) Power density [mW/cm2]
Figure 4.12 Characteristics of a solar cell at low light. (a) Output voltage versus
input power density. (b) Power conversion efficiency versus input
power density
Example 4.6: Optical position sensor

A position sensor is made as follows: a triangular slit is cut into an opaque material
that covers a solar cell to create a triangular area of exposure (Figure 4.13). This
makes the stationary part of the sensor. A moving part is placed above the sta-
tionary part and includes two items. One is a stationary source of light, and between
the source and the stationary slit is a thin rectangular opening that allows light to go
through that opening only. The opening is t m wide and the source supplies an
illuminance of I lux. The position of the slit is the output of the sensor. Since the
larger h (distance) is, the larger the light power on the cell, the larger the voltage of
the solar cell. Assuming the voltage output to be linear with the power incident on
the solar cell following the relation V ¼ kP, where P is the incident power and k a
constant of the cell, find a relation between the measured voltage of the cell and the
position of the slit, h.
Solution: As h increases so does the amount of light reaching the cell because the
width of the illuminated slit increases. Illuminance is measured in lux, which has
units of watts per square meter. Therefore, the power reaching the cell is propor-
tional to the area of the triangular slit covered by the rectangle of width t. We
calculate this area, multiply by the illuminance, and find the power P. It should be
recalled, however, that 1 lux is (1/683) W/m2. Then, of course, the output voltage is
immediately available.
Assuming the slit is at a height h as shown, the area exposed is
ð2x0 þ 2xÞt
S¼ ¼ ðx0 þ xÞt ½m2 :
2
To calculate x and x0 , we note the following:
b x x0
¼ ¼ :
a h hþt
Solar cell
b
Exposed
Moving solar cell
part Light Slit
x' t
Stationary part
a x
Moving part
Slit h Stationary
part
Top view Side view
Figure 4.13 An optical position sensor. The slit in the top layer indicates the
position by exposing a strip width that depends on position
Therefore,
b b
x ¼ h; x0 ¼ ðh þ tÞ ½m:
a a
We have

b b bt
S¼ ðh þ tÞ þ h t ¼ ð2h þ tÞ ½m2 :
a a a
The power reaching the cell is
btI
P ¼ SI ¼ ð2h þ tÞ ½W;
a
where I is the illuminance in [W/m2]. The output voltage is
btI bt2 I
V ¼ kP ¼ 2k hþk ½V:
a a
A linear relation between voltage measured and height h is obtained as expected.
Note also that the sensitivity can be increased by increasing b, t, and/or I.
4.5.4 Phototransistors
As an extension of the discussion on photodiodes, the phototransistor can be
viewed as two diodes connected back to back, as shown in Figure 4.14 for an npn
transistor. With the bias shown, the upper diode (the collector–base junction) is
reverse biased while the lower (base–emitter) junction is forward biased. In a
regular transistor, a current Ib injected into the base is amplified using the following
simple relation:
Ic ¼ bIb ½A; (4.18)
Collector (c) Ic
c
– – – Collector (c)
Light n c
– – –
Vce
+ + p+
b
– – –
n
Emitter (e) Ie = Ic + Ib
– – –
e
(a) Emitter (e) (b) (c) e (d)
Figure 4.14 An npn phototransistor. (a) Schematic structure and junctions.

(b) The circuit schematic. (c) The two junctions form diodes
as shown. (d) Currents in a phototransistor
where Ic is the collector current and b is the amplification or gain of the transistor,
which depends on a variety of factors, including construction, materials used, doping,
etc. The emitter current Ie is
Ie ¼ Ib ðb þ 1Þ ½A: (4.19)
The relations above apply to any transistor. What is unique in a phototransistor
is the means of generating the base current. When a transistor is made into a pho-
totransistor, its base connection is usually eliminated and a provision is made for
the radiation to reach the collector–base junction. The device operates as a regular
transistor with its base current supplied by the photon interaction with the collector–
base junction (which is reverse biased). The transistor described here is also called a
bipolar junction transistor. This name distinguishes it from other types of transistors,
some of which we will encounter later.
Under dark conditions, the collector current is small and is almost entirely due
to leakage currents, designated here as I0. This causes a dark current in the collector
and emitter as
Ic ¼ I0 b; Ie ¼ I0 ðb þ 1Þ ½A: (4.20)
When the junction is illuminated, the diode current is the current due to pho-
tons obtained in (4.13):
hPAe
I b ¼ Ip ¼ ½A: (4.21)
hf
The collector and emitter currents are then
hPAe hPAe
I c ¼ Ip b ¼ b ; Ie ¼ Ip ðb þ 1Þ ¼ ðb þ 1Þ ½A; (4.22)
hf hf
where the leakage current was neglected in the final relations, as was done for the
photodiode. Clearly then, the operation of the phototransistor is identical to that of the
photodiode except for the amplification, b, provided by the transistor structure. Since b,
for even the simplest transistors, is on the order of 100 (but can be much higher), and
the amplification is linear in most of the operation range (see Figure 4.15(a)), the
Ic Ib6
Breakdown region
Ib5
Ib4
Ib3
Ib2
Ib1
Vce
(a) (b)
Figure 4.15 (a) The I–V characteristics of a transistor as a function of base

current. In a phototransistor, the base current is supplied by photon
interaction. (b) Phototransistors equipped with lenses
phototransistor is a very useful device and is commonly used for detection and sensing.
The high amplification allows phototransistors to operate at low illumination levels. On
the other hand, thermal noise can be a significant problem, again because of the
amplification. In particular, the base–emitter junction behaves as a regular diode as far
as current through it. The latter is given in (4.12), where again, I0 is the dark current.
Although this current is small, the fact that the diode is forward biased, and due to the
amplification of the transistor, the effects of temperature are significant.
In many cases, a simple lens is also provided to concentrate the light on the
junction, which for transistors is very small. A phototransistor equipped with a lens
is shown in Figure 4.15(b).
Photoconducting sensors, photodiodes, and phototransistors can sense and
measure directly the radiation power they absorb. However, they can easily be used
to sense any other quantity or effect that can be made to generate or alter radiation
in the range in which the sensor is sensitive. As such, they can be employed to
sense position, distance, dimensions, temperature, and color variations, in counting
events, for quality control, and much more.
Example 4.7: Sensitivity of a phototransistor

Figure 4.15(a) shows the I–V characteristics of a transistor as a function of base
current. However, in a phototransistor, the base current is not measurable. Rather,
the current is a function of light power density on the junction. The following is an
experimental evaluation of the current in a phototransistor as a function of incident
light power density. The table below shows selected values, but all values measured
are plotted in Figure 4.16. Since the curve is linear for light power densities
between 0 mW/cm2 and about 400 mW/cm2, the sensitivity of the sensor can be
written using any two columns in the table. Taking the first and the eighth columns,
we find its sensitivity in the linear range as
0:13 0:00182 0:12818
s¼ ¼ ¼ 0:8545 mA= mW=cm2 :
152 2 150
0.35
0.3
0.25
Current [mA]
0.2
0.15
0.1
0.05
0
0 100 300 500 700 900
Light power density [μW/cm2]
Figure 4.16 Collector current in a phototransistor as a function of input light

power density. Note the saturation above 400 mW/cm2
Power density 2 9.57 20.7 46.2 60.4 83.9 113 152 343 409
[mW/cm2]
Current [mA] .00182 .00864 .0182 .0409 .0532 .0732 .0978 .13 .28 .324
4.6 Photoelectric sensors

Photoelectric sensors, including photomultipliers, are based on the photoelectric
effect (also called the photoemissive effect). As described in Section 4.4.2, the
photoelectric effect relies on photons with energy hf impinging on the surface of a
material. The radiation is absorbed by giving this energy to electrons and these are
emitted from the surface, provided the energy of the photon is higher than the work
function of the material. One can say that the collision between the photon and
electron releases the electron if the energy exchanged is sufficiently high. This effect
has been applied directly to the development of photoelectric sensors (sometimes
called photoelectric cells). In fact, this type of optical sensor is one of the oldest
optical sensors available.
4.6.1 The photoelectric sensor

The principle of the photoelectric sensor is shown in Figure 4.17. The photocathode
is made of a material with relatively low work function to allow efficient emission
of electrons. These electrons are then accelerated toward the photoanode because of
the potential difference between the anode and cathode. The current in the circuit is
then proportional to radiation intensity. The number of emitted electrons per photon
is the quantum efficiency of the sensor and depends to a large extent on the material
used for the photocathode (its work function). Many metals may be used for this
purpose, but for the most part their efficiencies are low. More often, cesium-based
materials are used because they have low work functions and fairly wide spectrum
responses down to about 1,000 nm (well into the IR region). Their response extends
into the UV region as well. In older devices, highly resistant cathodes were made of
a metal such as tantalum or chromium and coated with alkali compounds (lithium,
potassium, sodium, or cesium, or more often, a combination of these; see the
Light
– –
–
– –
C A
IL
V RL
Figure 4.17 Photoelectric sensor and biasing circuit

periodic table). This provides the low work function necessary. The electrodes are
housed in an evacuated tube or in a tube with a noble gas (argon) at low pressure.
The presence of gas increases the gain of the sensor (defined as the number of
electrons emitted per incoming photon) by internal collisions between emitted
electrons and the atoms of the gas through ionization of the gas. Newer devices use
the so-called negative electron affinity (NEA) surfaces. These are constructed by
the evaporation of cesium or cesium oxide onto a semiconductor’s surface.
Classical photoelectric sensors require relatively high voltages for operation
(sometimes, a few hundred volts) to supply useful sensing currents. NEA devices
operate at much lower potentials.
4.6.2 Photomultipliers
Photomultipliers are a development of the classical photoelectric sensor. Whereas,
in a photoelectric sensor, the current is low (the number of electrons emitted is
small), photomultipliers, as their names imply, multiply the available current,
resulting in sensors that are considerably more sensitive than the simple photo-
electric cell. The construction is shown schematically in Figure 4.18. It consists of
an evacuated tube (or a low-pressure gas-filled tube) made of metal, glass, or metal-
encased glass with a window for the incoming radiation. The photocathode and
photoanode of the basic photoelectric cell are maintained, but now there is a
sequence of intermediate electrodes, as shown in Figure 4.18(a). The intermediate
electrodes are called dynodes are made of materials with low work functions, such
as beryllium copper (BeCu), and are placed at potential differences with respect to
preceding dynodes, as shown in Figure 4.18(b). The operation is as follows: the
incident radiation impinges on the cathode and releases a number of electrons,
say, n. These are accelerated toward the first dynode by the potential difference, V1.
These electrons now have sufficient energy to release, say, n1 electrons for each
impinging electron. The number of electrons emitted from the first dynode is
n n1. These are accelerated toward the second dynode, and so on, until they
finally reach the photoanode. The multiplication effect at each dynode results in a
Dy
no Anode
e
od
de
th
1 Dynodes
Ca
C A
V1 V2 V3 Vn
e3
Dy
Incoming radiation
od
no
yn
de
(photons)
D
2
(a) (b)
Figure 4.18 (a) Basic structure of a photomultiplier. (b) Biasing of the dynodes
and photoanode. The typical potential difference between an anode
and a cathode is about 600 V, about 60–100 V between each of the
two dynodes
very large number of electrons reaching the photoanode for each photon impinging
on the cathode. Assuming there are k dynodes (10–14 is not unusual) and n is the
average number of electrons emitted per dynode (secondary electrons), the gain
may be written as
G ¼ nk : (4.23)
This gain is the current amplification of the photomultiplier and depends on the
construction, the number of dynodes, and the accelerating interelectrode voltages.
Clearly, additional considerations must be employed for maximum performance.
First, electrons must be “forced” to transit between electrodes at about the same
time to avoid distortions in the signal. To do so, the dynodes are often shaped as
curved surfaces that also guide the electrons toward the next dynode. Additional
grids and slats are added for the same purpose, to decrease transit time and improve
quality of the signal, especially when the photomultiplier is used for imaging.
As with all sensors of this type, there are sources of noise, but because of the
multiplying effect, noise is particularly important in photomultipliers. Of these, the
dark current due to thermal emission, which is both potential and temperature
dependent, is the most critical. The dark current in a photomultiplier is given as
I0 ¼ aAT 2 eE0 =kT ½A; (4.24)
where a is a constant depending on the cathode material, generally around 0.5, A is

a universal constant equal to 120.173 A/cm2, T is the absolute temperature, E0 is the
work function of the cathode material [e V], and k is Boltzmann’s constant.
Because all terms in (4.24) are constant except the temperature, the dark current
may be viewed as a thermionic current or noise. In photomultipliers, this current is
small because the cathode is cold and under these conditions the thermal emission
is low. Nevertheless, a dark current between 1 nA and 100 nA is present because of
the high gain of the photomultiplier. In addition, the shot noise due to fluctuations
in the current of discrete electrons and multiplication noise due to the statistical
spread of electrons limits the sensitivity of the device. A major concern with pho-
tomultipliers is their susceptibility to magnetic fields. Since magnetic fields apply a
force on moving electrons, they can force electrons out of their normal paths reducing
their gain and, more critically, distorting the signal.
Nevertheless, with proper construction (including possible cooling of the
sensor to reduce dark current) an exceedingly sensitive device can be made. These
sensors are therefore used for very low-light applications, such as in night vision
systems. For example, a photomultiplier sensor may be placed at the focal point of
a telescope to view extremely faint objects in space.
Photomultipliers are part of a broader class of devices called image intensi-
fiers that use various methods (including electrostatic and magnetic lenses) to
increase the current due to radiation. Because their output is sometimes the image
itself, they are sometimes called light-to-light detectors. Figure 4.19 shows a small
photomultiplier.
Figure 4.19 A photomultiplier. The light enters through the top circular surface
on the left. The dynodes are the curved surfaces seen on top
These devices have many disadvantages, including problems with noise, as

discussed above, size, the need for high voltages (in excess of 2,000 V for some), as
well as cost. For these reasons, except for some applications in night vision, they
have been largely displaced by charge-coupled devices (CCDs), which have many
of the advantages of photomultipliers while eliminating most of the problems
associated with photomultipliers.
Example 4.8: Thermionic noise in a photomultiplier

A photomultiplier with 10 dynodes has a cathode coated with potassium to increase
sensitivity. Calculate the thermally produced dark current at the cathode and at the
anode at 25 C, assuming that each incoming photon is energetic enough to release
six electrons and that each accelerated electron releases six electrons.
Solution: The work function of potassium is 1.6 eV (see Table 4.2). Room temperature
is 273.15 þ 25 ¼ 298.15 K. With the Boltzmann constant, k ¼ 8.62 105 eV/K,
we get
I0 ¼ aAT 2 eE0 =kT ¼ 0:5 120:173 104 ð298:15Þ2 e1:6=8:6210

5
298:15
¼ 4:9 1017 A:
This is a mere 4.9 108 nA. Since each accelerated electron releases six
electrons, the gain of the photomultiplier is
G ¼ nk ¼ 610 ¼ 6:05 107 :

The current at the anode due to thermionic emission is
Ia ¼ 4:9 1017 6:05 107 ¼ 2:96 109 A:

This is just under 3 nA.
It is because of these very low dark currents that the photomultiplier is so very
useful and has survived into the age of semiconductors.
4.7 Charge coupled (CCD) sensors and detectors

Charge coupled devices are typically made of a conducting substrate on which a p-
or n-type semiconductor layer is deposited. Above it lies a thin insulating layer
made of silicon dioxide to insulate the silicon from a transparent conducting layer
above it, as shown in Figure 4.20(a). This structure is called a metal oxide semi-
conductor (MOS) and is a simple and inexpensive structure. The conductor (also
called a gate) and the substrate form a capacitor. The gate is biased positively with
respect to the substrate (for an n-type semiconductor). This bias causes a depletion
region in the semiconductor and, together with the silicon dioxide layer, makes this
structure a very high-resistance device. When optical radiation impinges on the
device, it penetrates through the gate and oxide layer to release electrons in the
depletion layer. The charge density released is proportional to the incident radiation
intensity. These charges are attracted toward the gate but cannot flow through
the oxide layer and are trapped there. There are a number of methods to measure
the charge (and hence the radiation intensity that produced it). In its simplest form,
one can reverse bias the MOS device to discharge the electrons through a resistor,
as shown in Figure 4.20(b). The current through the resistor is a direct measure of
the light intensity on the device. However, the main value of CCDs is in building
one-dimensional (linear arrays) or two-dimensional arrays of MOS devices for the
purpose of imaging. In such cases, it is not possible to use the method in
Figure 4.20(b) directly. Rather, the basic method is to move the charges of each cell
to the next in a kind of “musical chairs” sequence by manipulating gate voltages.
In this method, the transfer is one cell per step and the current in the resistor for
each step corresponds to a particular cell. This is shown in Figure 4.21(a) for one
row in the two-dimensional array. At the end of this scan, the array can be scanned
again to read a new image. The scan for a two-dimensional array is shown sche-
matically in Figure 4.21(b). The data are moved vertically one row at a time, that is,
all cells move their data one row lower, whereas the lowest row moves it into a shift
resister. The scan stops and the shift register is moved to the right to obtain the
signal for one row (similar to that in Figure 4.21(a)). Then the next row is shifted
Radiation
Thin metal electrode
CCD
Metal oxide layer
− − − − − − − − − V
− Intrinsic region
+
R V0
Substrate
(a) (b)
Figure 4.20 The basic CCD cell. (a) In forward-biased mode, electrons
accumulate below the MOS layer. (b) In reversed-biased mode,
the charge is sensed by discharging it through an external load
+ + + + + +
1 2 N1 CCD array
− −− −− − −− − −− CCD 1
2
V0 Amplification,
1 2 3 4 5 6 N2
A/D conversion
V0 Register
(1) (2) (3) (4) (5) (6)
Steps
(a) (b)
Figure 4.21 Method of sensing the charge in a CCD array. (a) The charge is
moved in steps to the edge (by manipulating the gate voltages) and
discharged through a resistor. (b) Two-dimensional scan of an
N1 N2 image
until the whole array has been scanned. In practice, each cell is equipped with three
electrodes, each covering one-third of the cell and the time step described above is
made of three pulses or phases. All first electrodes in a row are connected to each
other, all second electrodes in a row form a second phase, and all third electrodes in
a row form the third phase. The phases are powered in sequence, moving all charge
in each row downward one-third of the cell. After three pulses, the charge of each
row is transferred to the row below. The signal obtained is typically amplified and
digitized and used to produce the image signal, which can then be displayed on a
display array such as a TV screen or a liquid crystal display. Of course, there
are many variations of this basic process. For example, to sense color, filters may
be used to separate colors into their basic components (red–green–blue [RGB] is
one method). Each color is sensed separately and forms part of the signal. Thus, a
color CCD will contain four cells per “pixel,” one reacting to red, two reacting to
green (our eyes are most sensitive to green), and one reacting to blue. In some
higher-quality imaging systems, each color is sensed on a separate array, but a
single array and filters arrangement is more economical.
One important issue in image transfer is the time required to do so. As the
resolution of CCDs increases, this issue becomes more important. Another problem
is associated with masking of the sensitive array during the transfer of the image.
These issues are resolved in a number of ways. The most obvious is the use of a
shutter, which opens to expose the array and closed during the period of image
transfer. Although this results in the best image quality, it is slow and results in
relatively low frame rates. Another method is the frame transfer method, which does
not require a shutter but rather, uses two identical CCD arrays side by side with one
exposed to create the image while the second is masked and serves to store the image
while the next image is acquired. The image is transferred onto the masked array at
the required frame rate and the image then transferred from the storage array during
the time the next image is acquired. This results in higher frame rates but because the
images are acquired continuously, it can also result in smearing of the images. It is
also a more expensive solution because of the need of two identical CCDs. A third
method is the interline transfer in which an array with double the number of columns
is used with alternating columns masked to serve as storage. Although the number of
pixels doubles as in the frame transfer method, the transfer of the image from the
exposed columns to the storage columns is much faster and hence there is less
smearing of the image resulting in higher resolution images. The main problem here
is that the production of an array with double the 2N pixels is much more expensive
that the production of two arrays with N pixels. Regardless of the method of transfer,
the image is created using the process described in Figure 4.21.
CCD devices are the core of electronic cameras and video recorders but are
also used in scanners (where linear arrays are typically used). They are also used
for very low-light applications by cooling the CCDs to low temperatures. Under
these conditions, their sensitivity is much higher, primarily due to reduced thermal
noise and hence have better signal to noise ratio. In this mode, CCDs have suc-
cessfully displaced photomultipliers in most applications.
Example 4.9: Some considerations in CCD imaging

CCDs are very common in still and video cameras, as they are inexpensive com-
pared with other imaging devices and can be produced on a chip, making them
ideal for miniaturization. The resolution, usually defined as the number of pixels,
can be very high while maintaining a small surface area, something that allows the
use of small lenses with minimal motion for focusing and zooming. In the extreme,
the imaging area can be quite large, with many millions of pixels. However,
because even a simple camera contains an imaging sensor of a few million pixels,
the transfer of images is not a mere formal process. Rather, it is a limiting process
that defines, for example, how quickly an image can be recorded.
Consider a digital camera imager with 12 megapixels in a 4:3 format. The
image has 3,000 columns with 4,000 rows (see Figure 4.21(b), for the schematic
structure). With three steps per row, the transfer of 3,000 rows will take 9,000 steps.
Each row is first transferred into a 4,000-position register that must be shifted out
one cell at a time to generate the row signal. Assuming that each operation takes the
same amount of time, there are 3,000 3 4,000 ¼ 36 106 steps to be per-
formed before the image is retrieved from the CCD. Assuming, arbitrarily that a
step can be performed in 50 ns, it will take a minimum of 1.8 s to do so.
This means of course that a camera cannot record video in that resolution.
Digital cameras that record video do so in a reduced resolution format, typically in
video graphics array (VGA) or high-definition (HD) formats. For example, in VGA
format the camera only records 640 480 pixels/frame. With the three steps
needed, the transfer takes 640 3 480 ¼ 921,400 steps. At the same 50 ns step,
this takes 46 ms and allows 21 frames/s, sufficient for a good quality video.
Of course, many methods can and are being used to improve performance, but
these simple and practical considerations give an idea of the issues involved. It
should be noted as well that much higher clock speeds are not very practical, as the
power consumption increases linearly with frequency, something that must also be
taken into considerations in small, battery-operated cameras.
There are cameras and imaging systems that may include hundreds of mega-
pixels. In these systems, extraction of the image may take considerable time, but
the quality and resolution are superior.
4.8 Thermal-based optical sensors

The thermal effects of radiation, that is, conversion of radiation into heat, are most
pronounced at lower frequencies (longer wavelengths) and are therefore most
useful in the IR portion of the spectrum. In effect, what is measured is the tem-
perature associated with radiation. The sensors based on these principles carry
different names—some traditional, some descriptive. Early sensors were known as
pyroelectric sensors (from the Greek pyr, “fire”). Bolometers (from the Greek
bolé, “ray,” so bolometer can be loosely translated as a ray meter) are also thermal
radiation sensors that may take various forms, but all include an absorbing element
and a temperature sensor in one form or another. Some are essentially thermistors
and can be used in the whole range of radiation, including microwave and milli-
meter wave measurements. Bolometers date back to 1878 and were originally
intended to sense low radiation levels from space. Other names like passive infra-
red (PIR) and active far infrared (AFIR) not only are more descriptive but also
broader, and encompass many types of sensors.
In effect, almost any temperature sensor may be used to measure radiation as
long as a mechanism can be found to transform radiation into heat. Since most
methods of temperature sensing were described in Chapter 3, we will discuss here
the specific arrangements used to sense radiation and will view the temperature-
sensing elements used in conjunction with various thermal radiation sensors as
given and known.
In general, thermal radiation sensors are divided into two classes: PIR
(including bolometers) and AFIR. In a passive sensor, radiation is absorbed and
converted to heat. The temperature increase is measured by a sensing element to
yield an indication of the radiative power. In an active sensor, the device is heated
from a power source and the variations in this power due to radiation (e.g., the
current needed to keep the temperature of the device constant) give an indication of
the radiation.
4.8.1 Passive IR sensors

A passive infrared sensor has two basic components: an absorption section that
converts radiation into heat and a proper temperature sensor that senses the tem-
perature rise due to this heat into an electrical signal. Without going into the
questions of heat transfer and heat capacities (these were discussed briefly in
Chapter 3), the absorption section of the sensor must be able to both absorb as much
of the incoming radiated power at the sensor’s surface as possible while at the same
time quickly respond to changes in radiated power density. Typically, the absorber
is made of a metal of good heat conductivity (gold is a common choice in high-

quality sensors) that is often blackened to increase absorption. The volume of the
absorber is kept small to improve response (quick heating/cooling) to changes in
radiation and hence keep the response time reasonable. The absorber and the sensor
are encapsulated or placed in a gas-filled or evacuated hermetic chamber to avoid
variations in sensing signals due to the cooling effects of air motion. The absorber
is located behind a transparent (to IR radiation) window, often made of silicon, but
other materials may be used (germanium, zinc selenide, etc.). The choice of the
sensor materials and structure dictates to a large extent the sensitivity, spectral
response, and physical construction of the device.
4.8.1.1 Thermopile PIR

In this type of device, sensing is done by a thermopile. A thermopile is made of a
number of thermocouples connected in series electrically but in parallel thermally
(i.e., they are exposed to identical thermal conditions). Based on the thermoelectric
effect, a thermocouple generates a small potential across a junction made of two
different materials. Any two materials can be used, but some material combinations
produce higher potential differences (see Section 3.3). Thermocouples can only
measure temperature differences, hence the thermopile is made of alternating
junctions, as shown in Figure 4.22. All “cold” junctions are held at a known
(measured) lower temperature and all “hot” junctions are held at the sensing tem-
perature. In practical construction, the cold junctions are placed on a relatively
large frame that has a high thermal capacity and hence the temperature will fluc-
tuate slowly while the hot junctions are in contact with the absorber, which is small
and has a low heat capacity (Figure 4.22). In addition, the frame may be cooled, or
a reference sensor may be used on the frame so that the temperature difference can
s Ab
on sor
ncti ber
ju
ld
Co
e
an
br
em
M
s
on
ncti
ju
ld
jun Hot Co
cti or
on
s ens
es
ur
Body erat
mp
Te
Figure 4.22 The structure of a PIR sensor showing the thermopile used to sense
temperature (under an IR absorber). A temperature sensor monitors
the temperature of the cold junctions
be properly monitored and related to the radiated power density at the sensor.
Although any pair of materials may be used, most PIRs use crystalline or poly-
crystalline silicon and aluminum because silicon has a very high thermoelectric
coefficient and is compatible with other components of the sensor, whereas alu-
minum has a low temperature coefficient and can be easily deposited on silicon
surfaces. Other materials used (mostly in the past) are bismuth and antimony. The
output of the thermocouple is the difference between the Seebeck coefficients of
silicon and aluminum (see Section 3.3).
PIR sensors are used to sense mostly near-IR radiation but within this range they
are quite common. When cooled, they can be used further down into the far-IR
radiation. One of the most common applications of PIR sensors is in motion detection
(in which the transient temperature caused by motion is detected). However, for this
purpose, the pyroelectric sensors in the following section are often used because they
are both simpler and less expensive than the structure described above.
Example 4.10: Thermopile sensor for IR detection

An IR sensor is used to detect hot spots in forests to prevent fires. The sensor is
made as in Figure 4.22, with 64 pairs of junctions using a silicon/aluminum junc-
tion (see Tables 3.3 and 3.9). The silicon/aluminum thermocouple has a sensitivity
S ¼ 446 mV/ C. The absorber is made of a thin gold foil, 10 mm thick and 2 cm2 in
area, coated black to increase absorption. Gold has a density r ¼ 19.25 g/cm3 and a
specific heat capacity Cs ¼ 0.129 J/g/ C (i.e., the absorber needs to absorb 0.129 J/g
for its temperature to rise 1 C). The absorber is not ideal and its conversion effi-
ciency is only 85% (i.e., 85% of the incoming heat is absorbed). We will denote
efficiency as e ¼ 0.85. The window of the sensor is A ¼ 2 cm2 and we will assume
that it takes the sensor t ¼ 200 ms to reach thermal steady state (i.e., for the tem-
perature of the absorber to stabilize to a constant value at a given radiation power
density). This temperature is due to the absorbed heat and heat loss. If a temperature
difference between the hot and cold junctions of 0.1 C can be reliably measured in
the sensor, calculate the sensitivity of the sensor assuming the input is the power
density of the IR radiation.
Solution: Because the input is a power density (i.e., measured in W/m2 or in lux)
and the temperature increase is due to heat (energy), the heat capacity is therefore
the product of power density and time. Given a power density Pin [Wm2], the
power received by the sensor is the product of power density and the area of the
absorber. With an 85% efficiency, the heat absorbed over 200 ms is
w ¼ Pin tAe ¼ Pin 0:2 2 104 0:85 ¼ 3:4 105 Pin ½J:
To find the temperature increase of the absorber due to this heat, we divide it
by the heat capacity of the foil, denoted as C. This is simply the specific heat
multiplied by the mass of the absorber. The latter is
m ¼ 10 106 2 104 19:25 ¼ 3:85 105 g:

The heat capacity for the absorber is therefore
Ca ¼ Cs m ¼ 0:129 3:85 105 ¼ 4:96665 106 ½J=K
We obtain the temperature increase in the absorber by dividing the incoming

heat by Ca:
w Pin tAe 3:4 105

T¼ ¼ ¼ Pin ¼ 6:846 Pin ½K:
Ca Ca 4:96665 106
Since the lowest measurable change in temperature is 0.1 C, we get the power
density needed to raise the temperature by that amount:
0:1
Pin ¼ ¼ 1:46 102 W=m2 :
6:846
This is 14.6 mW/m2.
The sensitivity, by definition (assuming, of course, a linear transfer function),
is the output divided by input. We have the input power density. Now, we need to
calculate the output of the thermocouple for that same temperature difference of
0.1 C. Since we have a thermopile with 64 pairs of junctions and a sensitivity of
446 mV/ C, the output for a temperature difference of 0.1 C is
Vout ¼ 446 64 0:1 ¼ 2;854:4 mV:
Thus, the sensitivity of the sensor is
Vout 2;854:4
s¼ ¼ 2
¼ 1:955 105 mV= W=m2 :
Pin 1:46 10
In practical terms, this means that, for example, the sensor will produce a
1.955 mV output at an input power density of 105 W/m2. This kind of sensitivity is
sufficient for most low-power sensing ranging from astronomy applications to
highly sensitive motion sensing.
4.8.1.2 Pyroelectric sensors

The pyroelectric effect is an electric charge generated in response to heat flow
through the body of a crystal. The charge is proportional to the change in tem-
perature and hence the effect may be viewed as heat flow sensing rather than
temperature sensing. However, in the context of this section, our interest is in the
measurement of radiation and thus pyroelectric sensors are best viewed as sensing
changes in radiation. For this reason, they have found applications in motion sen-
sing in which the background temperature is not important—only that due to the
motion of a “warm” source is sensed. Pyroelectricity was formally named in 1824
by David Brewster, but its existence in tourmaline crystals was described in 1717
by Louis Lemery. It is interesting to note that the effect is described in the writing
of Theophrastus in 314 BC as the attraction of bits of straw and ash to tourmaline
when the latter was heated. The attraction is due to the charge generated by the
heat. As early as the end of the nineteenth century, pyroelectric sensors were made of
Rochelle salt (potassium sodium tartrate [KHC4H4O6]). Currently, there are many
other materials used for this purpose, including barium titanate (BaTiO3), lead tita-
nate (PbTiO3), as well as lead zirconium titanate (PZT) materials (PbZrO3), polyvinyl
fluoride (PVF), and polyvinylidene fluoride (PVDF). When a pyroelectric material is
exposed to a temperature change DT, a charge DQ is generated as
DQ ¼ PQ ADT ½C; (4.25)
where A is the area of the sensor and PQ is the pyroelectric charge coefficient
defined as

dPs C
PQ ¼ ; (4.26)
dT m2 K
and Ps is the spontaneous polarization [C/m2] of the material. Spontaneous polar-
ization is a property of the material related to its electric permittivity.
A change in potential DV develops across the sensor as
DV ¼ PV hDT ½V; (4.27)
where h is the thickness of the crystal and PV is its pyroelectric voltage coefficient,

dE V
PV ¼ ; (4.28)
dT m2 K
and E is the electric field across the sensor. The two coefficients (voltage and charge
coefficients; see Table 4.4) are related through the permittivity of the material as
follows:
PQ dPs
¼ ¼ e0 er ½F=m: (4.29)
PV dE
By definition, the sensor’s capacitance is
DQ A
C¼ ¼ e0 er ½F: (4.30)
DV h
Table 4.4 Pyroelectric materials and some of their properties
Material PQ [C/(m2 K)] PV [V/(m K)] er Curie

temperature [ C]
TGS (single crystal) 3.5 104 1.3 106 30 49
LiTaO3 (single crystal) 2.0 104 0.5 106 45 618
BaTiO3 (ceramic) 4.0 104 0.05 106 1,000 120
PZT (ceramic) 4.2 104 0.03 106 1,600 340
PVDF (polymer) 0.4 104 0.4 106 12 205
PbTiO3 (polycrystalline) 2.3 104 0.13 106 200 470
TGS ¼ TriGlycine Sulfate.
Hence, one can write the change in voltage across the sensor as
h
DV ¼ PQ DT ½V: (4.31)
e0 er
Clearly, this change in voltage is linearly proportional to the change in
temperature. It should be noted again that our main interest here is not in mea-
suring the change in temperature, but rather the change in radiation that causes
this change in temperature. Also to be noted is that all sensors must operate below
their Curie temperature (at the Curie temperature their polarization vanishes).
Table 4.4 shows these properties for some materials commonly used for pyro-
electric sensors.
The structure of a pyroelectric sensor is quite simple. It consists of a thin crystal
of a pyroelectric material between two electrodes, as shown in Figure 4.23(a).
Some sensors use a dual element, as in Figure 4.23(b). The second element can be
used as a reference by, for example, shielding it from radiation, and it is often used
to compensate for common mode effects such as vibrations or very rapid thermal
changes, which can cause false effects. In Figure 4.23(b) the two elements are
connected in series, but they may also be connected in parallel.
The most common materials in pyroelectric sensors are triglycine sulfate
(TGS) and lithium tantalite crystals, but ceramic materials and, more recently,
polymeric materials are also commonly used.
In applications of motion detection, especially of the human body (sometimes
of animals), the change in temperature of IR radiation (between 4 mm and 20 mm)
causes a change in the voltage across the sensor, which is then used to activate a
switch or some other type of indicator.
An important property of all pyroelectric sensors is the decay time, during
which the charge on the electrodes diffuses. This is on the order of 1–2 s because of
the very high resistance of the materials; but, it also depends on the external con-
nection of the device. This response time is very important in the ability of the
sensors to detect slow motion.
Figure 4.24 shows a dual IR sensor used for motion detection. This device
includes a differential amplifier, operates at 3–10 V, and has a field of view of 138
horizontally (wide dimension of the window) and 125 vertically. The device has
an optical bandwidth (sensitivity region) between 7 mm and 14 mm (in the near IR
region).
Pyroelectric material Pyroelectric material Pyroelectric material
(a) (b)
Figure 4.23 The basic structure of a pyroelectric sensor. (a) Single element.
(b) Dual element in series connected in a differential mode
Figure 4.24 A PIR motion detection sensor. This is a dual sensor. Note the metal
package and the window (4 mm 3 mm)
Example 4.11: Motion sensor

A motion sensor based on a PZT ceramic is used to turn on lights in a room as a
person enters the room. The sensor is made of two conducting plates with a PZT chip
(8 mm wide, 10 mm long, and 0.1 mm thick) between them, forming a capacitor. One
plate is exposed to the motion, whereas the other is connected to the body of the
sensor and held at its temperature. As the person enters the room, the person’s body
temperature causes the exposed plate’s temperature to temporarily rise by 0.01 C
because of the IR radiation produced by the body. This temperature dissipates and
eventually both plates will reach the same steady-state temperature. For this reason,
the sensor can detect motion but not presence. Calculate the charge produced on the
plate and the potential difference across the sensor due to the rise in temperature.
Solution: The charge produced by the rise in temperature can be calculated from
(4.25) and the potential difference can be calculated from (4.31). However, we will
start with (4.31), calculating the change in voltage and then the change in charge
from the relation between charge and capacitance (4.30).
The change in voltage across the plates is
h 0:1 103
DV ¼ PQ DT ¼ 4:2 104 0:01
e0 er 1;600 8:854 1012
¼ 0:0296 V:
This is a small voltage, but because the reference voltage (i.e., the output in the
absence of motion) is zero, the small output voltage is easily measurable.
The charge produced by the change in temperature depends on the capacitance.
The latter is
e0 er A 1;600 8:854 1012 0:008 0:01

C¼ ¼ ¼ 1:1333 108 F:
h 0:1 103
The charge produced is

DQ ¼ CDV ¼ 1:1333 108 0:0296 ¼ 3:355 1010 C:
Clearly, to produce a useful output (i.e., to turn on a relay or an electronic
switch), the output must be amplified. For example, if the output requires (typi-
cally) 5 V, amplifying the voltage by about 170 will produce the required output.
We will discuss these issues in Chapter 11, where we will see that in fact the
amplification can be done using a charge amplifier. The main reason for that
approach, rather than performing a classical voltage amplification, is that the
impedance of pyroelectric sensors is very high, whereas the input impedance of
conventional voltage amplifiers is much lower. The charge amplifier, which has
very high input impedance, is better suited for this application.
4.8.1.3 Bolometers
Bolometers are very simple radiation power sensors useful over the whole spectrum
of electromagnetic radiation, but they are most commonly used in microwave and
far-IR ranges. They consist of any temperature-measuring device, but usually a
small RTD or a thermistor is used. The radiation is absorbed by the device directly,
causing a change in its temperature. This temperature increase is proportional to the
radiated power density at the location of sensing. This change causes a change in
the resistance of the sensing element that is then related to the power or power
density at the location being sensed. Although there are many variations of the
basic device, they all operate on essentially the same principle. However, since the
temperature increase due to radiation is the measured quantity, it is important that
the background temperature (i.e., air) be taken into account. This can be done by
separate measurements or by a second bolometer that is shielded from the radiation
(usually, in a metal enclosure in the case of microwaves). The sensitivity of a
bolometer to radiation can be written as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
αes ZT R0 DT
b¼ ; (4.32)
2 ð1 þ α0 DTÞ½1 þ ðwtÞ2
where α ¼ (dR/dT)/R is the temperature coefficient of resistance (TCR) of the

bolometer, es is its surface emissivity, ZT is the thermal resistance of the bolometer,
R0 is its resistance at the background temperature, w is the frequency, t is the
thermal time constant, and DT is the increase in temperature. Clearly then, the ideal
bolometer should have a large resistance at background temperatures and high
thermal resistance. On the other hand, they must be small physically, which favors
low thermal resistances.
In terms of construction, bolometers are fabricated as very small thermistors or
RTDs, usually as individual components or as integrated devices. In all cases, it is
important to insulate the sensing element from the structure supporting it so that
its thermal impedance is high. This can be done by simple suspension of the sensor by its
lead wires or, as is sometimes done, by suspension of the sensor over a silicon groove.
Equation (4.32) is rather complex, taking into account all parameters involved
in sensing. Analysis of bolometers is often simpler than this although it involves the
calculation of absorbed energy and rise in temperature due to this energy as was
done in Example 4.11. The very first bolometer, invented by Samuel Langley in
1878, was made of two thin platinum strips (later replaced with iron strips) coated
with carbon black to increase heat absorption, one strip exposed to radiation, the
second shielded from it. The change in resistance due to heat absorption was
measured to obtain sensitivities of 105 C. Langley used this bolometer to mea-
sure cosmic electromagnetic radiation. Similarly, a small thermistor, coated with an
absorbing layer can serve as a reasonably sensitive bolometer. Microbolometers
and arrays of bolometers also exist and used in many IR cameras.
Bolometers are some of the oldest devices used for the measurement of radi-
ated power and have been adapted for use in many applications in the microwave
region, including mapping of antenna radiation patterns, detection of IR radiation,
testing of microwave devices, and much more.
4.9 Active far infrared (AFIR) sensors
In its simplest form an AFIR sensor can be thought of as a power source that heats
the sensing element to a temperature above ambient and keeps its temperature
constant. When used to sense radiation, additional heat is provided to the sensor
through this radiation. The power supplied to the sensor must be reduced to keep
the temperature constant and the difference in power is a measure of the radiation
power. In practice, the process is more complicated. The power supplied to the
sensor through an electric circuit that heats it up to a constant temperature Ts is
P ¼ PL þ F ½W; (4.33)
2
where P ¼ V /R is the heat supplied by a resistive heater (V is the voltage across the
heating element and R its resistance) and F is the radiation power being sensed. PL
is the power loss, mostly due to conduction through the body of the sensor:
PL ¼ αs ðTs Ta Þ ½W; (4.34)
where αs is a loss coefficient or thermal conductivity (which depends on materials
and construction), Ts is the sensor’s temperature, and Ta is the ambient temperature.
Given the power supplied as P ¼ V2/R, a sensor with surface area A, total emis-
sivity e and electric conductivity s, the sensed temperature, Tm, is
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ffi
4 1 V
Tm ¼ Ts4 αs ðTs Ta Þ ½ C: (4.35)
Ase R
By measuring the voltage across the heating element, a reading of the radiation
power is readily obtained. If Ts and Ta are given in C, Tm is also in C. Alternatively,
the three temperatures may be given in K.
Although AFIR devices are much more complex than simple PIRs, including
bolometers, they have the advantage of a much higher sensitivity and a degree of
independence from thermal noise that other IR sensors do not possess. This is
particularly true at low ambient temperatures. Hence, AFIR devices are used for
low-contrast radiation measurements where PIRs are not suitable.
4.10 Optical actuators

When discussing optical devices, it is not immediately clear what an optical
actuator might be because of our tendency to think of actuators as devices that
perform some kind of motion. However, based on our definition of actuators, in fact
there are many optical actuators and they are quite common. The use of a laser
beam to perform eye surgery, to machine a material, or to record data in a magneto-
optical hard drive are some examples. Others are transmission of data on an optical
fiber, transmission of a command using an IR remote control device, use of an LED
or laser to illuminate a CD to read the data, scanning of an IPC code in a super-
market, or even turning on a light in a room. In Chapter 10, we will discuss addi-
tional examples such as optical switching.
Optical actuation can be low or high power. In an optical link such as in optical
fiber communication or an optical isolator (to be discussed in Chapter 12), the
transmitting element (the actuator) is a low-power LED in the IR or visible range
(Figure 4.25). The power produced by the LED may only be a few milliwatts. On
the other hand, industrial lasers, such as carbon dioxide (CO2) lasers (in which CO2
gas is excited and produces a beam in the IR region around 10 mm), can produce
hundreds of kilowatts of useful power for a variety of industrial processing pur-
poses, including machining, surface treatment, and welding. In between are lasers,
mostly CO2 lasers of moderate power (a few watts to a few hundred watts) used for
medical applications, including surgery, skin ablation, and suturing. Other appli-
cations are for range finding, particularly in the military, and for speed detection
and measurement. Laser actuation is also used for the production of electronic
components, for trimming of devices, and even for recording of data on CD-ROMs,
where they actually scribe the surface with the pattern representing the data.
A good example of an optical actuator of the type described here is the
magneto-optical recording of data, a common method used for high-density data
recording. The principle is shown schematically in Figure 4.26 and is based on two
principles. First, a laser beam, focused to a small point, can heat the surface of the
disk to a high temperature in a few nanoseconds. Second, when a ferromagnetic
material such as iron or its oxides is heated above a certain temperature (about
Sheath
LED
Optical fiber
Figure 4.25 An optical link. The fluctuations in the light intensity of the LED
represent the data transmitted along the optical fiber
Ferromagnetic oxide coating

Laser
Aluminum disk
Magnetic recording/reading head

Spindle
Figure 4.26 Magneto-optical recording. The laser beam heats the recording
medium above the Curie temperature and the magnetic recording
head applies the magnetic field needed to record the data
650 C), the material loses its magnetic properties. This temperature is called the
Curie temperature and is characteristic of the particular material used as the
recording medium (mostly Fe2O3). When cooled, the material becomes magnetized
with the field supplied by the recording head. To record data, the laser is turned on
to heat the point above the Curie temperature and the datum that needs to be
recorded at that point is supplied in the form of a low-intensity magnetic field by a
magnetic recording head. The beam is then switched off and the spot cools below
the Curie temperature in the presence of the magnetic field, retaining the data
permanently. Erasure of data is done by heating the spot and cooling it off without a
magnetic field. The data are read using the magnetic recording head alone. The
advantage of this method is that the data density is much higher than purely mag-
netic recording, which requires larger magnetic fields that in turn extend over larger
surfaces and hence is only practical at lower data densities.
4.11 Problems
Optical units
4.1 Optical quantities. An isotropic light source (a source that radiates uni-
formly in all directions in space) produces a luminance of 0.1 cd/m2 at a
distance of 10 m. What is the power of the source?
4.2 Optical sensitivity. Many optical instruments such as video cameras are
rated in terms of sensitivity in lux, especially to indicate low-light sensi-
tivity. One may encounter specifications such as “sensitivity: 0.01 lux.”
What is the sensitivity in terms of power density?
4.3 Electron-volts and joules. The Planck constant is given as 6.6261 1034 Js
or as 4.1357 1015 eV s. Show that the two quantities are identical but
expressed in different units.
The photoelectric effect
4.4 Photon energy and electron kinetic energy. A photoelectric device
intended for UV sensing is made of a platinum cathode. Calculate the range
of kinetic energies of the electrons emitted by UV light between 400 nm and
1 pm, assuming that a photon emits a single electron.
4.5 Work function and the photoelectric effect. The work function for copper
is 4.46 eV.
(a) Calculate the lowest photon frequency that can emit an electron from
copper.
(b) What is the wavelength of the photons and in what range of optical
radiation does that occur?
4.6 Electron density in a photoelectric sensor. A photoelectric sensor has a
cathode in the form of a disk of radius a ¼ 2 cm coated with an alkali
compound that has a work function e0 ¼ 1.2 eV and a quantum efficiency of
15%. The sensor is exposed to sunlight. Calculate the average number of
electrons emitted per second assuming a power density of 1,200 W/m2 and
uniform distribution of power over the spectrum between red (700 nm) and
violet (400 nm).
4.7 Work function, kinetic energy, and current in a photoelectric sensor.
A photoelectric sensor with unknown cathode material is subjected to
experimental evaluation. The current in the sensor (see Figure 4.17) is
measured while the wavelength of the radiation is recorded. Emission is
observed starting with IR radiation at 1,150 nm. The power density of the
incoming radiation is kept constant at 50 mW/cm2.
(a) Calculate the work function of the cathode.
(b) If the photoelectric sensor is now illuminated with the same power
density, but with a blue light at a 480 nm wavelength, what is the
kinetic energy of the electrons released?
(c) Assume each photon releases one electron. Calculate the current in the
sensor if the cathode has an area 2.5 cm2.
The photoconducting effect and photoconducting sensors
4.8 Bandgap energy and spectral response. A semiconductor optical sensor is
required to respond down to 1,400 nm for use as a near-IR sensor. What is
the range of bandgap energies of the semiconductor that can be used for this
purpose?
4.9 Germanium silicon and gallium arsenide photoconducting sensors. The
intrinsic concentrations and mobilities for germanium, silicon, and gallium
arsenide at 298 K are as follows:
Germanium Silicon Gallium arsenide

(Ge) (Si) (GaAs)
Intrinsic concentration, ni 2.4 1013 1.45 1010 1.79 106
[per cm3]
Mobility of electrons, me 3,900 1,500 8,500
[cm2/(V s)]
Mobility of holes, mp 1,900 450 400
[cm2/(V s)]
Compare the dark resistance of an identical sensor for the three materials
made as a rectangular bar of length 2 mm, width 0.2 mm, and thickness
0.1 mm to be used as photoconductors. The resistance calculated here is the
nominal resistance of the sensor.
4.10 Gallium arsenide photoconductive sensor. A gallium arsenide (GaAs)
photoconductive sensor is made as a small, rectangular chip 2.5 mm long,
2 mm wide, and 0.1 mm thick (see Figure 4.4(a), for the construction). A red
light of intensity 10 mW/cm2 and wavelength 680 nm is incident perpen-
dicularly on the top surface. The semiconductor is n-type with an electron
concentration of 1.1 1019 electrons/m3. The mobilities of electrons and
holes in GaAs are 8,500 cm2/(Vs) and 400 cm2/(Vs), respectively. Assuming
all incident power on the top surface of the device is absorbed and a quantum
efficiency of 0.38, calculate
(a) The “dark” resistance of the sensor.
(b) The resistance of the sensor when light shines on it. The recombination
time of electrons is approximately 10 ms.
4.11 Improved gallium arsenide photoconductive sensor. In an attempt to
improve the sensor described in Problem 4.10, the meander shape in
Figure 4.4(b) is adopted, keeping the total exposed area the same (5 mm2)
but reducing the length between the electrodes to 0.5 mm.
(a) Calculate the “dark” resistance of the sensor.
(b) Calculate the resistance of the sensor and its sensitivity when light
shines on it.
(c) In an attempt to improve performance, the length between electrodes is
reduced to 0.25 mm but the total exposed are is reduced to 3.5 mm2 to
accommodate the additional electrode area needed. Calculate the sen-
sitivity of the sensor and compare to that in (b).
(d) Compare the results in (a) through (c) with those for the rectangular
sensor and comment on the sensitivity of the two devices.
4.12 Intrinsic silicon optical sensor. Suppose, a photoconductive sensor is made
of intrinsic silicon with structure and dimensions as shown in Figure 4.27.
The intrinsic concentration is 1.5 1010 carriers/cm3 and mobilities of
electrons and holes are 1,350 cm2/(V s) and 450 cm2/(V s), respectively.
The carrier lifetime for electrons and holes depends on the concentration and
changes with illumination, but for simplicity we will assume these are
constant at 10 ms. Assume also that 50% of the incident power is absorbed
by the silicon and the sensor has 45% quantum efficiency.
(a) Find the sensitivity of the device to input power density at a given
wavelength in general terms.
(b) What is the sensitivity at 1 mW/cm2 at a wavelength of 480 nm?
(c) What is the cutoff wavelength, that is, the wavelength beyond which
the sensor cannot be used?
1 mm
Exposed
Electrodes
surface 1 mm
0.1 mm
15 mm
15 mm
(a) (b)
Figure 4.27 A photoconducting sensor
Photodiodes
4.13 Photodiode in photoconductive mode. A photodiode is connected in
reverse mode with a small reverse voltage to ensure low reverse current. The
leakage current is 40 nA and the sensor operates at 20 C. The junction has
an area of 1 mm2 and operates from a 3 V source, as shown in Figure 4.28.
The resistor is 240 W. Calculate the voltage across R in the dark and when
illuminated with a red laser beam (800 nm) with power density of 5 mW/cm2.
Assume a quantum efficiency of 50%.
R V=?
V0
Figure 4.28 A photodiode in photoconductive mode
4.14 Photodiode in forward bias. A photodiode is forward biased as shown in

Figure 4.29 so that the voltage across the diode is 0.2 V. The dark current
through the diode is 10 nA.
(a) What must be the voltage V0 to obtain this bias at 20 C in the dark?
(b) When illuminated with a certain power density, and with the voltage
V0 found in (a) applied, the bias changes to 0.18 V. Calculate the total
power absorbed by the diode at 800 nm.
R = 10 kΩ
V0
+
Figure 4.29 Forward-biased photodiode

4.15 Dusk/dawn light switch. Many lighting systems, including street lightning,
turn on and off automatically based on light intensity. To do so, it is pro-
posed to use a photodiode in the configuration shown in Figure 4.30. The
diode has negligible dark current, an exposed area of 1 mm2, and a quantum
efficiency of 35%. The electronic switch is designed so that to turn on the
lights the voltage across R must be 8 V or less and to turn off the lights the
voltage must be 12 V or greater. On a normal sunny day, the power density
available at ground level is 1,200 W/m2.
(a) Calculate the resistance R so that the lights turn on when available
power density (in the evening) is 10% (or less) of the normal daylight.
Assume radiation at an average wavelength of 550 nm.
(b) At what power density will the lights turn off in the morning?
(c) Repeat (a) and (b) if it is known that in the evening the average light
wavelength tends to be more red, with an average wavelength of
580 nm, and in the morning it tends to be more blue, with an average
wavelength of 520 nm.
Load
+
V0
Electronic
R switch
Mains
Figure 4.30 A light-activated switch
Photovoltaic diodes
4.16 Solar cells as actuators: power generation. The use of solar cells for
power generation is common in small-scale installations and for use in
stand-alone equipment such as remote sensors and monitoring stations. To
get some idea of what is involved, consider a solar cell panel with an overall
power conversion efficiency of 30% (i.e., 30% of the power available at the
surface of the solar cell is converted into electrical power). The panel is
80 cm 100 cm in area and the maximum solar power density at the
location is 1,200 W/m2. The panel is divided into 40 equal-size cells and the
cells are connected in series. Assume that the cells are equally responsive
over the entire visible spectrum (400–700 nm), have a quantum efficiency of
50%, and the internal resistance of the 40 cells in series is 10 W. The leakage
current of each cell is 50 nA. Calculate the maximum power the solar cell
can deliver and indicate what the conditions must be for that to happen.
4.17 Overall efficiency of solar cells. A solar panel supplies a current of 0.8 A
into a 10 W load when exposed to the sun under optimal conditions (i.e., the
sun’s radiation is vertical, power density is maximal) at a location where the
sun’s radiation intensity is 1,400 W/m2. The panel is made of cells, each
10 cm 10 cm, and the cells are connected in series. Use the average
wavelength in the visible range of 550 nm as the wavelength of radiation.
(a) Calculate the overall conversion efficiency of the solar cells under the
stated conditions.
(b) What is the maximum power the cell can deliver into a 10 W load if its
overall efficiency can be increased to 30%? Assume the internal
resistance is 10 W.
4.18 Solar cell as a light power density sensor. A simple light sensor to sense
background illuminance is made of a small solar cell in which the open
circuit voltage is measured. The solar cell has an average quantum spectral
efficiency of 80% over the spectrum between 700 (red) and 400 nm (violet)
and an exposed area of 2 cm2. The dark current is 25 nA and the efficiency
constant is n ¼ 1. Calculate at room temperature (25 C):
(a) The no-load voltage of the solar cell in the middle of the spectrum
(550 nm).
(b) The sensitivity of the sensor to power density for red and violet lights.
4.19 Photovoltaic temperature sensor. A photovoltaic cell can be used to sense
temperature as follows: a small silicon solar cell is illuminated with a blue
LED operating at a wavelength of 450 nm. The open voltage is measured as
an indication of temperature. The cell has a quantum spectral efficiency of
75%, an efficiency constant (ideality factor) of 2 and a dark current of 25 nA.
The LED’s output is 28 lumens. Because of the pattern of light radiation from
the LED and reflections from the cell and its structure, only 64% of the light
output power reaches the surface of the sensor. Calculate and plot:
(a) The cell’s output voltage as a function of temperature.
(b) The sensitivity of the sensor.
(c) What is the useful span of the device as a temperature sensor?
Phototransistors
4.20 Phototransistor as a detector. Consider again Example 4.4, but now the
photodiode is replaced with a phototransistor biased as shown in Figure 4.31.
The phototransistor has a gain of 50. For a given input pulse train, show the
output in relation to the input and calculate the voltage levels expected.
Assume that all power that reaches the phototransistor is absorbed in the
base–emitter junction.
+5 V
R 100 kΩ
Laser Optical fiber V0
16 km
Figure 4.31 Phototransistor as a detector in an optical link

4.21 Phototransistor and saturation current. The phototransistor in Figure 4.32

operates in its linear range. At an input power density of 1 mW/cm2, the
voltage measured between the collector and emitter is Vce ¼ 10.5 V.
Calculate the span of the phototransistor as a light intensity sensor. That is,
calculate the minimum power density (for which the current through the
transistor is zero) and maximum power density (at which the current through
the transistor saturates). The saturation Vce voltage is 0.1 V. Neglect the effect
of the dark current.
c 12 V
Vce
Figure 4.32 A phototransistor and its operation
4.22 Temperature effects on phototransistors. Consider again the configuration

in Figure 4.32, where R ¼ 1 kW and the amplification (gain) of the transistor
is 100. The collector–emitter voltage at a light intensity of 1 mW/cm2 is equal
to 8 V at 20 C. If the light is removed, the collector–emitter voltage climbs
up to 11.8 V. Calculate the collector–emitter voltage if the temperature rises
to 50 C. Assume that Vbe does not change with temperature or with light
intensity and the dark current is 10 nA. Discuss the result.
Note: Vbe decreases with temperature at a rate of 1.0–2.0 mV/ C (see
Section 3.4), but we will neglect this change here.
Photoelectric sensors and photomultipliers
4.23 Current and electron velocity in a photomultiplier. To get some idea of
the processes occurring in a photomultiplier, consider the following sim-
plified configuration. A circular cathode and a circular anode each of radius
a ¼ 20 mm are separated d ¼ 40 mm apart and connected to a potential
difference V ¼ 100 V as shown in Figure 4.33. The cathode is made of a
potassium-based compound with a work function e0 ¼ 1.6 eV and a quan-
tum efficiency h ¼ 18%.
(a) If blue light at a wavelength of 475 nm and intensity of 100 mW/cm2
impinges on the cathode, calculate the current expected in the device.
(b) Calculate the velocity of the electrons when they reach the anode,
given the mass of the electron me ¼ 9.1094 1031 kg.
Anode a
Radiation
V
d
−
− a
−
Cathode −
Figure 4.33 A basic photomultiplier
4.24 Limit sensitivity of a photomultiplier in the visible range. Given a pho-

tomultiplier, its highest sensitivity occurs at the shorter wavelengths.
Assume that the wavelength is 400 nm (violet light). If the cathode has a
work function e0 ¼ 1.2 eV and quantum efficiency of 20%, calculate the
lowest illuminance the photomultiplier can discern if detection requires
emission of at least 10 electrons/mm2 of the cathode area per unit time.
CCD sensors and detectors
4.25 Digital video camera image transfer from CCD. A digital color video
camera requires a picture format of 680 pixels 620 pixels for display on a
TV screen. Assuming that smooth video requires 25 frames/s (PAL or
SECAM formats), calculate the minimum clock frequency for the stepping
process necessary to retrieve the image from the CCD. Neglect the time
needed to create the image.
4.26 CCD sensor for HD video image transfer. The CCD sensor in an HD
video recorder is arranged as 1,080 lines with 1,920 pixels/line and produces
48 frames/s. The sensor uses an interline transfer method in which alter-
nating columns are transferred while the other columns are exposed.
(a) Calculate the minimum clock frequency needed to transfer the image
assuming the basic process in Figure 4.21(b) is used for the transfer.
(b) What is the effective frame rate?
Bolometer
4.27 Direct sensing bolometer. Samuel Langley in 1880 used a bolometer made
of two iron strips each 7 mm long, 0.177 mm wide, and 0.004 mm thick.
Unlike later bolometers, which use a separate absorber and a temperature
sensor, the resistance of the strips was measured directly, with one strip
exposed to radiation and the second kept at ambient temperature. The dif-
ference in resistance between the strips was used as an indication of tem-
perature. The bolometer described above is used to measure IR radiation
from a star. At an ambient temperature of 25 C, the smallest resistance
difference between the strips measurable is 0.001 W. Calculate the lowest
power density of radiation measurable. Assume absorption efficiency of
85% and a time constant of 0.8 s. Properties of iron are: conductivity:
1.0 107 S/m, TCR ¼ 0.0065/ C, density: 7.86 g/cm3, specific heat capa-
city: 0.46 J/g/K.
Thermopile PIR sensors

4.28 Thermopile PIR sensor. An IR sensor designed to operate at high tem-
peratures is made as in Figure 4.22 with 32 pairs of carbon/silicon carbide
junctions (see Chapter 3, Table 3.3). The carbon/silicon carbide thermo-
couple has a sensitivity S ¼ 170 mV/ C. The absorber is made of thin
tungsten foil, 10 mm thick and 2 cm2 in area, coated black to increase
absorption. Tungsten has a density of 19.25 g/cm3 (same as gold) and a
specific heat capacity Cs ¼ 24.27 J/mol/K. The absorber’s conversion effi-
ciency is 80%. The window of the sensor is A ¼ 5 cm2 and it takes the sensor
t ¼ 300 ms to reach thermal steady state, that is, for the temperature of the
absorber to stabilize to a constant value at a given radiation power density.
If a temperature difference between the hot and cold junctions of 0.5 C can
be reliably measured in the sensor, calculate the sensitivity of the sensor.
4.29 Thermopile IR sensor. An IR sensor is required to develop an output of 5 mV
for an IR radiation of 20 mW/cm2. For increased sensitivity, aluminum/silicon
thermocouples are used because they have an output of 446 mV/ C. An alu-
minum absorber of area 4 cm2 and thickness 20 mm is used with an absorption
efficiency of 80%. Aluminum has a density of 2.712 g/cm3 and a specific heat
capacity of 0.897 J/g/K. The required output must be obtained within 100 ms.
(a) What is the increase in the temperature of the absorber?
(b) Calculate the number of thermocouples needed to obtain the required
output.
(c) If the output of the device is measured using a digital voltmeter with a
resolution of 100 mV, what is the effective resolution of the sensor?
Pyroelectric sensors
4.30 Pyroelectric motion sensor. A PZT motion sensor is required to detect the
motion of a body within its range. The sensor is built as a dual element (see
Figure 4.23) to reduce the influence of temperature changes that are not due
to the motion of a body.
(a) Calculate the sensitivity if each element is 0.1 mm thick. One element
is exposed to the IR source, the other is shielded from it.
(b) Show how temperature compensation for common-source heat (heat
sources that affect both elements equally) is accomplished.
4.31 Time constant in motion sensors. A pyroelectric sensor is made of a small
barium titanate (BaTiO3) chip, 10 mm 10 mm in area and 0.2 mm thick,
sandwiched between two metal electrodes. In addition to the properties
given in Table 4.4, barium titanate has a conductivity of 2.5 109 S/m.
(a) Calculate the time constant of charge decay after heat has been removed.
(b) The sensor is used as a motion sensor to automatically trigger a camera
for wildlife photography at night. A cat runs by the sensor, generating
a temperature differential of 0.1 C, triggering the camera. If at least
half the charge generated across the plates must discharge before the
sensor can retrigger the camera, how much time is needed before the
next event can be detected?
(c) To discharge the sensor faster, one can connect a resistor across the
sensor. If the sensor must be ready to trigger within 250 ms, what must
be the resistance connected across the sensor? What is the side effect
of connecting a smaller resistor across the sensor other than a quicker
retrigger time?
Optical actuators
4.32 Coupling of power in an optical link. Optical links are very common in
data communication (see Figure 4.10 and Problem 4.20). However, linking
optical power to optical fibers can be a very low efficiency affair if not done
properly. Consider two ways of coupling the power from an LED to an
optical fiber, shown in Figure 4.34. In Figure 4.34(a), the optical fiber is
simply held in front of the LED, whereas in Figure 4.34(b) an intervening
light guide is used.
(a) If the LED radiates 10 mW uniformly over a 5 cone and the optical
fiber has a diameter of 130 mm, calculate the power coupled to the
optical fiber using the method in Figure 4.34(a). Neglect any reflec-
tions that may occur at the interface between the optical fiber and air.
The distance between the LED source and surface of the fiber is 5 mm.
(b) How much power is coupled in Figure 4.34(b)? Assume that all power
follows the light guide with uniform power density across the light
guide cross-section and none can escape through its outer surface. The
light guide has circular cross-section as does the fiber.
(c) Suppose, the LED is replaced with a laser with the same power. The
laser beam is collimated (the cross-section of the beam remains con-
stant as light propagates), is 150 mm in diameter with uniform power
density in the cross-section of the beam. How much power is coupled
in the two configurations in Figure 4.34? Compare it with the results in
(a) and (b).
Light guide
LED 5° 130 μm 5° 130 μm
LED
Optical 200 μm
fiber
(a) 5 mm (b) 5 mm Optical fiber
Figure 4.34 Coupling light to optical fibers. (a) Direct coupling. (b) Use of a light
guide to increase coupled power
4.33 Magneto-optical recording. In a hard disk storage device, writing of data is

done by magneto-optical means. Writing is done by heating the location
where the data are written to the Curie temperature of the storage medium
and applying the magnetic field representing the data to that spot while it
cools below the Curie temperature (see Figure 4.26). The data are written on
iron oxide (Fe2O3) coated on a conducting disk. Assume that 80% of the
writing time is needed to heat the spot. The laser beam is 1 mm in diameter
and supplies a power of 50 mW. The storage medium is 100 nm thick and
has a heat capacity of 23.5 J/mol/K and a density of 5.242 g/cm3. The curie
temperature is 725 C.
(a) Calculate the maximum writing data rate of the drive at an ambient
temperature of 30 C.
(b) Discuss possible ways the data rate can be increased.
(c) Discuss effects that will reduce the data rate in practical applications.
4.34 Laser skin ablation. A process used to resurface skin for cosmetic purposes
makes use a laser to ablate the surface of the skin to a small depth. The
typical process uses a pulsed IR laser. Consider the following: a laser
operating in the IR region (typically, at a wavelength of 2,940 nm) delivers
energy to the skin in pulses 250 ms wide, which then vaporize the water
within the tissue. The laser beam is 0.7 mm in diameter and penetrates
10 mm into the skin. Assuming all power is absorbed by the skin, and the
pulse width is too short for heat to conduct into neighboring tissue, calculate
the power of the laser beam and the energy absorbed by the skin in each
pulse. Skin tissue contains 64% water and its normal temperature is 34 C.
Specific heat of water is 4.187 J/g/K and the latent heat of evaporation is
2,256 J/g.

Optical Sensors and Actuators: The Eye

Uploaded by

Copyright:

Available Formats

Optical Sensors and Actuators: The Eye

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Optical Sensors and Actuators: The Eye

Uploaded by

Copyright:

Available Formats

Chapter 4

Optical sensors and actuators

Microwaves Infrared Ultraviolet X-rays

700 nm (red) 400 nm (violet)

Figure 4.1 Spectrum of infrared, visible, and ultraviolet radiation

4.2 Optical units

Table 4.1 Optical quantities and their units

Quantity Name Unit Derived unit Comments

Example 4.1: Conversion of optical units

4.4 Effects of optical radiation

4.4.2 Quantum effects

Table 4.2 Work functions for selected materials

Material Work Material Work

Example 4.2: Longest wavelength for photoelectric emission

ch 3 108 4:1357 1015

4.4.2.2 Quantum effects: the photoconducting effect

radiation intensity (visible light, UV radiation, and, to a lesser extent, IR radia-

(a) hf >Wbg (b) hf < Wbg

Material Band gap Longest wavelength Working

4.4.2.3 Spectral sensitivity

4.4.2.4 Tunneling effect

Figure 4.3 Spectral sensitivity of a semiconductor

4.5 Quantum-based optical sensors

4.5.1 Photoconducting sensors

Figure 4.4 Structure of a photoconductive sensor. (a) Simple electrodes.

Example 4.3: Properties of a photoresistor

s ¼ eðme n þ mp pÞ ¼ 1:602 1019 ð210 1016 þ 20 1016 Þ

Thus, we can rewrite (4.7) as

[mA] Forward current

Figure 4.7 A photodiode connected in the photoconductive mode (reverse biased).

Id ¼ I0 ðeeVd =nkT 1Þ ½A (4.12)

Figure 4.8 Various structures of photodiodes. (a) Common planar structure.

Example 4.4: Photodiode as a detector for fiber-optic communication

P ¼ 10 103 W ! P ¼ 10 log10 ð10 103 Þ ¼ 20 dB:

10 log10 P0 ¼ 58:4 ! log10 P0 ¼ 5:84 ! P0 ¼ 106:84

V0 ¼ 10 109 1 106 ¼ 10 mV:

V0 ¼ 946:3 109 1 106 ¼ 946:3 mV:

4.5.3 Photovoltaic diodes

Figure 4.11 (a) Photodiode connected in the photovoltaic mode—equivalent

Example 4.5: Properties of a solar cell at low light

Example 4.6: Optical position sensor

To calculate x and x0 , we note the following:

Top view Side view

The power reaching the cell is

Figure 4.14 An npn phototransistor. (a) Schematic structure and junctions.

Figure 4.15 (a) The I–V characteristics of a transistor as a function of base

Example 4.7: Sensitivity of a phototransistor

Figure 4.16 Collector current in a phototransistor as a function of input light

4.6 Photoelectric sensors

4.6.1 The photoelectric sensor

Figure 4.17 Photoelectric sensor and biasing circuit

I0 ¼ aAT 2 eE0 =kT ½A; (4.24)

where a is a constant depending on the cathode material, generally around 0.5, A is

These devices have many disadvantages, including problems with noise, as

Example 4.8: Thermionic noise in a photomultiplier

I0 ¼ aAT 2 eE0 =kT ¼ 0:5 120:173 104 ð298:15Þ2 e1:6=8:6210

G ¼ nk ¼ 610 ¼ 6:05 107 :

Ia ¼ 4:9 1017 6:05 107 ¼ 2:96 109 A:

4.7 Charge coupled (CCD) sensors and detectors

Id ¼ I0 ðeeVd =nkT 1Þ ½A (4.12)

I0 ¼ aAT 2 eE0 =kT ½A; (4.24)

Ca ¼ Cs m ¼ 0:129 3:85 105 ¼ 4:96665 106 ½J=K