July 1961 Electronics World
People old and young
enjoy waxing nostalgic about and learning some of the history of early electronics. Electronics World
was published from May 1959 through December 1971. See all
Electronics World articles.
thermoelectric coolers have been around for so long now
that most people probably assume there is nothing wondrous about
the discovery that makes them possible. I still marvel at the
process that allows the application of a current through physical
junction of two dissimilar metals (certain
a cooling effect rather than the I2R heating normally
associated with conductors. This article from a scientist at
Westinghouse Electric's research laboratories provides a nice
introduction to the subject of thermoelectricity from both electric
current generation based on the application of heat to a dissimilar
metals junction, and the aforementioned cooling effect possible
from passing a current through the same junction.
The Coming Breakthrough in Thermoelectricity
By Dr. John H. Kelly / Research Laboratories, Westinghouse
An important new method of energy conversion which promises
increasingly widespread applications for power generation, for
heating, and for refrigeration without the use of rotating machinery
or moving parts.
Although most of us take it for granted, the conversion of
energy from one form to another is essential to anything we
do, or expect our machines to do, and is a concern as old as
civilization itself. Indeed, the very progress of man is closely
related to his ability to uncover, tap, and convert sources
of energy into usable power. The quest for raw materials has
involved probing beneath the earth's surface, plumbing the depths
of its oceans, analyzing the microcosm of the atom, and sending
out feelers to the voids of space.
The search for conversion methods, while less spectacular,
is not one whit less important, since all the fuel in the world
would be useless without a way of using it to do some work.
One of the most important forms of energy conversion involves
the relation of heat to electricity, or "thermoelectricity."
In one form or another, thermoelectricity is about 150 years
old, but research and development in the last five years may
well point to a new breakthrough in understanding and application,
both in generating electricity and in products - industrial
and consumer - in everyday use.
Thermoelectric generators, which produce electrical power
by means of heat conversion, have not, of course, replaced conventional
generators, nor are they likely to do so in the foreseeable
future. Yet their use to supplement or augment existing generators,
is more and more indicated both by the need for additional electrical
power as well as continuous improvements in thermoelectric materials
and the technology needed to use them. Such devices, fired by
conventional or nuclear fuels - for land-based or space applications
- are constantly emerging from the nation's laboratories. Their
efficiency, cost, and other related aspects appear to be favorable
enough to indicate that they represent a "coming thing."
A striking thing about thermoelectricity is that it is a
two-way affair. Just as heat can be used to produce electrical
energy, so can an electric current be used to produce changes
in heat and on a scale suitable for anticipated future use in
major appliances such as refrigerators. No doubt small appliances
designed to take advantage of the compactness, versatility,
and convenience of thermoelectricity will precede the larger
products, but both types appear to be on the way. Thermoelectricity's
greatest impact is expected not so much in replacing existing
products, but in helping to create new and different ones. It
could, for example, usher in an era of separate, independently
controlled, refrigerated storage compartments throughout the
home, as contrasted to the single, all-purpose refrigerator
of today. The uniform efficiency of a thermoelectric cooler
has a decided advantage over the conventional refrigerator in
that in the latter type, refrigeration compressors become inefficient
and bulky for small size applications.
Fig. 1 - Distribution of electrons in unheated
and heated material.
Other useful items could also result from the new technology.
In 1958, for example, Westinghouse announced a prototype of
a device for heating and cooling a baby's bottle automatically
and another device, a mobile hostess cart, has both refrigeration
and oven compartments. Last year, two larger, full-scale devices
were demonstrated. One was a household refrigerator of ten-cubic-foot
capacity; the other, a "hot-cold-light" panel which combined
thermoelectric heating and cooling with electroluminescent lighting.
With such a panel, a home could be cooled in summer, heated
in winter, and lighted the year round - all by solid-state devices
having no moving parts and under the control of a few dials.
Presently being marketed, is a line of thermoelectric components
for cooling electronic equipment such as transistors. In addition,
small heat pumps for both cooling and heating, based on the
thermoelectric principle, are not too far in the future.
Fig. 2 - Difference in charge in the material
with one end heated.
Thermoelectricity is actually one aspect of a larger general
search for new ways of generating power which involves the use
of heat in a changing state, or thermodynamics. It is thus related
to three other distinct, although allied, fields. One is a magnetohydrodynamic
power, in which electric currents are produced by ionized gas
that is blasted at speeds of 1000 to 2000 miles-per-hour through
a magnetic field. Generators using this principle and producing
up to 10-kw. output have been demonstrated. Another form is
thermionic power, based on the electron flow from cathode to
anode in a vacuum tube. Special tubes, "thermionic converters,"
have been used to produce low-voltage power to do modest chores
such as lighting an indicator lamp or running a small fan. In
addition, combinations of thermionic-thermoelectric devices
have been successfully tested to produce small amounts of power.
Yet another form is the fuel cell, which generates usable electricity
from the combined consumption of gases, such as carbon dioxide
and oxygen, inside a high-temperature furnace. One such cell,
resembling a large snare drum, heated to about 800 degrees C,
has been said to produce 2-kw. per cubic foot.
Of all these new methods, thermoelectric power generation
is today the most well-developed and has, in fact, been used
to build large and efficient "unconventional" generators.
How It Works
Thermoelectricity, simply, is a property of two different
materials joined together and subjected to a temperature difference.
A simple explanation of this is given in Figs. 1 through 4.
In Fig. 1, the left drawing represents material at uniform temperature;
the distribution of electrons is the same throughout. As heat
is applied (Fig. 1, right), electrons leave the warmer portion,
tending to concentrate in the cooling area. As a result, the
material becomes polarized, or charged (Fig. 2) with the cooler
section being negative (more electrons).
Fig. 3 - When two dissimilar materials are
joined together and this junction is heated, the electrons are
distributed as shown. When the free ends of the two materials
are connected together through an external circuit, current
will flow in that circuit. This is the principle of the thermocouple,
which converts heat energy into electrical energy.
To use this effect to do work electrically, a closed circuit
is needed. As shown in Fig. 3, two dissimilar metals have been
joined, and the junction heated. This causes differences in
the distribution of electrons in each material. By connecting
a wire between the unjoined ends of the materials, the circuit
is closed and current will flow. This process is somewhat like
the expansion and compression of a gas.
Fig. 4 - Demonstration of the reversible
characteristics of thermoelectric materials is shown here. The
application of the proper polarity of electric current (from
a battery in this case) produces the movement of electrons shown.
As a result of this, the junction region of the two dissimilar
materials is cooled.
It should be noted that this explanation applies only to
a metal-to-metal junction (thermocouple), which has a very low
efficiency. Modern methods take advantage of the characteristics
of both the n-type (electronic) and p-type (hole, positively
charged) conductors. With these, it is possible to use a thermal
gradient to produce, in n-type material, a charge gradient as
shown in Fig. 2, and in the p-type material, a charge gradient
of opposite polarity. In other words, the cold end can be charged
positively. This is effectively adding potential, rather than
creating a difference in potential (as shown in Fig. 3). With
such materials and an understanding of semiconductor physics,
it is possible to produce efficient thermoelectric generators.
The thermoelectric effect, of course, can be reversed. If
heat can be used to pump electrons, then electrons can be used
to pump heat. As an electron moves through a material, it carries
not only its negative charge, but its associated heat as well.
In Fig. 4, a battery is used to send current through a thermocouple
circuit. Again, the electrons build up at the junction and tend
to expand across it. This expansion requires additional energy,
which is extracted from the heat energy in the region of the
junction. Consequently, the junction cools. If the direction
of the current is reversed, the effect is reversed and the junction
Thermoelectricity dates back to the "Seebeck Effect," discovered
in Germany in 1821, when Thomas Seebeck observed current flow
across a thermocouple. In 1834, the French physicist Jean Peltier
gave a partial explanation of the phenomenon and observed the
opposite effect: heat energy is absorbed at one junction and
is liberated at the other (cooling and warming, respectively)
when an electric current flows in such a circuit. This discovery
is known as the Peltier Effect. Twenty years later, the English
physicist, Lord Kelvin, offered the first detailed theoretical
explanation for both the Seebeck and the Peltier Effects.
Fig. 5 - Efficiency vs. power for conventional
machines compared to thermoelectricity.
One practical use of these phenomena has been in measuring
temperatures. The voltage generated by a pair of metals, such
as copper and constantan, is measured with a sensitive voltmeter.
This voltage, which is a function of the temperature being measured,
is typically on the order of a few millivolts for a difference
in junction temperature of 100 degrees F.
Serious attempts to use the thermoelectric effect to convert
heat to electrical power date back to 1885 and the work of another
English physicist, Lord Rayleigh. However, for the next 50 years
little progress was made toward realizing practical amounts
of power. No known pair of metals permitted efficiencies beyond
about one per-cent, too low for most use. Some limited commercial
uses were developed; for example, the automatic safety pilot
control on home gas furnaces and hot water tanks. But no large-scale
applications were in view or even seemed to be theoretically
New Power Generators
The present-day "renascence" of thermoelectricity has been
spurred by the pressing need for electric power in many new
applications, such as in remote or undeveloped areas of the
earth, as well as an active interest in the space around it.
The new developments have been accomplished largely by the work
in solid-state physics and semiconductors. With the combination,
for example, of n-type and p-type materials described earlier,
it is possible to achieve much higher efficiencies than before.
Scientists forecast that the efficiency of thermoelectric generators
in the next 15 or 20 years may reach a practical limit of 30
to 36 per-cent. This is still lower than what today's turbine
generators can produce, but it nevertheless indicates the extent
of development in a very short time and points the way to foreseeable
use of thermoelectricity in some areas of power generation.
The 135-degree heat of the test chamber melts
a stick of butter, but does not affect comfort of the man holding
it. He is wearing experimental air-conditioned suit that keeps
him at a temperature of 80 degrees F. A thermoelectric heating
and cooling unit is fitted into the back of the suit; the battery
pack in the front provides the power to make the suit portable.
Because thermoelectricity is a simple and competitively efficient
way to provide electrical power, one such use could be to serve
as a peaking complement to a central power station.
A modern, efficient power plant may have a thermal efficiency
of 42 per-cent. This figure is made up of a thermodynamic cycle
efficiency of about 45 percent and a generator efficiency of
better than 95 per-cent. In a thermoelectric generator, much
higher temperatures can be tolerated simply because there are
no mechanical stresses on the material. This permits cycle efficiencies
of 70 per-cent or better. As for generator efficiency, 40 to
45 per-cent is a reasonable-forecast. Thus, the thermodynamic
cycle efficiency is the most desirable feature of a thermoelectric
generator and the electrical generator efficiency is the more
desirable feature of a turbo-generator.
Four typical thermoelectric generators for
industrial applications. These units are presently available
and are representative of the range of such devices that can
be produced. Fuels that supply the heat required for operation
include propane or natural gas. Radiating fins on the outside
keep the "cold side" of the thermoelectric material at a lower
temperature than the heated junctions. The generators shown
have outputs of 5, 10, 50, and 100 watts. Voltage outputs range
from about 1.7 to 9.4 volts at operating currents of approximately
4 to 13 amperes.
The following table shows, for heat rate only, the rather
startling differences between thermoelectric and conventional
power conversion systems.
In the laboratory, thermoelectric generators delivering up
to 5000 watts have been built. Their efficiencies range from
less than one per-cent to about 10 to 12 per-cent. The calculated
efficiency, based on materials at hand, is about 18 per-cent.
The forecast, as stated earlier, is up to 36 per-cent, based
on the ability to solve problems regarding materials. With the
growing body of theory in this field, it is now possible to
apply the "prescription blank" method of defining the composition
of matter. This means it is possible to state a problem, consult
the physicist, transfer his prediction of best composition to
the chemist, and expect - and get - the desired property in
the final material that is produced.
Small-scale production line for final assembly
and testing on several thousand thermoelectric couples which
comprise the largest thermoelectric generator ever built. The
couples fit into interchangeable trays which line the inside
walls of a 5000-watt experimental Navy generator intended for
evaluation of materials.
It should be pointed out, of course, that the thermoelectric
generator is inherently a low-voltage, high-current, d.c. source.
To be used, this d.c. must undergo inversion and transformation
to multiphase a.c. To meet these needs, development of compatible
static inverters is under way; indeed, inverter technology is
ahead of generator technology at this time. Thus, the road ahead
to efficient and economical power conversion seems clear.
A convenient way of evaluating the competition for various
generator tasks is shown in Fig. 5, which plots efficiencies
under average use conditions of auxiliaries, prime movers, and
central power plants. The slope of the curve is a function of
three variables: economy, convenience, and geometry. Actually,
geometry is the only exact criterion and refers to the well-known
fact that the ratio of surface area (through which heat may
be lost) to volume (which, as displacement, governs power) increases
as smaller sizes are considered. This, of course, leads to increasing
efficiency. The other two factors tend to depress the left side
of the curve more than the right.
In a thermoelectric system, heat leak is the only mechanism
for power generation. Hence, the factor which reduces efficiency
in small conventional heat engines is inoperative here. What
is more, it is a fact that four thermoelectric couples give
twice the power output of two. This increased output further
substantiates the conclusion that comparable efficiencies can
be expected whether the generator is rated at one watt or one
megawatt. Thus, all of the area above the conventional power
curve, but below the present or limiting thermoelectric curve,
can be regarded as noncompetitive. The area common to both means
of power generation must be decided on the basis of initial
cost only, since there will be little difference in future maintenance
Thermoelectric cooling modules adapted with
different mounting fixtures for cooling various sizes and types
of transistors. These cooling devices are compact, provide a
controlled cooling rate, are extremely rugged, and provide reliable
On the basis of capital cost, there seems little doubt that,
in the absence of moving parts, thermoelectric generators can
be built at a lower cost than their conventional counterparts.
This fact would indicate their preference for use in many short-time
or intermittent operations, such as peaking or emergency generator
The one major problem in thermoelectricity today is to discover
and develop suitable materials. In practice, successful thermoelectric
junctions depend on a delicate balance of the properties of
the materials used. This balance is among thermal conductivity,
electrical resistance, and a special property called "thermoelectric
power," i.e., volts-per-degree of temperature difference.
Three general classes of materials must be considered. These
are metals, semiconductors, and insulators. Their physical parameters,
which govern their thermoelectric properties, are shown in Fig.
Fig. 6 - Physical parameters that determine
the best types of thermoelectric materials.
Metals have enough electrons to be very good electrical conductors,
yet standard metals are too inefficient for either refrigeration
or power generation. Semiconductors have an intentionally restricted
number of electrons, which gives them their desired characteristics
as a class of materials. They are the most efficient, although
their efficiency does drop off with an increase in temperature
and very high temperatures are needed for efficient power generation.
Finally, insulators have so few electrons that they make very
poor conductors. Yet insulators do have the highest thermoelectric
voltage, a desirable characteristic for power generation.
All told, semiconductors remain the most attractive class
of materials for use in thermoelectric cooling and power generation.
Present research to produce materials for thermoelectric cooling
is based largely on the compound bismuth telluride (Bi2Te3).
This formula, suitably modified, can demonstrate a heat pumping
efficiency of 10 to 12 per-cent. However, because of mechanical
properties, cost of raw materials, and other reasons, it has
been replaced by a number of proprietary substitutes which have
Recently, Westinghouse scientists have discovered a new class
of materials which promises thermoelectric power generation
at temperatures of about 2000 degrees F. These are known as
mixed valence compounds of the transition elements. From a junction
efficiency of one per-cent, achieved shortly after their discovery,
efficiency has risen to about ten per-cent. Such a value is
practical for a wide assortment of small-scale, specialized
applications. Eventually, junction efficiencies with these new
materials are expected in the 20 to 30 per-cent range, which
would be comparable to what is achieved with semiconductors.
Beyond the question of materials are other factors that cannot
be assessed at this time. Thus, it is impossible to predict
just what thermoelectric devices and applications will be generally
available-and when. On the basis of what has been accomplished
up to now, and the continuing progress indicated by research,
it does seem certain that these applications and devices will
come, as welcome supplements to what we now have and as thoroughly
new ways and means in our expanding electronic horizon.
This month's cover shows an experimental setup that demonstrates
a practical application of thermoelectricity. The photo shows
a standard gas furnace including a burner assembly and a squirrel-cage
blower and fan (at bottom of assembly). The furnace has been
modified so that four thermoelectric-generator modules, each
about 8" high x 4" wide x 2 1/2" deep, are mounted around the
outside of the furnace combustion chamber. The "cold side" (free
ends of the thermoelectric material) is attached to aluminum
plates that are exposed to the air. The "hot side" (junction
ends) of these modules is composed of a number of thermoelectric
pellets whose surfaces are exposed to the air as it is heated
in the gas combustion chamber. Therefore, heat is applied to
the hot side of the thermoelectric materials and is removed
through the cold-side fins. In doing so, the thermoelectric
couples generate enough electricity to power the motor that
can be seen in the base of the furnace. This motor powers the
furnace blower. Hence, this gas furnace needs no conventional
electric source to power the blower as other furnaces available
This furnace is an experimental unit being developed by the
C. A. Olsen Mfg. Co. and is not available on the market today.
It does, however, give an idea of the kind of exciting applications
available from the new science of thermoelectricity.
(Cover photo: Westinghouse Electric
Posted September 20, 2015