C H A P T E R

1
General principles of
thermoelectric technology

1.1 Introduction

The first thermoelectric effect, namely the Seebeck effect, was discov-
ered in 1821, which describes the electromotive force generated by the
temperature difference. In the following thirty years or more, Peltier
effect and Thomson effect were successively discovered. These effects
are the three main physical effects in thermoelectric technology that
describe the direct conversion between thermal and electrical energies
[1
3]. Although the discoveries of both Seebeck and Peltier effects were
made using a circuit composed of two different conductors and the
effects were only observed at the junctions between dissimilar conduc-
tors, they are actually the bulk properties of the materials involved, not
the interfacial phenomena. Solid state physics developed in the follow-
ing century reveals that all the three thermoelectric effects originate
from the energy difference of carriers in different materials and/or in
the different parts of materials under different temperatures.
Thomson built the relationship among the three effects, and devel-
oped the basic thermodynamic theories for thermoelectric effects [3].
Thomson’s work showed that a circuit composed of two conductors
with positive and negative Seebeck coefficients (usually called the ther-
mocouple) is a type of heat engine. Such heat engine can generate elec-
trical power by virtue of the temperature difference, or pump heat to
realize refrigeration. However, since the reversible thermoelectric effects
are always accompanied by the irreversible Joule heat and heat conduc-
tion, its energy conversion efficiency is principally low. Thermoelectric
effects have been widely used for temperature calibrations as thermo-
couples, but they had no practical application as heat engine, and there

Thermoelectric Materials and Devices Copyright © 2021 China Science Publishing & Media Ltd.
DOI: https://doi.org/10.1016/B978-0-12-818413-4.00001-6 1 Published by Elsevier Inc. All rights reserved.
2 1. General principles of thermoelectric technology

had been no useful theory to guide the design and fabrication of ther-
moelectric heat engines for a long time. Such situation did not change
until 1911, when Altenkirch, for the first time, analyzed the relationship
between the energy conversion efficiency and materials’ physical para-
meters (Seebeck coefficient, electrical conductivity, and thermal conduc-
tivity) in thermoelectric devices [4]. He pointed out that to enhance the
energy conversion efficiency, large Seebeck coefficient and electrical
conductivity, and low thermal conductivity are required. This outlines
the embryo for the criterion that is nowadays used to judge the thermo-
electric performance of materials—figure of merit (Z) or dimensionless
figure of merit (ZT).
This chapter will briefly illustrate the thermoelectric effects and the
relationship between the thermoelectric conversion efficiency and the
physical properties of materials.

1.2 Thermoelectric effects

1.2.1 Seebeck effect

The direct conversion from heat to electricity in solid materials was
discovered by a German scientist, Thomas Johann Seebeck, in 1821. It is
thus named as the Seebeck effect. In the next 20
30 years, the research-
ers successively discovered the Peltier and Thomson effects. These three
effects as well as the Joule effect are the physical foundations during the
thermoelectric conversion processes.
Thomas Johann Seebeck connected two different metal wires end-to-
end to form a loop, and then he observed a magnetic field around the
circuit when heating one junction and holding the other at low tempera-
ture, as shown in Fig. 1.1A. He wrote in his paper that “From the above
described experiments, it follows that the main and important condition
for the emergence of magnetism in these metal circuits is the presence
of temperature difference in the circuit links” [1]. Therefore, he named
it as thermomagnetism. Soon after that, this phenomenon was reex-
plained by Hans Christian Oersted. Oersted’s experiment demonstrated
that the magnetic field around the circuit was not directly contributed
by the temperature difference. Instead, the temperature difference gen-
erated a voltage Vab and thus an electric current in the circuit to provide
magnetism observed in experiment. Accordingly, he proposed the con-
cept of thermoelectricity. Nonetheless, since the phenomenon was firstly
discovered by Seebeck, it was named as the Seebeck effect until today.
As shown in Fig. 1.1B, when two conductors (a) and (b) connect each
other with cold-end temperature T and hot-end temperature T 1 ΔT, the

Thermoelectric Materials and Devices

 1.2 Thermoelectric effects 3

FIGURE 1.1 (A) Experimental phenomenon and (B) equivalent diagram of Seebeck
effect.

electrical potential difference Vab in the circuit can be measured at the

free ends (having the same temperature) of (b), which is expressed as

Vab 5 Sab ΔT (1.1)

where Sab is the differential Seebeck coefficient of the two conductors

with the unit of μV/K. Vab is directional, which depends on the intrinsic
properties of two constituting materials and the direction of tempera-
ture gradient. Sab is defined as positive when the thermoelectric current
flowing from hot end to cold end in conductor (a). Seebeck coefficient is
also called thermopower or thermal EMF coefficient.
The generation of thermoelctric potential can be simply but principally
explained by the fluctuation of charge distribution under a temperature
gradient. As shown in Fig. 1.2, taking the p-type semiconductor (holes are
majority carriers) as an example, when the temperature field is uniform,
the distribution of carriers (concentration, energy, and velocity) is also
uniform and the material as a whole is electrically neutral. When there is
a temperature difference between the two ends of material, the hole car-
riers at the hot end (the temperature is T 1 ΔT) gain higher energy
(E 1 ΔE) than the cold end (the temperature is T), and thus become more
prone to diffuse toward the cold end. Driven by the temperature/energy
difference, more holes diffuse to and accumulate at the cold end, and the
distribution of charges becomes nonuniform anymore, forming an inner
electric field. The inner electric field yields a reversed drift charge cur-
rent. When a dynamic equilibrium is established between the thermal
activated diffuse and inner field driven drift charge flows, a steady volt-
age V is formed.
Based on the definition of thermoelectric potential described earlier, the
absolute Seebeck coefficient of a material at temperature T is defined as

V
S 5 lim (1.2)
ΔT-0 ΔT

Thermoelectric Materials and Devices

4 1. General principles of thermoelectric technology

FIGURE 1.2 Schematic depiction of Seebeck effect.

The relation between the differential Seebeck coefficient Sab and the
absolute Seebeck coefficients Sa, Sb is
Sab 5 Sa 2 Sb (1.3)

The absolute Seebeck coefficient is independent with the direction

of temperature field and thus it is an intrinsic property of material. In
p-type semiconductors, the majority charge carriers (holes) diffuse
from hot end to cold end driven by the energy gradient, which has
the same direction as thermoelectric potential inside the material.
According to Eqs. (1.1)
(1.3), the absolute Seebeck coefficient is posi-
tive. Correspondingly, the direction of the thermoelectric potential in n-
type semiconductors is from cold end to hot end, and the absolute
Seebeck coefficient is negative. Normally, the Seebeck coefficient of
metals is very small with the values about several microvolts per Kelvin
(μV/K), while the Seebeck coefficient of semiconductors can reach sev-
eral tens or hundreds of microvolts per Kelvin (μV/K).

Thermoelectric Materials and Devices

 1.2 Thermoelectric effects 5

1.2.2 Peltier effect

Peltier effect is the inverse process of Seebeck effect, which describes
the phenomenon of directly pumping heat by carriers (holes and/or
electrons). When applying a current in the circuit composed of two dif-
ferent conductors, in addition the generated Joule heat, extra heat will
be released or absorbed at these two junctions (Fig. 1.3). This effect was
firstly discovered by a French scientist, J.C.A. Peltier, in 1834 and thus
was named as the Peltier effect. Peltier connected Bi and Sb wires and
observed the freezing of the water droplets in one of the junctions of the
two metals when applying an electric current on the circuit. After the
current was reversed, the ice was melted (Fig. 1.3A).
As shown in Fig. 1.3C, when two pieces of conductors with different
Fermi levels are connected and if an electric current is applied on this
link, the electrons will jump either from the high energy level to the low
energy level or in the opposite direction across the interface potential bar-
rier, and therefore either release heat or absorb heat at the junctions. For
example, in the metal/n-type semiconductor link, when the electrons
flow from latter to the former driven by the electric field, the electrons
jump from high energy level to low energy level accompanied with heat
release at the junction. Experimental results show that the heat absorbed
or released per unit time is proportional to the electric current
dQ
5 πab I (1.4)
dt
when current flows from a to b. Here πab is the differential Peltier coeffi-
cient with the unit of V, t is the time, and I is the current. When current

FIGURE 1.3 (A) Experimental phenomenon and (B) schematic depiction of Peltier effect.

Thermoelectric Materials and Devices

6 1. General principles of thermoelectric technology

flows from metals to p-type materials (electron flows from low-energy-

level conductor to the high one), heat is absorbed, and πab is negative.
Apparently,
πab 5
πba (1.5)

In analogy to Seebeck coefficient, the differential Peltier coefficient at

the junctions is related to the absolute Peltier coefficients of the two con-
stituting materials via
πab 5 πa 2 πb (1.6)

1.2.3 Thomson effect

The fact that both the Seebeck and Peltier effects occur only at junc-
tions between different conductors might suggest that they are interfa-
cial phenomena, but they are really dependent on the bulk properties of
the materials involved. It is known nowadays that these two effects
stem from the different properties of the materials connected together,
that is, the difference in electron energies between the two conductors.
The correlation between Seebeck and Peltier effects had not been real-
ized until William Thomson (later became Lord Kelvin) recognized this
issue in 1855. He analyzed the relationship between Seebeck effect and
Peltier effect and then proposed that there must be the third effect, that
is, when an electric current passes through a piece of uniform conductor
with a temperature gradient, reversible heat absorption or release
should occur through the whole piece beside the Joule heat. This effect
was experimentally verified in 1867 and was termed as Thomson effect.
When an electric current I passes through a piece of conductor with
temperature difference ΔT along the current direction, the heat released
or absorbed per unit time is
dQ
5 βΔTI (1.7)
dt
where β is the Thomson coefficient with the unit of V/K. If the direction
of current is consistent with that of the temperature gradient (from cold
side to hot side) and the conductor absorbs heat, β is positive and vice
versa. Considering the analogy of this expression with the definition of
material’s specific heat, Thomson vividly called β as “specific heat of
the current.” The origin of the Thomson effect is similar to the Peltier
effect. The difference is that the potential difference in the Peltier effect
comes from that of the carriers in different conductors, while it is
caused by the temperature gradient in a single conductor for the
Thomson effect. Compared to the aforementioned two effects, the

Thermoelectric Materials and Devices

 1.2 Thermoelectric effects 7

Thomson effect contributes little to thermoelectric conversion and is

therefore often neglected in the analysis of energy conversion processes
and device design.

1.2.4 Relations between thermoelectric effects and coefficients

The Seebeck, Peltier, and Thomson effects are the intrinsic properties
of bulk materials and these three coefficients are related to each other.
Thomson derived the relations among these three coefficients according
to equilibrium thermodynamics [3]
πab 5 Sab T (1.8)
TdSab
βa 2 βb 5 (1.9)
dT
These are called the Kelvin relations. The exact derivation of the
Kelvin relations should rely on irreversible thermodynamics [5]. These
two equations are verified by the experimental investigations on numer-
ous metals and semiconductors. For a single conductor, Eq. (1.9) can be
rewritten as
TdS
β5 (1.10)
dT
It can be also rewritten as
ðT
β
S5 dT (1.11)
0 T

The thermoelectric coefficients in Eqs. (1.8) and (1.9) are the differen-
tial values of two conductors. As demonstrated in Eqs. (1.3) and (1.6),
the absolute Seebeck (or Peltier) coefficient becomes equal to the differ-
ential Seebeck (or Peltier) coefficient if the second material in the circuit
is regarded as having zero Seebeck (or Peltier) coefficient. This can be
realized in practice by using a superconductor as the second material,
because both the Seebeck and Peltier coefficients are zero at the super-
conducting state. Generally, the absolute Seebeck coefficient of lead is
calibrated by measuring the differential Seebeck coefficient in the circuit
composed of lead and superconductor. If the absolute Seebeck coeffi-
cient of a material at low temperature is determined by connecting a cir-
cuit using a superconductor as the reference material, by using the
Eq. (1.11), one can find the values at higher temperatures above the crit-
ical superconducting temperature after measuring the Thomson coeffi-
cient [6,7]. Absolute Seebeck coefficients of other materials can be
calibrated by measuring the differential Seebeck coefficients in the

Thermoelectric Materials and Devices

8 1. General principles of thermoelectric technology

circuit composed of lead and the target materials. The Peltier coefficient
is difficult to measure in the experiment, and therefore it is often calcu-
lated via the Kelvin relation by using the measured Seebeck coefficient.
It is clear that, the Thomson effect is a spontaneous phenomenon as
the Seebeck coefficient changes along a temperature gradient inside a
conductor. Obviously, all the thermoelectric effects take place through-
out the whole material caused by temperature gradients and/or electric
current, though the Seebeck and Peltier effects are observed macroscop-
ically at the junctions.

1.3 Theory of thermoelectric power generation and

refrigeration

A practical thermoelectric device is usually constituted by n- and

p-type materials (legs) connected electrically in series and thermally in
parallel. A pair of n- and p-type legs, conveniently called as π-shape ele-
ment, is the basic unit of a thermoelectric device. Usually, a number of
π-shape elements make up a practical thermoelectric module via a par-
allel or series connection. The working principle of thermoelectric
power generation and refrigeration can be schematically shown in
Fig. 1.4.
Thermoelectric devices can be designed into several configurations
such as plate-like, cascaded, film, and ring-shaped devices for different
applications and/or working environment. Among them, the plate-like
device is the most typical one (Fig. 1.5), which has been widely used as

FIGURE 1.4 Schematic depiction of thermoelectric (A) power generation and

(B) refrigeration.

Thermoelectric Materials and Devices

 1.3 Theory of thermoelectric power generation and refrigeration 9

FIGURE 1.5 Structure and photo of plate-like thermoelectric devices.

power generation and refrigeration. Taking this type of device as an

example, this chapter will describe the relationship between energy con-
version efficiency and material performance (thermoelectric coeffi-
cients). To simplify the model and obtain a concise relation, the
materials’ physical parameters are taken as temperature-independent
constants. In addition, it is assumed that heat flows in the one way from
the hot end to the cold end through thermoelectric legs, and there is no
heat exchange (such as thermal radiation, conduction, and convection)
between the legs and the surrounding medium. However, practically,
the physical parameters (thermal conductivity, electrical conductivity,
and the Seebeck coefficient) of thermoelectric materials are usually
dependent on temperature, and it is not easy to keep unidirectional heat
flow from heat source and sink because the heat exchange between ther-
moelectric legs and surrounding medium cannot be completely pre-
vented. The performance prediction and design of practical devices are
much more complex and will be discussed in detail in Chapter 7,
Design and Fabrication of Thermoelectric Devices.

1.3.1 Thermoelectric power generation

1.3.1.1 Efficiency η
Thermoelectric devices can generate electric power and drive the
load when there is a temperature gradient between the two ends. The
energy conversion efficiency is the most important performance indica-
tor for thermoelectric devices. Giving the temperatures at the hot and
cold ends of the π-shaped device as Th and Tl (shown in Fig. 1.6),
respectively, the thermoelectric energy conversion efficiency (η) is
defined as
P
η5 (1.12)
Qh

Thermoelectric Materials and Devices

10 1. General principles of thermoelectric technology

FIGURE 1.6 Thermoelectric power generation.

where, P is the output power on the load, and Qh is the heat input at
the hot end (supplied by heat source). Here we do not consider the ther-
mal and electrical resistances at the interfaces as well as the Thomson
heat within the legs. On the assumption of unidirectional heat flow
without side heat dissipation, the net income heat at the hot junction
will be transferred from hot end to cold end by thermal conduction [K
(Th 2 Tl)] and Peltier pump. According to the Peltier effect, when taking
a p-type conductor as example, heat will be absorbed at the current-in
end (hot end in Fig. 1.6) and be released at the current-out end. The
amount of Peltier pumped heat from hot end to cold end in Fig. 1.6 is
πpnI, where I is the current and πpn is the total Peltier coefficient of the
two legs. On the other hand, the net heat income at the hot junction is
composed of two parts, the heat input (supplied from heat source) at
the hot end (Qh) and the Joule heat (I2R/2, where R is the total electrical
resistance of the two legs). Here, it is reasonable to assume that the
Joule heat I2R transfers equally to the hot end and cold end, therefore
only half of the Joule heat (I2R/2) reaches to the hot end. Then, we can
obtain the following equation
1 2
πpn I 1 KðTh 2 Tl Þ 5 Qh 1 I R (1.13)
2

Thermoelectric Materials and Devices

 1.3 Theory of thermoelectric power generation and refrigeration 11

Combining Eq. (1.8), the heat input is

1 2
Qh 5 Spn Th I 2 I R 1 KðTh 2 Tl Þ (1.14)
2
where Spn is the total Seebeck coefficient of n and p legs. The Seebeck
voltage in the circuit is
V 5 Spn ðTh 2 Tl Þ (1.15)

Giving the resistance of the load as Rl, the loop current and output
power are
Spn ðTh 2 Tl Þ
I5 (1.16)
R 1 Rl
 
Spn ðTh 2TÞl 2
P5 Rl (1.17)
R1Rl

Eq. (1.12) can be refined as

P I 2 Rl
η5 5
Qh 1
Spn Th I 2 I 2 R 1 KðTh 2 Tl Þ
2
(1.18)
S2pn ðTh 2 Tl ÞRl
5
1 2
S RðTh 1 Tl Þ 1 S2pn Th Rl 1 KðR1Rl Þ2
2 pn

Here it is convenient to define a parameter Z, called thermoelectric

figure of merit, as
S2pn
Z5
RK
where Z is mainly determined by the properties of the thermoelectric
lp Ap
legs; R 5 ln
An ρn 1 Ap ρp , K 5 An
ln κn 1 lp κp in which ρ and κ are the electri-
cal resistivity and thermal conductivity, respectively; A and l are the
sectional area and length of the thermoelectric legs. Then the efficiency
can be expressed as
T h 2 Tl Rl =R
η5 (1.19)
Th   ð11R1 =RÞ
2

1 1 R1 =R 2 Th2T2hTl 1 ZTh

Th 2 Tl
where Th is the Carnot cycle efficiency. One can define ε 5 Rl/R and
@η @η
obtain the p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi value of η when @ε 5 0. In Eq. (1.19), when @ε 5 0,
maximum
ε 5 Rl =R 5 1 1 ZT (T is the average temperature of the hot and cold

Thermoelectric Materials and Devices

12 1. General principles of thermoelectric technology

ends, T 5 ðTh 1 Tl Þ=2). Then the thermoelectric generator exhibits the

maximum energy conversion efficiency
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Th 2 Tl 1 1 ZT 2 1
ηmax 5 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (1.20)
Th 1 1 ZT 1 T1 =Th

It is seen from Eq. (1.20) that the maximum efficiency is only related
to the temperature difference and the ZT value. Like other heat engines,
thermoelectric generator takes the Carnot cycle efficiency as the up limit
of its energy conversion efficiency.
As discussed earlier, the parameter ZT of a device is a dimensionless
value, which is determined by the properties of the thermoelectric mate-
rial. It is conventionally to define the dimensionless figure of merit (ZT)
of a material to evaluate its thermoelectric performance
S2 σ
ZT 5 T (1.21)
κ
Clearly, for a given operation temperature range, a larger ZT will
produce higher efficiency. Fig. 1.7 demonstrates the relationship
between the efficiency of thermoelectric power generation and the aver-
age ZT of the material for a given Tl 5 300K and different Th. For exam-
ple, if we want to obtain a 25% efficiency using thermoelectric
technology, a comparable level to the conventional heat engine, the
average ZT of the constituent materials should be larger than 2.0 even
under a hot side temperature of 1000K. Nowadays, for most of the
state-of-the-art thermoelectric materials, the average ZT over wide tem-
perature range is smaller than unity, and thus the conversion efficiency
of practical device is much inferior to the conventional heat engines
(Fig. 1.8). Therefore, enhancing materials’ ZT takes always priority in
field of thermoelectricity.

1.3.1.2 Output power

The Seebeck voltage produced in a device under a temperature gradi-
ent (Th 2 Tl) is V 5 Spn(Th 2 Tl) as shown in Eq. (1.15). Obviously, this
voltage is the sum of the internal voltage drop and the voltage drop on
the road. The latter is the output voltage V0
Rl
V0 5 Spn ðTh 2TÞl (1.22)
Rl 1 R
The current is
Spn ðTh 2 Tl Þ
I0 5 (1.23)
Rl 1 R

Thermoelectric Materials and Devices

 1.3 Theory of thermoelectric power generation and refrigeration 13

FIGURE 1.7 Dependence of ηmax on ZT when the temperature of the cold end is fixed
at 300K.

FIGURE 1.8 Dependence of ηmax on the temperature difference when the temperature
of cold end is fixed at 300K.

Thermoelectric Materials and Devices

14 1. General principles of thermoelectric technology

Therefore, the output power P0 is derived as

S2pn ðTh 2Tl Þ2 Rl
P0 5 (1.24)
ðRl 1RÞ2

or

ε Spn ðTh 2Tl Þ

2 2
P0 5 (1.25)
ðε11Þ2 R

When the load resistance Rl is equal to the internal resistance R, that

is, ε 5 Rl/R 5 1, the output power reaches the maximum value Pmax
S2pn ðTh 2Tl Þ2 S2pn ΔT2
Pmax 5 5 (1.26)
4R 4R
The corresponding efficiency at the maximum output is
 
Th 2 Tl
Th
η5 h   i (1.27)
Th 2 Tl
22 1
2 Th 1 4
ZTh

Let Al 5 An 1 Ap, the sum of the sectional areas of the two thermo-
electric legs. And then the output power per unit area is

P0 ε S2pn ðTh 2Tl Þ2

5 (1.28)
Al ðε11Þ2 ðAn 1 Ap Þðρn ln =An 1 ρp lp =Ap Þ

1.3.2 Thermoelectric refrigeration

1.3.2.1 Coefficient of performance
Fig. 1.9 shows the working principle of thermoelectric refrigeration.
The main parameters describing the refrigeration performance include
coefficient of performance (COP), maximum refrigerating capacity, and
maximum temperature difference. COP is defined as
Qc
COP 5 (1.29)
P
where Qc is the amount of heat absorbed at the cold end (refrigeration
capacity) and P is the input electric power. As described in Fig. 1.3,
when a current I flows from the n-type leg to the p-type one at the
upper junction in Fig. 1.9, heat absorption and heat release will take
place at the upper junction and the lower junction, respectively, result-
ing in the formation of a temperature drop (ΔT 5 Th 2 Tl) between the

Thermoelectric Materials and Devices

 1.3 Theory of thermoelectric power generation and refrigeration 15

FIGURE 1.9 Working principle of thermoelectric refrigeration.

two ends. The heat pumped from the upper to the lower end per unit
time by Peltier effect is πpnI. At the same time, thermal conduction is
inevitable against the Peltier heat pumping due to the temperature gra-
dient. Giving the total thermal conductance of the legs K, the heat flow
driven by temperature difference is KΔT. Similar as in power generator,
Joule heat will be generated when the current is passed through the
refrigeration device. The Joule heat I2R transfers equally to the hot end
and cold end, therefore only half of the Joule heat (I2R/2) reaches the
cold end. Based on the aforementioned analyses and assumptions, the
thermal balance equation at the cold end can be established:
1 2
QC 1 I R 5 πpn I 2 KðTh 2 Tl Þ (1.30)
2
The refrigerating capacity per unit time at the junction Qc is
1
QC 5 πpn I 2 I 2 R 2 KðTh 2 Tl Þ (1.31)
2
or
1
QC 5 Spn Tl I 2 I 2 R 2 KðTh 2 Tl Þ (1.32)
2

Thermoelectric Materials and Devices

16 1. General principles of thermoelectric technology

l A
where R 5 Alnn ρn 1 App ρp , K 5 Alnn κn 1 lpp κp . Here the subscripts n and p
represent n- and p-type thermoelectric legs, respectively.
The voltage applied on the two thermoelectric legs is the sum of the
internal voltage drop (VR 5 IR) and the “anti-Seebeck voltage” to over-
come the thermoelectric voltage induced by temperature difference
[VS 5 Spn(Th 2 Tl)]:
V 5 VR 1 VS 5 IR 1 Spn ðTh 2 Tl Þ (1.33)

Therefore, the input power P (the rate of expenditure of electrical

energy by the thermoelectric legs) is
P 5 IV 5 I 2 R 1 Spn ðTh 2 T1 ÞI (1.34)

where the first term is the internal resistive loss and the second term is
the rate of working to overcome the thermoelectric voltage. Substituting
Eqs. (1.32) and (1.34) into the definition of COP, Eq. (1.29), we obtain
Spn Tl I 2 12 I 2 R 2 KðTh 2 Tl Þ
COP 5 (1.35)
I 2 R 1 Spn ðTh 2 Tl ÞI

Apparently, if fixing temperature difference Th 2 Tl, the COP varies

with the applied current. Let d(COP)/dI be zero, one can find the opti-
mum current ICOP at which the COP reaches the maximum value
 
Sp 2 Sn ðTh 2 Tl Þ
ICOP 5 hpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi i (1.36)
R 1 1 ZT 2 1

or
ðSp 2 Sn ÞðTh 2 Tl Þ
ICOP 5  lp
hpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi i (1.37)
ln
An ρ n 1 Ap ρ p 1 1 ZT 2 1

where Z holds the same definition as that in Eq. (1.20). Under this opti-
mum current, the maximum efficiency COPmax is
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Tl 1 1 ZT 2 Th
COPmax 5 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Tl (1.38)
Th 2 Tl 1 1 ZT 1 1

Correspondingly, the applied voltage and input power are

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðTh 2 Tl Þ 1 1 ZT
ðVCD ÞCOP 5 Spn pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (1.39)
1 1 ZT 2 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  
1 1 ZT ðTh 2Tl Þ 2
PCOP 5 Spn pffiffiffiffiffiffiffiffiffiffiffiffiffiffi (1.40)
R 11ZT 21

Thermoelectric Materials and Devices

 1.3 Theory of thermoelectric power generation and refrigeration 17

respectively, where T 5 ðTh 1 Tl Þ=2 is the average temperature of the

thermoelectric elements. Further, one can obtain the released heat per
unit time at the hot end
1 2
Q 5 Qc 1 P 5 Spn Th I 1 I R 2 KðTh 2 Tl Þ (1.41)
2
Similarly, the COP of a thermoelectric device at heating mode is
Spn Th I 1 12 I 2 R 2 KðTh 2 Tl Þ
COP 5 (1.42)
I 2 R 1 Spn ðTh 2 Tl ÞI

1.3.2.2 Maximum depression of temperature ΔTmax

Another key parameter of thermoelectric refrigerator is the tempera-
ture difference that can be established between the two ends
ΔT 5 Th 2 Tl. Obviously, this temperature difference is related to the
refrigerating capacity and current. Based on the thermal balance equa-
tion [Eq. (1.30)], it is obtained that
Spn Th I 2 12 I 2 R 2 Qc
ΔT 5 (1.43)
K
When the cold end is adiabatic, that is, Qc 5 0, let dΔT/dI be 0 under
which ΔT reaches the maximum value, and the optimum current IT is
Spn Tl
IT 5 (1.44)
R
The corresponding maximum temperature depression ΔTmax is
1
ΔTmax 5 ZTl2 (1.45)
2
This can also be used to know the minimum temperature that can be
reached at the cold end of the refrigerator
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 2ZTh 2 1
ðTl Þmin 5 (1.46)
Z
ΔTmax or (Tl)min is usually used to evaluate the performance of ther-
moelectric device as refrigerator. They are determined by the figure of
merit (Z) of the constituent thermoelectric materials and the tempera-
ture of heat sink (Th).

1.3.2.3 Maximum refrigerating capacity Qc,max

The heat absorbed per unit time by the refrigerator (or called the
refrigerating capacity) is given by Eq. (1.31). Obviously, for giving

Thermoelectric Materials and Devices

18 1. General principles of thermoelectric technology

thermoelectric materials, Qc, is related to the current (I) and the tem-
peratures at the two ends (Th, Tl). Similarly, let dQc/dI be 0, one can
obtain the optimum current Iq which satisfies Qc reaching the maximum
value at a certain temperature,
Spn Tl
Iq 5 IT 5 ; (1.47)
R
The corresponding Qc is
S2pn Tl 2
Qc 5 2 KðTh 2 Tl Þ (1.48)
2R
Next, the maximum refrigerating capacity Qc,max is obtained when
the temperature difference is 0, that is,
S2pn Tl 2
Qc;max 5 (1.49)
2R
Apparently, Qc,max is independent of the thermal conducting proper-
ties of the thermoelectric material. In addition, based on Eqs. (1.48) and
(1.49) as well as (1.43) and (1.45), the relationship between the refrigerat-
ing capacity and temperature difference is
Qc 5 KðΔTmax 2 ΔTÞ (1.50)
1
ΔT 5 ðQc;max 2 Qc Þ (1.51)
K
It is seen that Qc varies with ΔT in a linear way. To plot Qc against
ΔT, one can easily find ΔTmax and Qc,max from the intercepts onthe
A
two axes, and the total thermal conductance K K 5 Alnn κn 1 lpp κp of
the thermocouple from the slop. It is convenient to estimate the thermal
conductance of the constituting thermoelectric materials in a device
from the experimental variation of Qc with ΔT.

References
[1] J. Seebeck, Magnetische Polarisation der Metalle und Erze durch Temperatur-
Differenz Abh, Akad. Wiss. Berl. (1822) 289
346.
[2] J.C.A. Peltier, Nouvelles Expériemences sur la Caloricite descourants électriques, Ann.
Chim. Phys. 56 (1834) 371
386.
[3] W. Thomson, On a mechanical theory of thermo-electric currents, Proc. R. Soc. Edinb.
(1851) 91
98.
[4] E. Altenkirch, Elektrothermische Kalteerzeugung, Phys. Z. 12 (1911) 920.
[5] I. Muller, Thermodynamics of Irreversible Processes, North-Holland Pub.Co-distribu-
tors for USA, Interscience Publishers, New York, 1951.
[6] G. Borelius, W.H. Keesom, C.H. Johansson, J.O. Linde, Establishment of an absolute
scale for the thermo-electric force, Proc. K. Akad. Wetensch. Amst. 35 (1932) 10
14.
[7] J.W. Christian, J.P. Jan, W.B. Pearson, I.M. Templeton, Proceedings of the royal society
of London series a-mathematical and physical sciences (1958) 213
245.

Thermoelectric Materials and Devices