Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

and

Metals at High Temperatures:


Thermoelectric Power
Thermopower or, more broadly, thermoelectric phenomena play an important role in basic science, as well
as in a wide variety of applications of metals. A unique
feature of thermoelectric phenomena is that they
originate from an energy dependence of conduction
electron properties, such as their mobility and concentration (this is not true for phonon drag thermopower, which is important only at low temperatures
and will not be discussed here). The electrical conductivity and the electronic thermal conductivity,
however, only depend on the average magnitude of the
conduction electron mobility and concentration.
Thermopower, therefore, provides complementary
information on the electronic properties and the
scattering mechanisms in conductors. This article is a
review of the state of experimental knowledge of
thermopower in metallic materials at high temperatures. It is assumed that the boundary between the
high- and the low-temperature region is determined by
the characteristic energy of the most important
temperature-induced scattering mechanism. In crystalline conductors the most common temperatureinduced scattering mechanism is the scattering of
conduction electrons with lattice vibrations. The
characteristic temperature of lattice vibrations (called
the Debye temperature) of most metals falls in the
range 100400 K. This is why in practice the boundary
between low and high temperatures is about 300 K.

Thermopower, or historically the Seebeck effect, consists of the generation of an electric field, E, in the
presence of a temperature gradient, ]T, such that E l
S]T (S is the thermopower coefficient), in a conductor
under the condition of zero net current: j l 0. In
general, the coefficient S is a tensor of second rank; in
the case of isotropic media the tensor degenerates into
a scalar (Barnard 1972). A microscopic consideration
of electronic transport phenomena is usually based on
the semiclassical Boltzmann equation for the electron
distribution function f (a detailed discussion is given in
Boltzmann Equation and Scattering Mechanisms). A
solution of this equation in a linear approximation for
a stationary and homogeneous temperature gradient
reveals the following expression for the diffusion
thermopower:

&! (, T ) (k)

cf !
d (1)
c H

k
F

where electrical conductivity, (T ), is given by


(T ) l

&! (, T )

e# l(k, kh, T )
dA
12$q Q](k)Q

cf !
d
c H

(2)

(3)

with f !, l(k, kh, T ), , and being the FermiDirac


distribution function, the free path (which depends on
the wave vectors of the Bloch electrons and on the
temperature), the drift velocity, and the electrochemical potential, respectively. The integration in Eqn. (3)
is taken over a constant energy surface (k) l const. in
the wave vector k-space. Equation (1) for the thermopower is general in the scope of the Boltzmann
formalism. It can be applied irrespective of a degeneracy of the conduction electron system. Usually
one employs a further approximation associated with
the fact that the conduction electron gas in a metal is
strongly degenerated. The function kcf !\c has a
sharp peak at l , thus the integrand in Eqns. (1)
and (2) is nonzero in a narrow (as compared to )
energy region near l . Expanding the function
(, T ) in a Taylor series in the vicinity of l and
taking only the first nonvanishing terms of this
expansion, the following expressions for the electrical
conductivity and the thermopower can immediately be
written:
(T ) l ( , T )

(4)

and
E

1. Theoretical Introduction

1 1
Slk
QeQT (T )

(, T ) l

# kB
1 c
S lk
k T
3 QeQ B F c H

(5)
=

For phonon-induced scattering processes Eqn. (4)


leads to the well-known BlochGru$ neisen law, which
predicts for the phonon-induced electrical resistivity a
linear dependence of (T ) above about the Debye
temperature (Dugdale 1977).
Equation (5) is known as Motts expression for the
thermopower. According to Eqn. (5) the thermopower
is also a linear function of the temperature. This is the
main and the only general theoretical result for the
diffusion thermopower of metals at high temperatures.
Equation (5) is valid at sufficiently low temperatures:
kBTc, with c depending on the electronic structure
of the conductor. Within the free-electron model, c
coincides with the Fermi energy: cl. Since values
are typically about 5 eV (corresponding to a temperature of about 6i10% K), the criterion for the
applicability of Motts formula is satisfied at all in
practice achievable temperatures. However, for metals
with complex electronic band structures (such as
transition metals), c can be considerably smaller than
the Fermi energy, and the Mott expression fails to
describe the thermopower at high temperatures. As is
1

Metals at High Temperatures: Thermoelectric Power


exist no sufficiently general theoretical models able to
describe, even qualitatively, electrical resistivity and
the thermopower of disordered alloys.
For the analysis of experimental resistivity data, the
so-called Matthiessen rule is frequently used. It states
that the total electrical resistivity of a dilute alloy can
be represented in the form

S ( V K1)

(T ) l impjph(T )
where ph(T ) is the temperature-dependent part of the
electrical resistivity due to electronphonon scattering
and imp is the temperature-independent impurity
contribution. Matthiessens rule is well validated for
very dilute alloys and for a weak scattering potential of
an impurity (see Boltzmann Equation and Scattering
Mechanisms). A combination of the Matthiessen rule
for the resistivity and Motts expression together with
the WiedemannFranz law finally gives the
NordheimGorter rule, which relates the thermopower and resistivity of an alloy for a given temperature:

T (K)
Figure 1
Temperature dependencies of the thermopower of metals
of group I: 5, Li; $, Na; #, K; j, Rb; =, Cs. The
dotted line denotes the thermopower according to the freeelectron model. The arrows (at the end of the S(T )
curves) indicate the change of the thermopower at the
corresponding melting points. The broken line in the case
of lithium is a linear extrapolation of the experimental
results from 150 K to 250 K towards the melting point. For
caesium, S has been measured at one temperature below
the melting point (300 K) and at one just above.

shown below, the experimental data frequently do not


confirm the linear relationship given by Eqn. (5). The
thermopower at high temperatures exhibits in most
investigated metals and alloys a complex temperature
dependence which cannot be described by the simple
linear relationship that follows from Motts formula
even to a first approximation.
Experimentally it is found that transition metals
belonging to the same group in the periodic table show
a very similar temperature variation of the thermopower at high temperatures. However, for metals from
different groups of the periodic table qualitatively
quite different S vs. T dependencies have been found
(Vedernikov 1969). This and other observations
(Burkov and Vedernikov 1995) indicate that the
temperature variation of thermopower is intimately
connected with details of the density of states around
the Fermi energy. A microscopic theory able to
describe this connection and providing a realistic
description of high-temperature thermopower has not
been formulated. The situation is naturally more
complicated in disordered alloys because there are
serious difficulties in understanding of the electronic
structure of disordered metal systems. Therefore, there
2

S l (SphkSimp)

ph
jSimp
phjimp

(6)

where S, Sph, and Simp are the total thermopower, the


thermopower due to electronphonon scattering, and
the thermopower due to impurity scattering. From
Eqn. (6) it follows that the dependence of the total
thermopower, S, of an alloy on ph\( phjimp) is
linear. From the slope of the function S vs.
ph\( phjimp) and the intercept on the vertical axis,
one can determine the values Sph and Simp. The
derivation of the NordheimGorter rule ( just as with
the theoretical derivation of the Matthiessen rule),
however, is based on fairly rigid assumptions (see also
Boltzmann Equation and Scattering Mechanisms),
namely:
independence of the electronic structure on alloy
composition, and
independence of the electronphonon and electronimpurity scattering.
Both assumptions can apparently be realized in very
dilute alloys.
Metallic compounds in general have a more complex electronic structure and consequently the scattering processes of the conduction electrons cause a
stronger temperature dependence of the transport
coefficients. This is especially true for the thermopower.
2. Experimental Data
The periodic table provides a natural classification
scheme of the properties of metals and their alloys.
The selected thermopower data of metals and their
alloys are arranged according to their position in the
periodic table.

Metals at High Temperatures: Thermoelectric Power

S ( V K1)

S ( V K1)

T-(Zr)

T (K)
Figure 2
Temperature dependencies of the thermopower of metals
of group II: 5, Be along the c-axis; #, Be normal to the
c-axis; j, Ca; i, Sr; $, Ba. The arrows indicate
temperatures of polymorphic transformations for calcium
and strontium. The lines are guides for the eye.

2.1 Pure Metals


The metals from the first group of the periodic table,
lithium, sodium, potassium, rubidium, and caesium
(francium has no stable isotopes, the transport properties are not known), have comparatively simple electronic band structures, and it is known that their
Fermi surfaces have the shape of a slightly distorted
sphere. Therefore, it is expected that their properties
will be close to those of a free-electron gas system. To
some extent, this expectation is confirmed by the
temperature variation of the thermopower, since at
high temperatures the thermopower is an almost linear
function of the temperature. However, the magnitude
of the thermopower is considerably larger than the
value expected for a free-electron system (see Fig. 1).
Moreover, lithium and caesium (the latter at low
temperatures and in the liquid state) show positive S
values. A detailed discussion and a theoretical interpretation of the thermopower behavior of group 1
metals can be found in Barnard (1972).
Divalent metals belonging to group 2, beryllium,
magnesium, calcium, strontium, and barium (radium
has no stable isotopes), have Fermi surfaces of more
complex shapes than the metals of group 1. This has
an immediate impact on the temperature variation of
the thermopower of these metals. The S vs. T curves of
all metals of group 2 take much larger values and are
essentially nonlinear up to the highest temperatures

T-(Ti)

T (K)
Figure 3
Temperature dependencies of the thermopower of metals
of the titanium group: #, Ti; $, Zr; j, Hf. The lines are
guides for the eye. Titanium and zirconium show
polymorphic transformations indicated by the arrows.

(see Fig. 2). Beryllium and magnesium crystallize in


the hexagonal structure. S(T ) of beryllium, measured
along the principal crystallographic directions a and c,
shows a pronounced anisotropy. The S vs. T curve of
magnesium (not included in Fig. 2) is known up to
room temperature only. From 100 K to 300 K S vs. T
is similar to that of strontium, the anisotropy is,
however, smaller in this temperature range.
The physical properties of the transition metals are
dominated by the d-electron density of states at the
Fermi energy. These electrons in the solid state form
bands; however, these bands are much narrower than
the bands derived from s or p atomic orbitals.
Therefore, the d electrons cannot be treated as free
electrons, even to a first approximation. Reflecting the
complicated electronic structures, the thermopower of
transition metals reveals a variety of complex temperature dependencies. There is, however, one generality: transition metals belonging to the same group
of the periodic table have qualitatively similar temperature dependencies of thermopower. Exceptions to
this rule are the metals of group III: scandium, yttrium,
and lanthanum. They show rather different S(T )
dependencies, probably due to the increasing importance the f electrons play in lanthanum metal. The
similarity of the S vs. T curves among the elements in
the titanium group is shown in Fig. 3. Representative
S(T ) data of other transition metals are given in Fig.
4. The thermopower of the three ferromagnetic tran3

S ( V K1)

Metals at High Temperatures: Thermoelectric Power

T (K)
Figure 4
Temperature dependencies of the thermopower of
transition metals and of copper: #, Sc; $, Nb; j, Mo;
=, Ru; i, Ir; 5, Pt; , Cu. The lines are guides for the
eye.

The lanthanide group of metals includes the 14


elements from cerium to lutetium. Most of these
metals exhibit a thermopower similar with respect to
the magnitude and the temperature dependence. As
typical examples, the thermopower of dysprosium and
holmium are shown in Fig. 6 (Burkov et al. 1996). The
measurements presented in Fig. 6 span a rather wide
temperature range, also in the liquid state of both
metals. For a review on the transport properties of
liquid rare-earth metals see Van Zytveld (1989). The
two divalent elements among the lanthanides, europium and ytterbium, show basically the same S(T )
dependence as the trivalent lanthanides. However, the
magnitude of the thermopower of europium and
ytterbium at maximum is about 30 V K", which is an
order of magnitude larger than that of the other
lanthanides. The thermopower of cerium is a monotonic function of temperature, decreasing with increasing temperature from 80 K to 1000 K, and being
about 7 V K" at room temperature. The experimental data of the thermopower of rare-earth metals
at high temperatures are reviewed in Vedernikov et al.
(1977).
The actinides, elements from actinium to lawrencium, include mostly unstable elements. Thermopower
data exist only for thorium, uranium, neptunium, and
plutonium (Foiles 1985).
2.2 Binary Alloys

Because of the practically unlimited variety of metallic

S (V K1)

sition metals, iron, cobalt and nickel, is depicted in


Fig. 5. Below the Curie temperatures (indicated by the
arrows) the S vs. T variation is mainly due to the
evolution of the spin-polarized electronic structure as
the splitting of the spin-up and spin-down sub-bands
decreases when approaching the Curie temperature.
Above the Curie temperature the thermopower, however, shows a temperature dependence similar to that
of the other metals from the corresponding group.
Note that S vs. T of nickel above the Curie temperature
(631 K) is very similar to that of palladium.
Near room temperature the noble metals, copper,
silver, and gold, all exhibit very similar S values to the
other transition metals; however, S vs. T of silver
increases at high temperatures faster than linear,
whereas S vs. T of gold shows a saturation tendency.
The polyvalent metals to the right of the noble
metals in the periodic table all have comparatively
small thermopower, within the range 3 V K" to
k2 V K" at room temperature. Of these metals, lead
is of a particular importance for thermoelectric
measurements. Lead is used as the primary reference
to form the Absolute Thermoelectric Scale. The
thermopower of lead is determined from Thomson
heat measurements. The most precise data for lead
have been published by Roberts (1977, 1981). The
other metals that are used as the primary reference
materials (especially at high temperatures) are copper
and platinum (see Roberts 1981, Roberts et al. 1985).

T (K)
Figure 5
Temperature dependencies of the thermopower of the
ferromagnetic transition metals: j, Fe; $, Co; #, Ni.
The lines are guides for the eye. The arrows indicate the
Curie temperature.

S ( V K1)

S ( V K1)

Metals at High Temperatures: Thermoelectric Power

T (K)
Figure 6
Temperature dependencies of the thermopower of two
rare-earth metals: j, Dy; $, Ho. The lines are guides for
the eye. The arrows indicate the melting temperature. Both
metals (as well as other rare-earth metals) have a
polymorphic transformation from hexagonal to the b.c.c.
structure just below the melting temperature. However,
this transformation is not resolved in these measurements
owing to a large temperature gradient across the samples.

alloys, generally valid tendencies can be given only for


the simplest cases, such as continuous binary solid
solutions (CBSSs). There are about 30 binary metallic
systems known with complete solid solubility. The
CBSS systems can be divided into two types, namely
alloys formed of metals from the same group of the
periodic table (isoelectronic alloys) and alloys consisting of metals from different groups (nonisoelectronic alloys). The thermopower of isoelectronic
alloys shows a comparatively weak variation with the
concentration and the S(T ) behavior is in most cases
not much different from that of the elements of which
the alloys consist. A notable exception to this rule is
the AuAg system, where the S(T ) curves differ
markedly from those of the pure metals. An example
of the thermopower for an isoelectronic alloy system
(VNb) is shown in Figs. 7 and 8. In alloys of
nonisoelectronic metals, the thermopower varies
strongly with the alloy composition. The concentration dependence of the thermopower of the NbMo
system at 293 K and 1700 K is given in Fig. 7. A
pronounced concentration dependence of the thermopower with an extremum in the middle concentration
range is frequently observed in these alloy systems. In
contrast to isoelectronic alloy systems, in noniso-

at.% of Nb
Figure 7
Thermopower of VNb and MoNb alloys dependent on
the alloy composition (on niobium content): j, VNb at
293 K; i, VNb at 1000 K; $, MoNb at 293 K; #,
MoNb at 1700 K. The lines are guides for the eye.

electronic systems remarkable changes in the S(T )


curves occur for different concentrations. Usually
these changes in S(T ) are not continuous. In most
nonisoelectronic alloys, the S vs. T curves preserve
features characteristic of the corresponding pure
metals within a broad region of composition on the
side of each alloy component. Qualitative changes in
the character of the temperature dependence occur in
a comparatively narrow region of alloy composition.
A classical example of these types of alloys is the
AgPd alloy system. Transport properties of these
alloys have been the subject of comprehensive studies.
A summary of the results and interpretations is given
in Dugdale (1977). As an example, the temperature
dependence of the thermopower above room temperature for selected concentrations of the NbMo
system is given in Fig. 9. A detailed discussion of
experimental data of the resistivity and the thermopower of CBSSs is given in Burkov and Vedernikov
(1995). For the thermopower of further alloys see
Foiles (1985).

2.3 Compounds
Metallic compounds, or ordered alloys, possess an
almost unlimited variety of properties. There are heavy
fermion compounds, magnetic and nearly magnetic
compounds, intermediate valence compounds, hightemperature superconductors, and others. Among
5

Metals at High Temperatures: Thermoelectric Power

S ( V K1)

them, the thermopower of heavy fermion compounds


has probably been studied most extensively (Brandt
and Moshchalkov 1984, Bauer 1991).
These studies, however, as a rule have been performed only at low temperatures (see also Kondo Systems and Heay Fermions: Transport Phenomena and
Rare Earth Intermetallics: Thermopower of Cerium,
Samarium, and Europium Compounds). Thermopower of a family of nearly magnetic and magnetic
compounds, RCo (where R designates a rare-earth
element), has been# investigated in detail up to 1000 K
(Gratz et al. 1995).

Bibliography

T (K)

S (V K1)

Figure 8
Temperature dependencies of the thermopower of VNb
alloys: #, V; , 28% Nb; j, 58% Nb; $, 100% Nb.
The composition is given in at.%. The lines are guides for
the eye.

T (K)
Figure 9
Temperature dependencies of the thermopower of NbMo
alloys: $, Nb; j, 36% Mo; , 64% Mo; i, 84% Mo;
#, 100% Mo. The composition is given in at.%. The lines
are guides for the eye.

Barnard R D 1972 Thermoelectricity in Metals and Alloys.


Taylor and Francis, London
Bauer E 1991 Anomalous properties of CeCu and YbCubased compounds. Ad. Phys. 40, 417534
Blatt F J, Shroeder P A, Foiles C L, Greig D 1976 Thermoelectric
Power of Metals. Plenum, New York
Brandt N V, Moshchalkov V V 1984 Concentrated Kondo
systems. Ad. Phys. 33, 373467
Burkov A T, Kolgunov D A, Hoag K, Van Zytveld J 1996
Thermopower and electrical resistivity of liquid and crystalline
Dy and Ho at temperatures 3002000 K. J. Non-Cryst. Solids
205/207, 3327
Burkov A T, Vedernikov M V 1995 Electrical and thermoelectric
properties of metallic binary continuous solid solutions. In:
Srivastava S K, March N H (eds.) Condensed Matter and
Disordered Solids. World Scientific, Singapore, pp. 361424
Dugdale J S 1977 The Electrical Properties of Metals and Alloys.
Edward Arnold, London
Foiles C L 1985 Thermopower of pure metals; and Thermopower of dilute alloys. In: Landolt-BoW rnstein. Numerical Data
and Functional Relationships in Science and Technology. New
Series. Group III, Vol. 15, Metals. Springer-Verlag, New
York, pp. 48104; 123209
Gratz E, Resel R, Burkov A T, Bauer E, Markosian A S,
Galatanu A 1995 The transport properties of RCo com#
pounds. J. Phys.: Condens. Matter 7, 6687705
Roberts R B 1977 The absolute scale of thermoelectricity. Phil.
Mag. 36, 91107
Roberts R B 1981 The absolute scale of thermoelectricity II.
Phil. Mag. 43, 112535
Roberts R B, Righini F, Compton R C 1985 The absolute scale
of thermoelectricity III. Phil. Mag. 52, 114763
Van Zytveld J B 1989 Liquid metals and alloys. In: Gschneidner
K A, Eyring L (eds.) Handbook on Physics and Chemistry of
Rare Earths. North-Holland, Amsterdam, Vol. 12, pp.
357407
Vedernikov M V 1969 The thermoelectric powers of transition
metals at high temperatures. Ad. Phys. 18, 33770
Vedernikov M V, Burkov A T, Dvunitkin V G, Moreva N I
1977 The thermoelectric power, electrical resistivity, and Hall
constant of rare earth metals in the temperature range
801000 K. J. Less-Common Met. 52, 22145

A. T. Burkov

Metals at High Temperatures: Thermoelectric Power

Copyright ' 2001 Elsevier Science Ltd.


All rights reserved. No part of this publication may be reproduced, stored in any retrieval system or transmitted
in any form or by any means : electronic, electrostatic, magnetic tape, mechanical, photocopying, recording or
otherwise, without permission in writing from the publishers.
Encyclopedia of Materials : Science and Technology
ISBN: 0-08-0431526
pp. 55485554
7

You might also like