Dissertation, No.

1491

Thermoelectric properties of conducting polymers

Olga Bubnova

Organic Electronics
Department of science and technology (ITN)

Linköping University, SE-601-74 Norrköping, Sweden

Thermoelectric properties of conducting polymers

Olga Bubnova

ISBN: 978-91-7519-741-8
ISSN 0345-7524

Copyright ©, 2013, Olga Bubnova

olgbu@itn.liu.se
Linköping University
Department of science and technology
SE-601-74 Norrköping
Sweden
Printed by LiU-Tryck, Linköping 2013.
Abstract
According to different sources, from forty to sixty percent of the overall energy
generated in the world today is squandered in waste heat. The existing energy
conversion technologies are either close to their efficiency limits or too costly to
justify their implementation. Therefore, the development of new technological
approaches for waste heat recovery is highly demanded. The field of
thermoelectrics can potentially provide an inexpensive, clean and efficient
solution to waste heat underutilization, given that a new type of thermoelectric
materials capable of meeting those requirements are available.

This thesis reports on strategies to optimize a thermoelectric efficiency (ZT) of

conducting polymers, more specifically poly(3,4-ethylenedioxythiophene)
(Pedot). Conducting polymers constitute a special class of semiconductors
characterized by low thermal conductivity as well as electrical conductivity and
thermopower that can be readily modified by doping in order to achieve the best
combination of thermoelectric parameters. Conducting polymers that have never
previously been regarded as hypothetically compatible for thermoelectric energy
conversion, can exhibit promising thermoelectric performance at moderate
temperatures, which is a sought-after quality for waste heat recovery.
A rather substandard thermoelectric efficiency of Pedot-Pss can be markedly
improved by various secondary dopants whose addition usually improves
polymer’s morphology accompanied by a drastic increase in electrical
conductivity and, consequently, in ZT. In order to enable further enhancement
in thermoelectric properties, the optimization of the charge carrier concentration
is commonly used. The oxidation level of Pedot-Pss can be precisely controlled
by electrochemical doping resulting in a tenfold increase of ZT. In contrast to
Pedot-Pss, another conducting polymer Pedot-Tos exhibits superior
thermoelectric performance even without secondary doping owning to its
partially crystalline nature that allows for an improved electronic conduction.
With the aid of a strong electron donor, positively doped Pedot-Tos gets
partially reduced reaching the optimum oxidation state at which its
thermoelectric efficiency is just four times smaller than that of Be2Te3 and the
highest among all stable conducting polymers. The downsides associated with
chemical doping of Pedot-Tos such as doping inhomogeneity or chemical
dopants air sensitivity can be surmounted if the doping level of Pedot-Tos is
controlled by acidity/basicity of the polymer. This approach yields similar
maximum thermoelectric efficiency but does not necessitate inert conditions for
sample preparation. Optimized Pedot-Tos/Pedot-Pss can be functionalized as a
p-type material in organic thermogenerators (OTEG) to power low energy
electronic devices. If printed on large areas, OTEGs could be used as an
alternative technique for capturing heat discarded by industrial processes,
households, transportation sector or any natural heat sources for electricity
production.
Acknowledgments

To my own surprise, this part of the thesis is the hardest one to put into
words, since it has nothing to do with research and science, but solely concerns
with the people to whom I owe a debt of gratitude for their being by my side
when I needed it most. So, let me publically acknowledge those remarkable
individuals who gave me the courage to carry on with my PhD.
The biggest “merci” goes to my SUPERvisor Xavier Crispin – a true
scientist and mentor. I will probably never know why he had given me this
opportunity 4 years ago, but for that I am more than grateful. One can hardly
imagine how much patience it takes over the years to guide, teach and
constantly cheer up a student. Xavier has done it all and I am more certain than
I’ve ever been that this PhD wouldn’t be accomplished without his creative
mind and positive attitude.
Next big “tack” is for Magnus Berggren, my co-supervisor whose
boundless optimism and belief in science will never cease to amaze me. Thanks
to his personality and incredible energy, our group today is stronger than ever
and I am proud to have been a part of this big scientific family.
Now, it’s time for my friends in chronological order:
Hiam: We’ve shared rooms, we’ve sang, we’ve played numerous board
and card games, we’ve been kayaking and killing spiders together we survived a
volcano eruption and came back home alive. Sounds like a serious commitment,
doesn’t it? At least I know that after all we’ve been through I can call you my
friend!
Artsem: The man of remarkable personal integrity who has been a true
inspiration and a great source of strength for me. I know you will be with me on
that special day, and just for that I am looking forward to my defense.
Loig: for teaching me french, taking care of my rabbit and being such a
good company. Special thanks for rillettes de canard and brioches.
Jiang: for great atmosphere in the office and just for being such a nice
guy
Hui: for your help and support in the lab and outside.
Amal: for your friendship and kind heart
Zia: for all sleepless nights in the lab because of me
Negar: for great company and lots of fun (especially on the plane from
Marseille)
Amanda: for singing, skating, “afro-dancing” and partying with me
Many thanks to Henrik, Simone, Jun, Xiadong, Maria, Kristin, Ek,
Darius and the entire Orgel group for being a part of this journey.
Special thanks to Sophie for her indispensible help in administrative
issues and great Orgel on tour trips (especially the last one!)
To my family in Russia and France, especially to my mom and dad. You
haven’t missed a thing even being so far away from me.
 To Brice: I don’t know who or what I have to thank for having you in
my life, but I feel very fortunate.
To my little brother: I hope you know how much you mean to me.
Special thanks to Franz-Renan for a wonderful cosmo-design of the
cover.
And finally, to Lucia: for being мой пушистик.
List of included papers
Paper I: Towards polymer-based organic thermoelectric generators
O. Bubnova, X. Crispin, Energy & Environmental Science, 2012, 5, 9345-9362.

Contributions: Wrote the first draft and contributed to the final editing of the
manuscript.

Paper II: Advantageous thermoelectric properties of a semimetallic

polymer
O. Bubnova, Z. Ullah Khan, H. Wang, D. Dagnelund, Y. Geerts, M. Berggren,
X.Crispin, submitted to

Contributions: Conductivity and thermopower measurements, temperature

dependence measurements. Wrote the first draft.

Paper III: Tuning the thermoelectric properties of conducting polymers in

an electrochemical transistor
O. Bubnova, M. Berggren and X. Crispin, Journal of the American Chemical
Society, 2012, 134 (40), 16456–16459.

Contributions: All experimental work. Wrote the first draft and contributed to
the final editing of the manuscript.

Paper IV: Optimization of the thermoelectric figure of merit in the

conducting polymer poly(3,4-ethylenedioxythiophene)
O. Bubnova, Z. U. Khan, A. Malti, S. Braun, M. Fahlman, M. Berggren and X.
Crispin, Nature Materials, 2011, 10, 429-433.

Contributions: Conductivity and thermopower measurements, design and

fabrication of the thermogenerators. Wrote the first draft and contributed to the
final editing of the manuscript.

Paper V: Ph effect on thermoelectric properties of poly(3,4-

ethylenedioxythiophene):tosylate
O. Bubnova, M. Berggren and X. Crispin, manuscript

Contributions: All experimental work. Wrote the first draft of the manuscript
Table of Contents
CHAPTER 1 – INTRODUCTION .................................................................... 1
CHAPTER 2 - THERMOELECTRIC POWER CONVERSION ................. 3
2.1 THERMOELECTRIC EFFECTS........................................................................... 3
2.1.1 Seebeck effect ........................................................................................ 3
2.1.2 Peltier effect .......................................................................................... 5
2.1.3 Thomson Effect...................................................................................... 6
2.1.4 Thermal conduction and Joule heating ................................................. 7
2.2 PHYSICS OF THERMOELECTRICS .................................................................... 8
2.2.1 Thermodynamics of thermoelectrics ..................................................... 8
2.2.2 Solid-state physics of thermoelectrics ................................................. 12
2.3 THERMOELECTRIC MATERIALS .................................................................... 17
2.4 THERMOELECTRIC APPLICATIONS ............................................................... 20
CHAPTER 3 – CONDUCTING POLYMERS .............................................. 25
3.1 ELECTRONIC AND CHEMICAL STRUCTURE OF CONJUGATED POLYMERS ....... 25
3.2 CHARGE CARRIERS IN CONJUGATED POLYMERS .......................................... 27
3.3 DOPING OF CONDUCTING POLYMERS ........................................................... 29
3.4 CHARGE TRANSPORT IN CONJUGATED POLYMERS ....................................... 31
3.4.1 Role of morphology in charge transport ............................................. 31
3.4.2 Density of states and sources of localization in disordered materials33
3.4.3 Role of insulator-to-metal transition in charge transport .................. 34
3.4.4 Nearest-neighbour hopping and variable range hopping .................. 35
3.5 POLY(3,4-ETHYLENEDIOXYTHIOPHENE) ..................................................... 36
CHAPTER 4 - EXPERIMENTAL TECHNIQUES ...................................... 39
4.1 ELECTRICAL CONDUCTIVITY MEASUREMENTS ON THIN FILM ....................... 39
4.2 SEEBECK COEFFICIENT MEASUREMENTS ..................................................... 40
4.3 THERMAL CONDUCTIVITY MEASUREMENTS................................................. 42
CHAPTER 5 - GOAL OF THE THESIS ....................................................... 45
CHAPTER 6 – CONCLUSIONS AND PERSPECTIVE .............................. 47
REFERENCES: ................................................................................................ 48
Paper 1
Paper 2
Paper 3
Paper 4
Paper 5
 Chapter 1 – Introduction

Despite the energy resources limitation problem, roughly a two third of

the energy consumed in industrial processes and private households is lost to the
environment in the form of low-grade heat commonly referred to as waste heat.
Inefficiency of the waste heat recovery techniques is mostly associated with low
quality of waste heat making its harvesting economically impractical and
expensive. Besides anthropogenic heat, there are numerous natural sources of
energy such as solar, wind or geothermal energy, which is underutilized given
its great potentials. The inexorable depletion of oil and gas dictates a strong
need for “green” energy generation technologies as well as more thrifty use of
the energy that is being produced in a conventional way.
Thermoelectric energy conversion is one possible solution that could
efficiently address these problems as it offers extremely reliable, small in size,
light and silent in operation power sources with no moving parts and no need for
maintenance. Thermoelectrics that has been known for nearly two centuries is
far from being considered as a possible alternative to fossil-fuel power plants.
As it turn out, thermoelectric generation fails in that respect due to very low
efficiencies achieved in thermogenerators (TEGs)1 and high production cost that
further diminishes the already low dissemination of this technology on the
electricity production market. Thermoelectric generation only requires a small
temperature difference across a TEG to recover some portion of the waste heat
even at low temperatures, meaning that thermoelectrics can be used there where
most other energy conversion technologies fail.
Obviously, if thermoelectric installations are too bulky and rigid there is
no way to effectively adapt them for a given industrial or natural environment.
Instead, a miniaturized version of TEGs on a flexible substrate could do
exceptionally well covering large heat exchanger areas of whatever shape and
size is required. As for now, the main challenge has been to find materials
capable of highly efficient thermoelectric energy conversion. Another problem
to tackle is materials scarcity. If this condition is not satisfy a higher production
cost results. Hence natural abundance along with good processibility is
essential.
After a period of oblivion in the field of thermoelectrics, the research
into this matter has been intensified anew as a result of significant
advancements in material science. Regardless of the material complexity and
dimensionality the overwhelming majority of the thermoelectrics are highly
doped inorganic semiconductors. High temperature solid-state synthesis used to
form more complex materials out of single elements represents a big limitation
on the way of creating multinary compounds. Low temperature processing from

1
solution is a great option that enables patterning of thermoelectric elements by
means of printing techniques. Moreover, flexible substrates equally imply low
temperature synthesis. For that reason, organic conducting polymers can offer a
surprisingly good alternative provided their relatively high electrical
conductivity in doped state, extremely low thermal conductivity and Seebeck
coefficient that can be altered by redox chemistry typical in these materials.
Many conducting polymers have already been tested as potentially
appropriate materials for thermoelectric applications. Thermoelectric properties
of polyaniline (PANI)2, polythiophene (PTH), poly(3,4-
ethylenedioxythiophene): poly(styrenesulfonate)/tosylate (PEDOT:PSS,
PEDOT-Tos)3, polyacetylene (PA)4, polypyrrole (PPY)5 6, polycarbazoles
(PC)7, polyphenylenevinylene (PPV) and their derivatives show a lot of promise
and certainly worth further investigation. The research on the thermoelectric
performance of Pedot and its derivatives is currently underway, but the question
for now is what are the main strategies one can use in order to improve its
thermoelectric efficiency (zT). For that a better understanding of charge
transport in organic semiconductors is required. In many ways it’s been a hard
subject to study due to polymers morphological complexity8 and anisotropy of
their physical properties. In spite of these obstacles, the thermoelectric
properties of PEDOT based polymers are easily tunable in practice by doping,
which has been a good and straightforward method for zT enhancement capable
of inducing a dramatic rise of this value (mainly due to thermopower increase).
The material is equally characterized with high environmental stability,
nontoxicity and great technological adaptability and flexibility. It can be
functionalized with inorganic substances including various types of
nanostructured materials in order to create better performing organic-inorganic
hybrids9 10.
In this thesis some of zT optimization approaches including different
doping and chemical treatment techniques are shown on the example of
PEDOT-PSS and PEDOT-Tos. In both cases thermoelectric efficiency is
drastically changed to yield higher resulting zT. Nonetheless, the PEDOT-based
polymers exhibit rather dissimilar thermoelectric properties with PEDOT-Tos
being a much better thermoelectric potentially capable of efficient room T
energy conversion.

2
 Chapter 2 - Thermoelectric power conversion

2.1 Thermoelectric effects

Any transport phenomenon that implicates an exchange of heat and

electrical potential energy can be referred to as a thermoelectric effect. The
process in which system’s final state can be restored to its initial state without
entropy production or dissipation of energy is defined as thermodynamically
reversible. There exist three reversible thermoelectric effects: the Seebeck and
Peltier effects used for thermoelectric generation and electronic refrigeration
respectively and the Thomson effect that holds no practicality. At the same time,
the performance of a real thermoelectric device always remains lower than
Carnot efficiency due to two present irreversible processes that include Joule
heating and thermal conduction. Since it is virtually impossible to separate
reversible and irreversible processes in thermoelectrics it can only be treated
within non-equilibrium thermodynamics.

2.1.1 Seebeck effect

The discovery of the first thermoelectric effect belongs to Estonian-

German physicist Thomas Johann Seebeck11, which was made in early 1820 as a
result of his experimental investigations into a possible relation between
electricity and heat. In that experiment he used an open circuit consisting of two
dissimilar materials (standard thermocouple configuration) that is depicted in fig
1 where the cold junction is kept at temperature T1 and the hot junction at T2
creating a temperature gradient ΔT. Under these conditions an electromotive
force (Seebeck voltage) is induced in the circuit. The measured voltage ΔV is
directly related to ΔT by a proportionality coefficient, which is best known as
Seebeck coefficient or thermopower12:

3
 Figure 1. a) Seebeck effect illustration in a circuit comprising of two
dissimilar materials b) Effect of a temperature gradient on the charge carrier
diffusion in a single conductor
!!
∆! = !! ! − !! !                (2.1)
!!

The Seebeck coefficient of a thermocouple can be expressed as:

∆!
!! − !! =                      (2.2)
∆!

Practically it is not possible to measure directly an absolute Seebeck

coefficient value of an individual homogeneous conductor since evidently the
resulting emf will be zero. For that reason the combination of two different
materials is always used in Seebeck coefficient measurements. It is very
important to mention herein that the Seebeck effect is in no way a contact
phenomenon. Individual coefficients can be calculated if at a given T the
Seebeck coefficient of one of the materials (reference) forming a thermocouple
is known with good precision or negligibly small as compared to the second
material13. One obvious solution is to use a superconductor whose Seebeck
coefficient is zero, however it can be only done at low temperatures while the
operating temperatures for thermoelectrics are significantly higher. Hence, the
limitation comes from the fact that the Seebeck coefficient is a T dependent
parameter and this dependence is nonlinear and can vary appreciably subject to
material nature. Charge carriers available for conduction in the material under
the influence of temperature gradient start to diffuse from the region with higher
energy to the cooler region of a conductor that is kept at lower T (fig 1). This
process continues until the steady state is reached characterized by a stable
Seebeck voltage, which arises as a force acting against further migration of
charge carriers. If the majority of charge carriers are holes the Seebeck
coefficient is positive whereas for electrons it is obviously negative. In a
junction between two dissimilar materials charge carriers of a conductor with
higher electronic pressure will tend to diffuse into a lower pressure region of the
other conductor and once again the Seebeck voltage will build up. This voltage
difference is of course dependent on a material type (n- or p-type) and
temperature gradient. At higher T the charge carriers’ velocity is substantially
increased, which corresponds to a greater “electronic pressure”. In this situation,
the voltage difference measured in a circuit will be further raised. As it is seen
from (2.1) one needs to use two conductors with opposite signs of Seebeck
coefficient to acquire high ∆!. As we shall see later, the materials with highest

4
thermopower are doped semiconductors14, while metals15 possess lowest
Seebeck coefficients and are only used in thermocouples16 for temperature
sensing. In the early years of thermoelectrics, thermocouples based on metals
and their alloys were the only practical implications of the Seebeck effect. In
1949 A. F. Ioffe developed the modern theory of semiconductor
thermoelectrics17 using the concept of the 'figure of merit' zT, which gave the
green light to a new Seebeck effect application, namely thermoelectric
generation.

2.1.2 Peltier effect

The second thermoelectric effect, which is complementary in nature to

the Seebeck effect, was discovered in 1834 by a French watchmaker Jean
Peltier. In his experiment, the same set-up as for the Seebeck effect
demonstration was used. However, no temperature gradient was established
between the junctions. Instead, the current was passed through the loop
consisted of two conductors producing heating or cooling at the junction18 (fig
2).

Figure 2. Peltier effect in a circuit of two different materials

Similar to Seebeck, Peltier didn’t completely understand the physics of

this phenomenon, and it was Lenz who recognized the reversibility of this effect
and demonstrated that the rate of heat ejection or absorption was proportional to
the electrical current with the proportionality coefficient that was subsequently
named after J. Peltier19. At a constant temperature the absolute Peltier
coefficient of an individual material is defined as:

5
 !
Π=                (2.3)
!

Analogous to the Seebeck coefficient, for a circuit made up of two

dissimilar materials A and B the heat absorbed at the junction is given by:

! = Π!" ! = (Π! − Π! )!            (2.4),

where I is the electrical current and ΠAB is the Peltier coefficient of the
junction between two conductors. The Peltier coefficient is measured in volts
and can be defined for an individual conductor using the approach previously
discussed for the Seebeck coefficient. The same rule applies for the Peltier
coefficient sign: p-type material would have a positive Peltier coefficient and n-
type typically would be characterized by a negative Peltier coefficient. Physical
representation of the Peltier effect can be interpreted as follows: In an n-type
material in the presence of an electric field electrons tend to leave the negative
side carrying away energy in a form of heat that is subsequently released at the
positive side. In a p-type semiconductor the holes drift towards the negative side
instead, opposite to a current flow and so does the heat. Hence, reversing the
polarity of the electrical current will change the direction of heat transfer.

2.1.3 Thomson Effect

The last of the three thermoelectric effects was discovered by Thomson

(Lord Kelvin) in 1851 and referred to as Thomson effect. This effect represents
a combination of the Seebeck and Peltier effects and describes the generation of
heat when an electrical current flows through an unequally heated single
conductor20 (fig 3).

Figure 3. Thomson effect in a single conductor

6
 The heating/cooling rate in the material is proportional to the electrical
current and the temperature gradient through a proportionality coefficient !
(Thomson coefficient):

!!!!"#!$ = !"(− ! !)      (2.5)

The sign of the Thomson coefficient depends on whether an electrical

current flows from warmer to cooler side of the conductor or, in opposite, from
cooler to warmer. The total heat production within the material is given by:

!!"#$% = !" − ! ! + ! ! ! − ! k − ! !     2.6 ,

where ! ! ! is irreversible Joule heating and the term in 2.6 is the heat
conduction rate in the material. The first term in the equation is the reversible
Thomson heat that changes sign whenever the direction of the electrical current
is reversed. The Thomson coefficient is related to the Seebeck coefficient by the
first Kelvin relation:

!"
!=!          (2.7)
!"

Essentially, the Thompson effect represents the heat generated (or

absorbed) due to the fact that the Peltier heat changes with temperature. Similar
to the Seebeck and Peltier coefficients, the Thomson coefficient can be positive
or negative depending on the material’s type, if ! is temperature independent !
is zero. The Thomson effect offers no interest in the operation of thermoelectric
devices, however it shall not be neglected in experimental analysis that requires
good accuracy.

2.1.4 Thermal conduction and Joule heating

Along with the aforesaid thermoelectric effects there are two irreversible
processes that have to be mentioned: Joule heating and thermal/heat conduction
that always tend to lower the thermoelectric device performance21. Their
presence precludes device efficiency from attaining its thermodynamic limit
better known as Carnot efficiency.
Joule heating inevitably occurs as a consequence of any electric current
and in case of thermoelectrics is particularly undesirable as it converts useful
electric energy into heat that dissipates resulting in so-called ohmic loss. Joule

7
heat is given by the second term in 2.6, where i is the electrical current density
and ρ is the resistivity.
In a conductor subject to a temperature gradient the increase of its cold
end temperature is always observed as a result of heat conduction, which is
expressed as following:

!"
! = −!"           2.8 ,
!"
where A is the cross-section area of the conductor and k is material’s
thermal conductivity, the expression is known as Fourier law. The
thermoelectric efficiency loss caused by heat conduction is directly proportional
to k, hence the materials with low thermal conductivity are required for efficient
energy conversion.

2.2 Physics of thermoelectrics

2.2.1 Thermodynamics of thermoelectrics

Despite the reversibility of thermoelectric effects, the processes taking

place in a thermoelectric system are not in equilibrium and therefore beyond the
scope of classical thermodynamics. Still it can be assumed that the
transformations persisting in the macroscopic system over time are quasi-static
and the problem can be treated as near equilibrium. In other words, the system
tends to restore the state as close as possible to equilibrium and is said to be in
steady state, which is conditioned by minimum entropy production rule22. This
special case of non-equilibrium is generally treated by Onsager theory of
irreversible thermodynamics23 that establishes linear relations between
conjugate “flows” and “driving forces” through proportionality coefficients24.
The most straightforward examples that illustrate this principle are: Ohm’s law,
where the electrical flow is proportional to the electrical potential and Fourier’s
laws (the heat flow is proportional to the temperature gradient). In a
thermoelectric system the situation gets more complicated as both electrical and
heat currents are present concomitantly and interact with one another.
Furthermore, both forces can cause either of two flows or both of them
simultaneously25.
By virtue of the first law of thermodynamics the expression for the total
energy flux JE in terms of heat flux JQ and particle flux JN is given by:

8
 J E = JQ + µ e J N(2.9),
where µe is the electrochemical potential. The corresponding set of
conjugated thermodynamic potentials for energy and particles are 1/T and µe/T.
Finally, using kinetic coefficient matrix the linear combination between fluxes
and forces can be written as:

! J N $ ! LNN LNE $ !'((µe / T )$

# &=# & # & (2.10)
" J E % #" LEN LEE &% #" '(1 / T ) &%

The off-diagonal terms in the matrix are equal by virtue of Onsager

reciprocal relations that imply the symmetry of the transport coefficients in the
system, which is asymptotically close to equilibrium26.
In order to thermodynamically express the thermoelectric coefficient, we
need to rewrite (2.10) in terms of particle and heat flux density:

! J N $ ! LNN LNQ $ !'(1 / T )(µe $ (2.11)

# &=# & # &
" JQ % #" LQN LQQ &% #" ((1 / T ) &%

In the absence of thermal gradient (2.11) yields for the electrical current
density:
!eLNN
J= "µe (2.12),
T

where the proportionality coefficient is nothing but the electrical

conductivity:
e2
!T = LNN (2.13)
T

Analogously, in the absence of the electrical current (2.10) the heat

density is given by:

" L L ! LNN LQQ %

JQ = 1 / T 2 $ QN NQ ' (T = kJ (T (2.14)
# LNN &

Here, kJ is the thermal conductivity in zero electrical current limit and

2.14 is simply Fourier’s law. If instead the electrochemical gradient is zero
(JN=0), then the thermal conductivity is:

9
 kE=LQQ /T2 (2.15)

The thermoelectric effects are so-called coupled processes since

different irreversible transport phenomena are involved simultaneously and may
interfere with each other. In a circuit consisting of two dissimilar materials an
electric current will generate heat absorption/evolution at the junctions (Peltier
effect), or the electromotive force will be established in a circuit if the junctions
are maintained at different temperatures. Moreover, heat by itself can transport
electrically charged particles or these particles under electrical force can diffuse
along a conductor carrying some heat.
The Seebeck effect is a coupled thermoelectric process that requires a
presence of temperature gradient in an open circuit created by two dissimilar
conductors. The Seebeck coefficient is then represented by a ratio between the
electrochemical gradient and the temperature difference at J=0:

1 LNQ
!= (2.16)
eT LNN

To define the Peltier coefficient, one needs to consider an isothermal

case ( ! T=0), then from (2.11) and provided that J=eJN we get:

1 LNQ
JQ = J = !J (2.17)
e LNN

It follows immediately that the Seebeck coefficient is related to the

Peltier coefficient by a well-known relation: Π=αT.
Finally, using 2.14 and 2.15 we can we can express kE and kJ:

kE = T! 2" + kJ (2.18)

Since the kinetic coefficients are known, the Onsager expressions now
can be rewritten:

!T !"T 2 $
! J N $ # e2 ! e
& !'(1 / T )(µe $
(2.19)
# &=# & # &
" JQ % #!"T 2 & #" ((1 / T ) &%
#" T 3!" 2 + T 2 kJ &
e %

10
 or in terms of J and JQ:

(2.20)

Under zero electric field condition 2.20 reduces to JQ. The heat flux
divergence is given by:

!JQ = !(!TJ " kJ !T ) = TJ!! + !!TJ + !T!J + !("kJ !T ) (2.21),

where TJ!! is the Peltier–Thomson term, which under temperature

gradient leads to the Thomson contribution:

#"& # d" &

TJ!! = TJ! % ( = J(!" ) !!T ) = J % ) ! ( !T = " J!T (2.22)
$T ' $ dT '

The Thomson coefficient τ is defined via the well-known Kelvin

relation. The Peltier contribution can be obtained setting !T =0.

Finally, substituting 2.18 in 2.19 we get in terms of J and JQ:

! J $ !! "! $ ! E $
# &=# & # & (2.23)
" JQ % #"!"T kE &% "'(T %

Note, that this result defines α as a coupling term between heat and
electrical conduction and in absence of this term the matrix reduces to Ohm’s
law and Fourier’s law that are decoupled.
The thermoelectric effects are characterized by the coupling between
heat flux and electrical flux represented by the off-diagonal coefficients in 2.23.
Therefore, in a thermoelectric device in open circuit configuration, the voltage
is high, but the heat flux is too small, so that no power can be generated. If
instead the current (electrical flux) is increased, the voltage drops as the
temperature gradient is diminished due to improved thermal conduction.
Therefore, neither of two configurations provides a satisfactory solution and
hence, subject to certain provisions, there is a specific condition such that the
highest thermoelectric efficiency is found. The thermal conductivity is expected
to be maximal in the short-circuit configuration (E=0) and minimal in the open-
circuit one (i=0). Hence kE/kJ must be maximized to get the highest possible
power output. Using 2.18 we get the expression:

11
 ! 2"
kE / k J = T +1 (2.24)
kJ

The first term of the right hand side of the equation is so-called Ioffe
formula that introduces the central concept in the theory of thermoelectrics:
dimensionless thermoelectric figure of merit zT. This expression defines the
thermoelectric efficiency of a given thermoelectric material. Even though
temperature is not an intrinsic property of the material, it represents certain
working conditions that are of a great importance in practice and, besides, the
three thermoelectric parameters are temperature dependent. The figure of merit
results from the band structure, lattice dynamics, and scattering mechanisms of
charge carriers of a particular material, thus zT optimization is beyond the scope
of thermodynamics and shall be considered in the frame of solid-state physics.

2.2.2 Solid-state physics of thermoelectrics

In the following section electrical and heat transport in thermoelectric

materials will be briefly discussed with the aid of classical Boltzmann transport
equation, which is commonly used in transport properties analysis in the
systems with non-uniform particle density and temperature27. A non-equilibrium
distribution function reflects the probability of a particle to be found at some
specific location within the system associated with some local thermal velocity.
Globally, the system is described by the non-equilibrium distribution, which is
given by a sum of local equilibrium distributions. Equivalently, equilibrium
systems properties provide a basis for a study of an overall non-equilibrium
system. Since the external forces are steady, the average rate of change of the
distribution function must be zero. In other words the “in-and-out” effects of
scattering, forces and drift have to be balanced out as shown in Figure 5.
Under the steady state conditions and the relaxation time approximation,
the linear Boltzmann transport equation28 can be written as:

F f # f0
v ! "r f + "k f = # (2.25)
! !

12
 Figure 4. Boltzmann’s Equation Determines the Non-Equilibrium
Occupation of an Energy State by Balancing the Effects of Scattering (both
phonons and charge carriers can be involved), Forces (any external forces that
act on a system causing a change in the momentum of a wave) and Drift (waves
"drifting" due to differences in the distribution function) on a Small Group of
Energy States, in a Small Region of the Material

Here f (r,k) represents the non-equilibrium distribution function where r

and k are coordinates and wavevector in phase space. The right hand side is
given by f0 - the local equilibrium distribution defined by Fermi-Dirac function
for electrons and by Bose-Einstein function for phonons and τ is the relaxation
time or an amount of time it takes the non-equilibrium distribution to restore its
equilibrium upon removal of all driving forces. Note, that in Boltzmann’s
equation each particle has its momentum and position that are precisely
described and the uncertainty principle does not apply here.

E(k) E(k)

f0(k) f(k)

k k

13
 Figure 5. Illustration of equilibrium and non-equilibrium distribution
function for electrons at the conduction band

In a conductor subject to a small electric field, temperature gradient, and

concentration gradient the local equilibrium distribution function for electrons
in given by the Fermi-Dirac distribution f0=1/exp((E- ! )/kBT)+1. Here ! is the
Fermi level, kB is the Boltzmann constant. Provided that the external forces are
small, no big deviation from equilibrium is expected and the solution of the non-
equilibrium distribution function is:

# E ! EF 1 &# )f &
f = f0 + ! v % ! " rT + q(" + " r EF )(% ! 0 ( (2.26)
$ T q '$ )E '

In order to derive the expressions for thermoelectric parameters, we shall

first recall that current density and heat flux carried by electrons can be written
as:

1 3
J(r) =
4! 3
!!! qv(k) f (r, k)d k
(2.27)

1
JQ (r) = !!! (E(k) " E F (r))v(k) f (r, k)d 3k
4! 3

Substituting the expression 2.26 in the above equations gives:

!" 1 !" q !"

J(r) = q 2 L0 (F ! " r EF ) + L1 (!" rT )
q T
(2.28),
# 1 & 1
JQ (r) = qL1 % ! "µe ( + L2 (!"T )
$ q ' T

where Φ is the electrochemical potential (- ! µe/q= ! + ! EF/q) and

transport coefficients Ln are defined by the integral:

1 $ #f '
Ln = !!! " (k)v(k)v(k)(E(k) " E F )n & " 0 ) d 3k (2.29)
4! 3 % #E (

14
 From the expressions 2.28 thermoelectric parameters such as the
electrical conductivity, electronic thermal conductivity and the Seebeck
coefficient can be calculated. Assuming that the current flow and the
temperature gradient are both in the x-direction we get:

! = J x / (!"µe / q ) " T =0 = q 2 L0
x

1 !1
" = (!"µe / q) / " x T J x =0
= L0 L1 (2.30)
qT

!1 !1 1
kE = J Q / " x T J x =0
= L1L0 L1 + L2
x
T T

Or in the form of integrals:

$ #f '
! ! * ! (E) & " 0 )dE
% #E (
(2.31)
(E " EF ) $ #f0 '
k
* ! (E) & " )dE
kBT % #E (
"! B + E "#
q $ #f0 '
* ! (E)&%" #E )(dE

In 2.31 σ(E) is the differential conductivity that evaluates the

contribution of electrons with energy E to the total conductivity. This parameter
can be expressed by the product of the density of states D(E) (see Figure 6):

! (E) = q 2" (E) !! vx2 (E, k y , kz )dk y dkz " q 2" (E)vx2 (E)D(E) (2.32)

As it follows from the above equation, the electrical conductivity is

limited by a derivative of the Fermi-Dirac distribution ! ƒ0/ ! E known as the
Fermi window factor, is a bell-shape function centered at E= EF with a width of
kBT. This function is nonzero only in the region of several kBT near the Fermi
level (see Figure 6). This condition arises from the fact that at finite temperature
among all available charge carriers in the system, only those close to the Fermi
surface contribute to the conduction process. Therefore, electrical conductivity
requires large area under differential conductivity while the magnitude and sign

15
of the Seebeck coefficient are related to an asymmetry of the electron
distribution around the Fermi level. The Seebeck coefficient is the average
energy transported by the charge carriers corresponding to a diffusion
thermopower. The factor (E-EF) appears in the thermopower equation because

Figure 6. Schematic illustration of the Fermi-Dirac probability

distribution and its derivative (light green curves), the density of states D(E)
(purple curve), and their product (shaded blue area) for a semiconductor
thermoelectric material

heat is nothing but the difference between the total energy and the internal
energy. Therefore, increasing the electron density of states D(E) and electron
energy E near the Fermi level can result in a high zT.
The electron thermal conductivity is given by the third expression in
2.30 and in the simplest can be calculated using Wiedemann-Franz law that
relates the electrical conductivity and electronic part of thermal conductivity
(kE/σ=LT). The Lorenz factor (L=π2kB2/3e2) is constant for metals and can vary
in semiconductors depending on the charge carrier concentration. In order to
determine the second contribution to the thermal conductivity the knowledge
about the phonon distribution is required. Here again we turn to the Boltzmann
transport equation. Phonon transport properties cannot be affected by external
forces such as electric or magnetic field, hence 2.25 reduces to29:

f # f0
v ! "r f = # (2.33),
!
where f and f0 once again represent the non-equilibrium and equilibrium
distributions except that phonons are described by the Bose-Einstein statistics30

16
31
, which provides the average number of phonons in a system at equilibrium.
The distribution is given by:

1
f0 = (2.34)
exp [(E ! EF ) / kBT ] !1

The solution of the non-equilibrium function has the form:

f = f0 ! ! v " # r f0 (2.35)

And the phonon heat flux is given by:

$ #f '
J p = !"T *** ! vv & 0 )d 3k = !k p "T (2.36)
% #T (

Finally for the thermal conductivity contribution due to lattice vibration

in homogeneous materials we get:

1
kJ = ! C(! )v 2p (! )" (! )d! (2.37),
3
where C is the specific heat of phonons at frequency ω, vp the phonon
group velocity, and τ the phonon relaxation time.

2.3 Thermoelectric materials

According to the definition of zT, a good thermoelectric material should

have a large Seebeck coefficient, high electrical conductivity and low thermal
conductivity32 33. These parameters are uniquely defined for each specific
material.

17
 Figure 7. Seebeck coefficient, electrical and thermal conductivity as a
function of free carrier concentration for different classes of materials, after 34

The choice of materials with proper thermoelectric performance is based upon

the precise identification of the specific electronic makeup that can only be
provided by the theory of solids that connects the structure of solids to their
thermoelectric properties. Several criteria that the best thermoelectric materials
must have include35 36:
• Many valleys electronic bands near the Fermi level for large Seebeck
coefficient and low average electronegativity between the elements for
high mobility
• Large unit cell size ternary, quaternary compounds made of elements
with large atomic number and large spin orbit coupling for low thermal
conductivity
• Energy gaps equal to 10kBT for higher Seebeck coefficient.

Figure 7 provides an overview of existing material types and their

respective appropriateness for thermoelectric energy conversion. Among them,
metals have best electrical conductivities but their Seebeck coefficients are way
too low for their being even remotely considered as decent thermoelectrics. In
general, the carriers with energy above the Fermi level tend to increase the
Seebeck coefficient and those with lower energy contribute to making it smaller.
Consequently, in metals that typically have half-filled bands the thermopower is
relatively small.

18
 The resulting thermoelectric efficiency is further decreased by high
thermal conductivities. Therefore the average zT does not exceed 10-3 (see
Table 1). Thus, metals are only used in thermocouples for temperature sensing
but certainly not for power generation or cooling.
Likewise, insulators with their gigantic thermopowers (up to several
mV/K) and lowest among all known materials thermal conductivities (less than
1W/mK) are equally unfit for thermoelectric applications due to poor electronic
conduction. In principle, there is only one class of materials that can truly meet
the requirements for high zT - highly doped semiconductors. If their doping
level is properly adjusted high electrical conductivity may result, however it
should always go hand in hand with Seebeck coefficients as large as several
hundreds of µV/K and acceptably low thermal conductivity. In semiconductors
the latter parameter can be extraordinarily large exceeding typical metallic
values as in case of diamond37 or, to the contrary, be as small as that of glasses,
one example is clathrates38. The best material for room temperature thermo-
generation and cooling is Bi2Te339 40. Its thermoelectric efficiency of zT=1 for
bulk and 1.2 for nanostructured Bi2Te341has been a longstanding benchmark
which, despite numerous studies that have been performed on various mixed-
crystal semiconductors, remains unbeaten.

Metals Semiconductors Insulators

S≈5µV/K S≈200µV/K S≈1mV/K
σ=neµ≈108S/m σ ≈105S/m σ ≈10-10S/m
k≈10-1000Wm-1K-1 k≈1-100Wm-1K-1 k≈0.1-1Wm-1K-1
zT≈10-3 zT≈0.1-1.0 zT≈10-14

Table 1 - Comparison at room temperature of the thermoelectric

properties of metals, semiconductors and insulators, after 42

Because of renewed interest in thermoelectrics in recent years, the

search for efficient thermoelectric materials is on the fast track again. The list of
candidates is long and plentiful, among them the most studied thermoelectrics
are: Bismuth chalcogenides43, inorganic clathrates44, Half Heusler alloys45,
skutterudite materials46, silicides47, oxides48. Organic conducting polymers
represent a rather overlooked class of thermoelectrics that is nonetheless
suitable for room T energy conversion49 50. It is not only on account of their
generally low thermal conductivities and satisfactory thermopower and
electrical conductivity can they be considered for low T thermoelectric
generation, but also due to their technological compliance, reduced production
cost, nontoxicity and abundance. The detailed overview of their thermoelectric

19
performance as well as some examples of organic thermo generators are given
in Paper 1 of the present dissertation.

2.4 Thermoelectric applications

It was mentioned in the preceding sections that there exist three

thermoelectric effects: the Seebeck, Peltier and Thomson effects. However, only
the first two effects are of practical interest. The Seebeck effect is at the basis of
thermoelectric generation51 52, whilst the Peltier effect is used in refrigeration53
and can be equally adopted for another convenient albeit less common
application i.e. thermoelectric heating. Regardless of the application, the
appropriate device architecture is that represented by a typical thermoelectric
module (Fig 8).
The thermoelectric module consisting of many thermocouples connected
electrically in series and thermally in parallel does not have any moving parts
making it durable, silent and can be produced in a variety of sizes and shapes
depending on the specific application. Among the elements designated in Fig 8,
the thermoelectric legs of two types (n-type for electron conduction and p-type
for hole conduction) largely determine the efficiency of the device. Ideally, their
thermoelectric performance should be balanced such that the all three
parameters constituting zT are matched. This arises because the thermoelectric
legs should be ideally of the same size to facilitate the fabrication process and
enable a better performance. The thermoelectric figure-of-merit for a single
thermocouple consisting of two dissimilar materials is given by54:

20
 Figure 8. Thermoelectric module can be used in two different modes: 1)
for thermoelectric generation if an external load is connected creating a close
circuit while a temperature gradient is established across the device; 2) for
cooling if operated by a DC power source turning the surface where heat is
absorbed cold and heating the opposite side where heat is rejected.

(! p ! ! n )2 (! p ! ! n )2
Z= 2
= (2.38)
"#(k p " p )1/2 + (kn "n )1/2 $% KR

The denominator KR can be equally expressed using the geometrical

parameters of the TEG:

! k A k A $! ! L ! L $
KR = ## p p + n n &&## p p + n n && (2.39)
" Lp Ln %" Ap An %

Here L is the length, A – cross-sectional area and ρ is the resistivity, the

subscripts n and p have their usual meaning.
Since the Seebeck coefficient is not a geometry dependent parameter, in
order to maximize Z the denominator of 2.38 shall be minimized. This condition
is satisfied when:
1/2
Ln / An ! ! p kn $
=# & (2.40)
L p / Ap #" !n k p &%

In the thermoelectric generation (TEG) mode, the heat flow can be

directly converted into electrical power. The temperature difference across
thermoelectric elements controls the output voltage of the device (Fig 9).

As discussed previously the conversion of heat into electricity is

associated with some irreversible processes that cause heat losses and lower
TEG’s efficiency. If for simplicity we consider a TEG consisting of only one

21
 Figure 9. Two operational modes of a thermoelectric module: a)
thermoelectric generation mode b) thermoelectric cooling/heating mode.

thermocouple, the incoming thermal power at the hot side of the TEG is
expressed as following55:

QH = !TH I + K!T "1 / 2RI 2 (2.41)

where ! is the total Seebeck coefficient of a thermocouple (! n,p), K is

the thermal conductance similarly given by both materials, TH is the hot side
temperature and !T is the temperature gradient, R is the internal resistance of
the TEG. Hence there are three terms associated with the heat flow at the hot
side i.e. heat due to the Seebeck effect, the half of Joule heating and the thermal
conduction. The last two inputs are hugely detrimental for the TEG’s
performance and have to be minimized.
When the load with a resistance Rload is connected to the TEG the
current flowing in the closed circuit will be equal to:

I = Vopen / (R + Rload ) (2.42)

and the useful power produced by the TEG is simply Pload=I2Rload or

equivalently:

Pload = I 2 Rload = ! 2 !T 2 Rload / (R + Rload )2 (2.43)

It follows that the maximum power output is reached when the load
resistance and the device resistance are matched, which corresponds to the
condition ! P/ ! (Rload/R)=0. Therefore the Pmax is V2open/4R56

22
 The TEG’s efficiency is given by the ratio between the generated power
and the supplied power (incoming heat) φ=Pload/QH or equivalently:

I 2 Rload
!=
1
" ITH + K!T " I 2 R (2.44),
2

The efficiency of the maximum power output mode can be found setting
R=Rload and bearing in mind that the thermocouple figure-of-merit Z=α2/RK:

!T
!P = (2.45),
3 / 2TH +1 / 2TC +1 / 4Z "1

On the other hand R=Rload condition does not lead to the maximum
efficiency of the TEG. Instead, its maximum performance is obtained setting to
zero the derivative of the efficiency with respect to the internal-to-load
resistance ratio, the result for the maximum conversion efficiency is:

!max =
!T 1+ ZT "1 (2.46)
TH 1+ ZT + TC
TH

T is the average temperature of the TEG and !T / TH is the reversible

Carnot efficiency (ϕC) and the rest of the expression is the material-related
contribution since it depends on Z. Hence ϕmax ! ϕC when Z ! ∞.
Consequently, there are two possible modes of operation for a TEG, one
that provides the maximum efficiency and the other that enables its maximum
power output. If the ratio R/Rload= 1+ ZT , the TEG is in the maximum
efficiency mode, while R/Rload=1 (matched load condition) corresponds to the
maximum power generation. The discrepancy between the maximum power
efficiency and maximum conversion efficiency at ZT≈1 is within 3%. Therefore
the maximum power output mode shall be used as an optimum TEG
configuration. However, as soon as ZT ≥ 3, the deviation becomes larger than
10% and the higher heat transfer rate needs to be maintained at the hot junction
as the efficiency degrades.

23
24
 Chapter 3 – Conducting Polymers

3.1 Electronic and chemical structure of conjugated polymers

Polymers are carbon-based compounds consisting of numerous repeating

units (monomers). Conjugated conducting polymers constitute a special class of
organic materials whose ability to conduct electrical charge originates uniquely
from π –conjugation.
In regular saturated polymers, that are practically insulators, all the
available electrons are bound in strong covalent bonds and are characterized by
sp3 hybridization. In contrast to this situation, in conjugated polymers planar σ-
bond arises from hybridization of one s and two p orbitals (Fig10a). That is to
say, there is a region where p-orbitals overlap, bridging the adjacent single
bonds and creating a π collective orbital in which p-electrons do not belong to
one single bond, but rather delocalized over a group of atoms.

Figure 10. sp2 hybridization in conjugated π-system where electron

density can be delocalized over many carbon atoms.

Since the interaction between the parallel 2pπ-atomic orbitals is

weaker than between the hybrid 2pσ-atomic orbitals, the resulting occupied π-

25
molecular orbitals are less stable than the occupied σ-molecular orbitals.
Since σ-orbital is at lower energy, the highest occupied molecular orbital
(HOMO) is a π-orbital. As a result of bonding (in-phase interaction) and
antibonding (out-of-phase interaction) atomic orbital overlap. A system of N
2pπ-atomic orbitals, leads to N/2 π-molecular orbitals with a predominant
bonding character that are stable and occupied with N π-electrons. There are
N/2 unoccupied π*-molecular orbitals with a predominant antibonding
character. The most stable of those orbitals is called the lowest unoccupied
molecular orbital (LUMO). In the solid phase HOMO and LUMO rearrange to
form bands somewhat analogous to the valence band and the conduction band in
inorganic semiconductors. The existence of bond alternation in conjugated
polymers is in fact responsible for the band gap formation and semiconducting
properties in these materials. According to the Peierls theorem, the metallic state
with one electron per ion is unstable in (quasi) one-dimensional material with
the tendency to distort the geometrical structure and lower the symmetry.
A typical conjugated polymer consists of a large number of repeating
units. In a system containing many polymer chains that tend to intertwine
creating spaghetti-like structures it is virtually impossible to maintain a long-
range order. Therefore, conjugated polymers are rather amorphous in nature
with the conjugation length strongly dependent on intra- and intermolecular
structure. Their physical properties are very sensitive to any modifications in
chemical architecture and can vary significantly subject to the morphological
complexity of these materials. Generally, conjugated polymers are shorter in
length than saturated polymers and characterized by stiffness with respect to
chain twisting and bending. Besides, their solubility is very poor unless specific
side-chain groups are introduced. Some examples of conjugated polymers are
shown in Figure 11.

26
 Figure 11. Chemical structure of some most commonly used conjugated
polymers

3.2 Charge carriers in conjugated polymers

Depending on the ground state of a conjugated polymer, different exited

states can occur. A polymer whose ground state energy remains unaffected
when interchanging single and double bonds is referred to as a degenerate
ground system. In trans-polyacetylene, for example, it is of no importance for its
energetics on what side the double bond is (see Fig. 12)57. Conversely, in
polyphenylene the change of the double bond position transforms the aromatic
lower energy state (three double bonds in the ring) to the quinoid state with
much higher energy (two double bonds in the ring)58.

Figure 12. a) Total energy curve for an infinite trans-polyacetylene chain

as a function of the degree of bond length alternation b) The energy levels of
two poly(para-phenylene) states with stable aromatic configuration that can be
excited to quinoid state.

A phase kink in a single-double bond sequence in trans-polyacetylene

made of odd number of carbon atoms results in a formation of an in-band defect
in a form of an unpaired electron called a neutral soliton59 (see Fig 13a). In
doped trans-polyacetylene positive or negative spinless solitons can be
created60. However, most of the conjugated polymers have a non-degenerate

27
ground state, which impedes the soliton formation. It is suggested that in these
polymers polarons and bipolarons are created upon doping61 (Fig.13). The
oxidation (an electron removal) of the polymer leads to a formation of a radical
cation. Its presence causes an attenuation of neighbouring bond alteration
amplitudes and a half-filled electronic level above the valence band along with a
new anti-bonding level below the bottom of the conduction band results (Fig.
13b). Polaron is a single charge excitation that can be either positive as
discussed above or negative in case if an electron is added to a polymer chain. A
polaronic distortion can spread over three to four monomers depending on
polymer chemical structure. A removal/addition of the electron from a polaron
produces a bipolaron i.e. a radical ion pair associated with a strong local lattice
distortion.

28
Figure 13. a) Chemical structure of trans-polyacetylene with tree types of
solitons and the corresponding electronic structures. b) A schematic illustration
of the different excited states in polythiophene and the polymer’s electronic
structure corresponding to (from left to right) a neutral chain, a chain carrying a
polaron, a chain carrying a bipolaron, a chain or assembly of chains with a high
concentration of polarons (intra chain or interchain polaron network), a chain or
assembly of chains with a high concentration of bipolarons (intrachain or
interchain bipolaron bands). The relaxation of the lattice around a doubly charge
bipolaron is larger than in case of a polaron with only one charge, hence the
bipolaron in-gap states are moved even further away from the band edges. It is
suggested that a bipolaron formation is more thermodynamically favorable than
a formation of a polaron pair since the energy required for their creation is
nearly identical while in case of bipolaron the decrease in ionization energy is
greater.

3.3 Doping of conducting polymers

The presence of π–conjugated chain is the first fundamental requirement

for a polymer to become conducting. However, the band gaps of conjugated
polymers are ranging from 1 to 3eV62, which is consistent with semiconducting
or even insulating properties. In order to acquire metallic electronic
characteristics, a doping process is used. In contrast to inorganic doping
involving atoms replacement, doping in conducting polymers is due to an
oxidation (p-type doping) or reduction (n-type doping) process. The oxidation
level in conducting polymers can be extremely high, sometimes up to 50%63
exceeding that of inorganics by several orders of magnitude.
Oxidation/reduction can be achieved by removing/adding an electron from/to a
polymer via an electron transfer reaction with an chemical species (redox
reaction) or with an electrode (electrochemical reaction). The doping charge on
the polymer is stabilized by a counter ion (cation/anion) to ensure
electroneutrality of the materials. Polymer can either be synthesised either in
their doped state or their neutral state. The oxidation level can be controlled in a
number of ways.
Chemical doping is a method that involves an exposure of a polymer to a
solution or vapor of the dopant64. A suitable dopant (donor or acceptor) used in
this process should have a proper redox potential and the polymer’s ionization
potential (electron affinity) shall be such as to facilitate the doping process.
Throughout the exposure chemical dopants can physically penetrate into or
move out of the polymer putting a strain on the polymer film, which can cause
its swelling, shrinking or other morphology change. Increase in a charge carrier

29
concentration manifests in a tremendous change in electrical conductivity that
can be nearly metallic in some highly ordered polymers65.
Indeed, a choice of chemical dopant is rather limited since in each case
the requirement for a dopant would be dictated by a unique polymer electronic
structure. Typical dopants used in chemical doping of conjugated polymers and
the achieved oxidation levels are listed in the Table 2.

Dopant ion Formula

Chloride Cl−
Perchlorate ClO4−
Tetrafluoroborate BF4−
p-toluene sulfonate CH3-C6H5-SO3−
Trifluoromethane sulfonate CF3SO3−
Hexafluorophosphate PF6−
Polystyrene sulfonate (-CH2CH(C6H4SO3)-)n
Bromide Br-
Sulfate SO42-
Iodide I-
Polymer Maximum doping level
Poly(pyrrole) 33% (ClO4−)
Poly(thiophene) 30% (ClO4− ) 6% (PF6−)
Poly(aniline) 42% (Cl− )
Poly(p-phenylene) 44% (Li+)

Table 2 - Anionic and cationic dopants commonly used for chemical doping in
conducting polymers. Some examples of maximum doping level that can be
achieved in certain conducting polymers with the aid of chemical doping
(after66).

Chemical doping takes place via redox reaction (electron transfer with a
dopant species), this is a straightforward and technologically simple charge
concentration modification method. However, there are some known downsides
to this approach such as a limited choice of dopants and chemical stability
issues of the dopant due to their high reactivity. Another problem arises from
the homogeneity of the doped samples, which can be high in heavily doped
polymers but fails to be so at intermediate doping levels67. In some applications
where semiconducting properties of conjugated polymers are requested it
becomes crucial to ensure homogeneity of moderately doped samples. This
problem can be successfully tackled by electrochemical doping.
In electrochemical doping an electrolyte is used between a conducting
polymer and the electrode that supplies the charge to the polymer while ions

30
from the electrolyte diffuse in and out to maintain polymer’s charge neutrality.
In case of n-doping the polymer is reduced and the electrolyte cations are
introduced into the polymer as counterions. Upon p-doping, the polymer is
oxidised and the anions penetrate into the polymer film as charge balancing
ions. This method provides precise control of the doping level, which is set by
the applied potential between the electrode and the polymer film. Almost any
conjugated polymer can be electrochemically doped but the stability of the
obtained samples strongly depends on the applied voltage. Compared to
chemical doping the electrochemical doping procedure is technologically more
complicated to perform as it necessitates an electrolyte and a counter electrode.
Consequently, it is not ubiquitously used. This doping method is employed in
electrochemical batteries, light-emitting electrochemical cells68 and
electrochemical transistors69.
Other doping methods include photodoping70 based on photo-absorption
and charge separation (photovoltaic devices), doping by charge injection
without counterions involved (OFET)71, doping by acid-base treatment72.

3.4 Charge transport in conjugated polymers

3.4.1 Role of morphology in charge transport

A conducting polymer film can be described as large macroscopic

assemblies of charged polymer chains and ions. The cohesion in the solid is
maintained by electrostatic (dominant in doped polymers) and van der Waals
forces. Clearly, the structural order can’t be effectively maintained over large
distances provided a significant length of polymer chains. Most of the
conducting polymers appear to be amorphous, sometimes with some degree of
crystallinity73, as in the case of highly conducting samples. The latter example
pertains to the situation where the three-dimensional highly conducting
inclusions and disordered quasi one-dimensional regions with strong
localization effects coexist74 (Fig 14). The polymer chain packing is the only
difference between amorphous and crystalline regions as both of them are
obviously made of the same material.
High degree of disorder in amorphous materials translates into strong
charge localization. These systems are featured with low charge carrier mobility
and electrical conductivities even at high charge carrier concentrations. These
so-called “dirty” conductors are extremely disordered and can be attributed to
the Fermi glass concept75.

31
 In “metallic” conducting polymers the interchain order and crystalline
coherence length are expected to be large while the defect density has to be low
to enable delocalization of charge carriers with mean free paths significantly
exceeding one repeat unit76.
The charge transport in conducting polymers can proceed in three
possible ways: 1) propagation of a charge carrier along a polymer backbone, 2)
transfer across the polymer chains by hopping, 3) tunnelling between
conducting segments separated by amorphous regions. The delocalization of π-

Figure 14. Schematic representation of conducting polymer

morphology represented by metallic highly ordered regions randomly
distributed in an amorphous host made from chains of the same polymer.

electrons along the polymer backbone alone is not a sufficient condition for
conduction due to intrinsic localization in a one-dimensional system. For this
reason the interchain coupling is essential77. Usually the transfer integral for
interchain coupling represents a very small fraction of the one-chain transfer
integral. Nonetheless, even being weak it can still encourage a charge carrier to
hop from site to site diffusing across the polymer chains instead of being
trapped by a defect or having to move in the opposite direction. The interchain
transfer is further disrupted when side groups are introduced into initial
conjugated polymers. This modification of the chemical structure improves the
solubility of the polymer but may corrupts the backbone rigidity resulting in the
reduced intermolecular overlap78.
Another important structural aspect affecting the charge transport in the
conducting polymers is the molecular weight of polymers. Polymers with low
molecular weight have shorter chain lengths and are more crystalline in nature

32
than high molecular weight polymers. Yet, the mobility in these polymers is
lower possibly because the connections between the crystals are obstructed by
insulating boundaries. In high molecular weight polymers bends and twists
interrupt the conjugation, but at the same time longer polymer chains that may
spread trough several material segments connect more effectively ordered
regions between one another74.

3.4.2 Density of states and sources of localization in disordered

materials

As previously discussed, the conduction process in disordered systems is

generally limited by the localization of the electronic states. Conducting
polymers are commonly described by the density of states (DOS) where
delocalized-states band and smooth band-tails with localized states at the band
edges are separated by a "mobility edge"79 (see Fig 15a). When the Fermi level
is found below the mobility edge all the charge carriers are localized at zero
temperature and the conduction is ceased. An abrupt transition from insulating
to metallic state is predicted once the Fermi level overpasses the mobility edge.
This transition is known as Mott Transition80. Another type of localization
observed in organic materials is Anderson transition81, which may take place if
the mobility edge shifts deeper into the band. Once the random disorder
potential exceeds the bandwidth all states become localized and the system turns
into an insulator with continuous DOS even in the absence of the band gap (Fig
15b). The material is then called a Fermi glass as mentioned previously.

Figure 15: Density of states types in conjugated polymers including:

a) band of fully localized states, b) band with localized states in a tail
and free states above the mobility edge, c) in-gap band of localized states that
can be represented by polaron or bipolaron bands as shown in Figure 13. The
Fermi level position can vary, if, for example, the doping level is changed.

33
 In conducting polymers the in-gap localized states are induced by
doping in the form of (bi)polaron states/bands. In the lack of better knowledge
about the DOS shape, the distribution of these states is commonly assumed by
Gaussian DOS82 as shown in Figure 15c. If the distortion of lattice due to
polarization is strong (bi)polarons become self-trapped. A deep potential well is
created around a localized carrier when the lattice acquires a more stable
equilibrium state. Hence, the (bi)polaron transport can be carried out by means
of phonon assisted hopping between available states (Fig 16). Unlike the band
regime with strong electron coupling, a coherent transport is not enabled in
hopping transport since the charge carriers do not have a “memory” of where
they were previously and the process is virtually random83.

Figure 16. Hopping transitions from occupied (full circles) states to

available sates above the Fermi level.

3.4.3 Role of insulator-to-metal transition in charge transport

The mechanism of the charge transport in conducting polymers largely

depends on to what side of the metal-to-insulator transition the polymer
belongs84. Upon doping the (bi)polaron lattice can be formed progressively
filling the band gap. Eventually a distinct band structure can emerge. In reality
even the best conducting polymers are only partially crystalline suggestive to
inhomogeneous doping with highly conducting regions separated by insulating
boundaries85. The conductivity in these materials grows rapidly once the
percolation threshold is surmounted.
To give a rough estimation to whether a polymer is in its insulating
(semiconducting) or metallic state, the reduced activation energy is calculated:

d(ln ! )
W= (3.1)
d(lnT )

34
 To a first approximation negative slope of W versus T is suggestive of
the insulating regime, positive coefficients of W imply a metallic nature of a
sample, and finally, a weak temperature dependence of W is an indication of the
critical regime86.

3.4.4 Nearest-neighbour hopping and variable range hopping

With the exception of metallic polymers with the resonant quantum

tunneling transport among the chain-linked network of nanoscale grains,
hopping is considered as a basic transport mechanism for polymers87. For the
conjugated polymers on the insulating side of IMT the conduction of charges is
governed by thermally activated hopping. Subject to the localization degree two
types of hopping process can be distinguished i.e. 1) nearest-neighbor hopping88
and 2) variable-range hopping89 90. The probability of hopping is generally given
by:

W
Phop = ! phonon exp(!2" R ! ) (3.1)
kT

where, vphonon is the phonon frequency; ! is the decay rate of the

wavefunction or the inverse of the localization length; R and W are the distance
and the energy difference between two sites. If the condition ! R0>>1 (R0 is the
nearest neighbor distance) is satisfied it implies a strong localization with the
mobility edge shifted in the higher band where only nearest-neighbor hopping
within kT is allowed. This is to say, the charge transfer between localized states
is solely defined by spatial factors, which is valid at high temperature limit.
Once the energy distribution of the localized states becomes comparable or
larger than kT, the energy dependent term comes into play. At some point, the
energy difference between two closest states can happen to be larger than that
between more remote states lowering the nearest hop probability and increasing
the chance of a longer hop event. Therefore, at lower temperatures the variable-
range hopping (VRH) first suggested by Mott shall be assumed. The most
effective transport of charge carriers shall be expected in the vicinity of the
Fermi energy where many empty states with matching energy can be available
for occupation.
For the hopping conduction, the electrical conductivity is zero at T=0K
since the charge carriers are localized. At finite temperatures, charge carriers
hop by absorbing or emitting phonons. The typical temperature dependence of
conductivity is expressed by91:

35
 ( " T %1/d +
! = ! 0 exp *! $ 0 ' - (3.2),
*) # T & -,

where d=n+1, where n is the system’s dimensionality, T0 is the

characteristic temperature given by:

!" 3
T0 = (3.3)
kg(EF )

In this expression β is the numerical constant and g(EF) is the DOS in the
vicinity of the Fermi level. This model, however, does not take into account the
long-range electron-electron interactions assuming constant DOS.
VRH is consistent with d=4, while d=2 is generally explained in the
framework of the charging-energy limited tunneling model. It is been suggested
that conduction proceeds through tunneling between small conducting grains
separated by insulating barriers (Sheng’s model of granular metals)92.
According to this model, highly doped metallic-like islands are generated by
heterogeneities in the doping distribution, bridging neighboring chains and
improving the charge transport.
The best fit to the experimentally observed temperature dependent
conductivity can be attained using the heterogeneous model proposed by
Kaiser73, where based on complex polymer’s morphology the electrical
conductivity can be represented in terms of the fractions of ordered metallic
regions, insulating barriers and disordered metallic barrier regions. A more
detailed discussion of the temperature dependent conductivity and thermopower
behavior is provided in the Paper 1.

3.5 Poly(3,4-ethylenedioxythiophene)

Poly(3,4-ethylenedioxythiophene) better known as (PEDOT) does not need a

long introduction as over the course of many years this truly remarkable
conducting polymer has been used far and wide. A long list of PEDOT
applications includes photovoltaics93, light-emitting diodes94, sensors95,
actuators96, thin-film transistors97, bioactive coatings98 etc. In addition to a very
high conductivity (ca. 200 S/cm)99, PEDOT forms transparent and stable films
in its oxidized state, has moderate band gap ofapproximately 1.6 eV100 and low
redox potential101. Yet, it is virtually insoluble and unstable in its neutral state.
To circumvent the solubility problem, polystyrene sulfonate (PSS) is commonly
used as a supplement yielding an aqueous dispersion of PEDOT:PSS (see

36
figure17). The introduction of a bulky insulating counterion cannot but affect
the electrical conductivity of the polymer, which drops tenfold.
Subject to counterion choice, PEDOT can exhibit conductivities ranging
from 10-2 to 3500 S/cm. PEDOT-PSS typically has low conductivity
(<10S/cm)102, but it can be readily improved using various organic and
inorganic solvents treatment. Secondary doping with polar solvents such as
dimethyl sulfoxide (DMSO)103, sorbitol, dimethyl sulfate104 and ethylene glycol
(EG)102 or acidic treatment105 with HCl or H2SO4 have proven to be very
effective in electrical conductivity enhancement. Besides, there are some
commercially available aqueous dispersions of PEDOT-PSS under the
trademark Clevios that are highly conductive (up to 800S/cm). Insoluble films
of PEDOT doped with tosylate106 (Fig.17b) prepared by chemical
polymerization yield much higher conductivities (up to 1000S/cm) and unlike
amorphous PEDOT-PSS are partially crystalline.

Figure 17. Chemical structure of PEDOT doped with a) polystyrene sulfonate and b)
tosylate

Environmental stability and high conductivity of these polymers is a

good starting point for investigating their thermoelectric performance. Similar to
other conjugated polymers the oxidation level of PEDOT can be modified to a

37
large extent, providing an additional flexibility to its electronic properties. The
thermal conductivity in PEDOT derivatives is low as expected from typical
polymers (ca. 0.2W/mK), slightly higher for PEDOT-TOS. The Seebeck
coefficient is positive and small (almost metallic) in highly doped films.
Moderate thermopower is the major concern and the blocking point for high
thermoelectric efficiency. Once again, doping can be a relevant approach to
tackle this issue. As previously stated, various doping procedures are
appropriate for conducting polymers and can be adopted for optimization of
thermoelectric properties, which is the ultimate goal of this thesis.

38
Chapter 4 - Experimental techniques

4.1 Electrical conductivity measurements on thin film

Among all parameters required for thermoelectric characterization of

materials, the electrical conductivity is perhaps the easiest to measure. A four-
point probe method used for this purpose is known to be straightforward and
simple. In this measurement four equally spaced electrodes are coated on a
substrate whether it’s glass, silicon or another appropriate material. In order to
exclude possible errors due to the parasitic contact resistance between the thin
film surface and the probe one set of leads is used to source an electrical current,
while the other one is intended for the measurements of a resulting voltage. The
electrical conductivity of a thin film with the cross sectional area A and the
length l determined by the distance between two inner electrodes is given by:

I l I l
!= = (4.1)
V A V f !w

Here I is the electrical current injected through two outer leads and V is
the voltage measured between two inner leads. The cross sectional area is found
as a product of the film thickness f and its width w.
Presence of electrical current in a thermoelectric material can cause the
emergence of the Seebeck voltage due to the Peltier effect. The better the
material’s thermoelectric performance the larger this contribution may become.
Therefore, it is important to minimize this effect so that only the resistive
component of the measured voltage is used in the electrical conductivity
estimation. In order to eliminate this source of errors a fast DC or AC
measurements are used. The DC current should not be applied for more that 2-3
seconds and its direction can be reversed to obtain V by averaging the results of
measurements in both directions. Besides, all four ohmic contacts should consist
of the same material.
In the samples with arbitrary shapes the electrical conductivity can be
estimated by the van der Pauw technique107 that was designed for two-
dimensional system where the sample thickness is much smaller than its length
and width. The four contacts are chosen such that they are located at the
extremities of the sample, which by itself should preferably be as symmetrical

39
as possible. The resistances between the points (one along a horizontal edge and
the other along the vertical edge) are measured similar to the four-point probe
method and can be corrected by comparing them with their reciprocal values.
The resistance of the sample is found using the van der Pauw formula:

! ! RA /RSample ! ! RB /RSample
e +e =1 (4.2)

The schematic illustration of both methods is depicted in figure 17.

Figure 17. Schematic display of a) van der Pauw method and b) four-
point probe measurement technique used for thin film electrical conductivity
estimation

4.2 Seebeck coefficient measurements

It follows from 2.2 that the Seebeck coefficient can be readily calculated
from the voltage and the temperature difference measured between the junctions
made of two dissimilar conductors. If the Seebeck coefficient of one of the
materials is precisely known as a function of temperature or negligibly small
compared to the total measured Seebeck coefficient, the absolute thermopower
of the second conductor can be estimated108. Despite the apparent simplicity of
the measurement, in practice it is associated with many pitfalls. Since the
Seebeck coefficient is not geometry dependent, the error coming from the
wrong estimation of the sample dimensions is not a concern. Caution should be
exercised when calculating the absolute Seebeck value of the unknown material
as it usually has an opposite sign as compared to the total Seebeck coefficient
sign109. The accuracy of the measurements can be impaired due to the contact
potential, sometimes not only the value but also the sign of the thermopower can

40
be misinterpreted. Thus, a good ohmic contact is required between the sample
and the probes; the contact resistance can be measured separately taking into
account different temperatures at which probes are kept.
The Seebeck coefficient is a temperature dependent parameter that
rarely has a linear dependence on T. It is important to know its variation in a
working T range as it can have a big effect on the thermoelectric efficiency. To
estimate the thermopower at a given T, the measurement has to be done at
different ∆T to make sure that ∆V plotted vs. ∆T is linear.
Providing a good thermal contact between a thermoelectric material and
a temperature sensor is another serious issue as far as the accuracy of the
Seebeck coefficient measurements is concerned. The heat exchange area
between the sample and the thermocouple has to be maximized to ensure the
uniformity of temperature profile. Thermally conducting pastes or other
adhesives as well as soldering can be used to improve the thermal contact,
however it still not sufficient in case of the measurements on thin films. A
regular temperature sensor of spherical shape cannot be implanted into the
sample due to its limited thickness. Moreover, the probes for voltage
measurement and the thermocouples have to be placed as close to each other as
possible since by definition both ∆V and ∆T are supposed to be estimated at the
same points, which requires a good electrical insulation between them.

Figure 18. Schematic illustration of the set-up used for Seebeck coefficient
measurements on thin films.

In order to avoid errors in temperature gradient estimation, the set-up

shown in Fig 18 can be used. The main idea is based on the utilization of thin
metal thermistors for accurate temperature measurements. The sample
comprises two symmetrical parts: the reference and the polymer film under

41
investigation. The whole structure is fixed between two metallic blocks
mounted on an electrically and thermally insulating platform and separated from
one another by a distance of several centimeters. One block is heated and
therefore acting as a hot side, while the other one can be cooled serving as a
heat sink. The sample is kept between the blocks and the four golden electrodes
are aligned and have the same separation distance. The two golden thermistors
on the reference sample are used for T measurements. The T of each line can be
determined from the calibration slope (T vs. resistance) specific to each
thermistor. The width of the lines is so small (≈35µm) that its resistance is very
sensitive to any changes in temperature. In addition, the thermistors are covered
with an insulating material with the thermal conductivity comparable to
polymer’s thermal conductivity to imitate the heat transfer processes on the real
film. The other pair of identical electrodes coated underneath the polymer film
is used for the Seebeck voltage measurements. Since the sample is symmetrical
and the separation between the electrodes as well as the material they are made
of are the same, one can assume that the T profile along these lines is constant
as long as the heating and cooling rate is uniform on both sides. The generated
thermoelectric voltage (ΔV) can be measured with nanovoltmeter keithley 2200,
with one probe on the hot and cold Au lines joined by the polymer film in
between. Knowing the T of each thermistor, the Seebeck coefficient is
calculated as ΔV/ ΔT.
Even though, the thermopower is conceptually easy to estimate,
potentially it represents a big challenge, especially in thin films.

4.3 Thermal conductivity measurements

Thermal conductivity measurement are considered to be the most

difficult to execute as far as high accuracy is concerned. In amorphous
materials, which is what conducting polymers are, the thermal conductivity is
often very small (≤2 W/mK). Consequently, the measurement becomes even
harder to perform, provided that among all the elements constituting the set-up
the measured material is the least preferable for the heat to flow through.
The 3ω technique introduced by Cahill in late 80’s is commonly used for
the thermal conductivity measurements in thin films, bulk materials and
liquids110. An AC current at frequency ω applied to a thin heater causes joule
heating at 2ω. The fluctuation in the resistance of the heater creates an
oscillating Joule heat of frequency 2ω. The metallic heater is calibrated similar
to the thermistors in the Seebeck coefficient measurements described in the
previous section.

42
 Figure 19. a) Circuit diagram for the 3ω voltage measurement. The
input signal is applied from a function generator; a bridge circuit is used to
minimize first harmonic voltage signal. RS represents the gold heater while R1,
R3 are standard resistors and R2 is a variable resistor. Differential inputs (A &
B) of the lock-in- amplifier are used to extract the 3ω voltage signal after
comparing with frequency of a reference signal from function generator. b)
Cross-section of the device used to measure the thermal conductivity. It is
composed of a glass substrate partially coated with a polymer layer having SiNx
insulation layer and gold heater (H1) on top. Gold heater (H2) of the same size
on top of glass substrate with SiNx serves as reference.

Thus, it is used as a heater and a thermometer simultaneously. The

fluctuation in the resistance of the heater at the 2nd harmonic creates a 3ω
component in the voltage. This 3ω voltage provides information about the
amount of heat dissipated through the underlying layers and can be used to
calculate thermal conductivity (k) of those layers.
For a metal heater with width 2b, the temperature oscillation (∆T) can be
found by Cahill’s simplified of heat diffusion equation:

P "
sin 2 (kb)
!T =
L! k
# (kb)2 (k 2 + q 2 )1/2
dk (4.3)
0

i2"
q=
D

43
 1/q is thermal penetration depth, D is thermal diffusivity and P/L is
power per unit length of heater. Experimentally ∆T can be found as:

dT R
!T = 2 V3! (4.4)
dR V1!

Where R, V1ω, V3ω are resistance, first and 3rd harmonic voltages across
the heater. Differential 3ω method has been frequently used to investigate
thermal conductivity of thin films. Two heaters of the same dimensions are
patterned on top of reference sample without film and another one having the
film under test. Temperature drop in thin film (∆Tf ) can be found by:

!Tf !T1 !T2

= " (4.5)
P1 P1 P2

∆T1 and ∆T2 are the temperatures on the substrate and the film and the
substrate only, P1 and P2 are the respective power dissipation. An insulation
layer is added if the film is conducting.
The thermal conductivity is estimated using the equation111:

P1h f
k= (4.6)
2b' L!Tf

where hf and L are the thickness and the length of the film,
2b’=2b+0.76 hf is the effective line width of the heater.
Thermal conductivity characterization with 3ω technique is an
advantageous method for a number of reasons. Firstly, it is an AC measurement
where radiation effects are minimized and secondly, it readily provides the
thermal conductivity temperature dependence. In amorphous materials, such as
conjugated polymers, associated with high anisotropy of physical properties, it
is crucial to measure both vertical and lateral thermal conductivity. With 3ω
technique it can be done using the set-up shown in figure 19b.

44
 Chapter 5 - Goal of the thesis

This thesis provides an insight into the emerging field of organic

thermoelectrics, more specifically, thermoelectric power generation based on
conducting polymers as functional materials. Starting from the basic principles
of thermoelectrics, containing thermodynamics, solid-sate physics and
thermoelectric energy conversion aspects, the discussion is followed by the
section describing physical and electronic properties of conducting polymers as
well as the charge transport mechanisms characteristic to these materials. Paper
1 serves as the main introduction to organic thermoelectrics by merging together
conducting polymers and thermoelectrics. The review provides a brief
discussion of electronic and thermoelectric properties of various conducting
polymers along with an overview of the progress in the field of polymer-based
thermoelectrics. The potentials of organic thermoelectric materials as well as the
strategies of thermoelectric efficiency improvement are examined. Paper 2
attempts to compare and comprehend thermoelectric properties of some PEDOT
derivatives by looking into their morphology and structure as the origin of
different conduction mechanism in seemingly similar materials. Besides, the
effect of secondary doping on thermoelectric properties is studied. Paper 3
concerns with optimization of the thermoelectric efficiency of PEDOT-PSS
used as an active material in an electrochemical transistor. Applied gate
potential acts as a tuning parameter that induces a certain oxidation state in the
organic semiconductor and ultimately enables to obtain an optimum zT. In
Paper 4 another PEDOT derivative i.e. PEDOT-Tos is gradually reduced with
Tetrarkis(dimethylamino)ethylene (TDAE) from its highly doped state (ca. one
charge per three monomers) down to lower doping levels. The thermoelectric
efficiency reaches 0.25 at room T, which is almost comparable with the best
inorganic thermoelectrics. Finally Paper 5 shows another PEDOT-Tos zT
optimization method, this time using acid/base chemistry. This approach can be
seen as an alternative to chemical doping. The doping level in the latter process
can be challenging to control precisely, moreover it may involve highly reactive
and therefore unstable in air dopants. Changing pH of the solution with which
the polymer is treated, the thermoelectric properties of PEDOT-Tos can be
altered in a manner similar to chemical doping. Acidic pH increases electrical
conductivity without affecting thermopower, while basic pH reduces the
polymer and promotes higher thermopower values. As a result zT is optimized
at a certain pH value of the used solution.
The ultimate goal of the thesis is to demonstrate high potentials of
conducting polymers in thermoelectric energy conversion applications. PEDOT
derivatives, especially PEDOT-Tos are one of those materials that can exhibit
fairly high thermoelectric efficiency if properly modified. The polymers can be

45
easily functionalized in different sizes and shapes to pattern organic TEGs on a
variety of different substrates. Being a p-type material, PEDOT cannot be used
for both types of thermoelectric element and a stable n-type polymer is required
to complete the converter. The solution can be perhaps found in the organic-
inorganic symbiosis, not only in order to obtain n-type materials but also for the
sake of higher zT.

46
 Chapter 6 – Conclusions and perspective

Organic thermoelectrics is a rapidly emerging field that is yet to evolve

into a strong, independent and clearly oriented research area. In the
“department” of conventional thermoelectrics that is overly inorganic,
conducting polymers have not been recognized as potentially suitable materials
for thermoelectric applications and rarely mentioned in the related literature.
But things are certainly changing, as the advantages offered by this new
approach are considerable as compared to inorganic thermoelectrics. Loosing
the battle for high efficiency, organic materials can compensate for it with their
good processibility, mechanical flexibility, low material and manufacturing
cost, variety of synthesis techniques and light weightiness of the polymer based
devices. The efficiency problem can be taken care of by controlling the structure
and morphology of polymers in order to achieve higher crystallinity and
therefore improve the charge transport in the materials. Optimizing the
oxidation level in conducting polymers enables to achieve the balance of
thermoelectric properties and acquire favorable DOS at the Fermi level. Extra
improvement can be obtained by decreasing lattice thermal conductivity in
conducting polymers. It can be done by incorporating scattering centers into the
polymer structure in the form of voids, pores or any other inclusions of
nanosize. Equivalently, inorganic nanostructures with high Seebeck coefficient
or electrical conductivity can be functionalized within polymer matrix in order
to enhance the overall zT. Polymers themselves can form nanostructures, which
allows for DOS engineering and much higher thermopower can be expected in
such materials.
All in all, there are numerous directions in which organic thermoelectrics
can potentially progress. There is no doubt that in the years to come it shouldn’t
be and won’t be abandoned because of the constantly growing need in energy.
Based on today’s energy consumption patterns, it’s clearly just a matter of time
before the renewable energy resources become a substantial part of the total
energy supply and at that point organic thermoelectrics shall not be overlooked.

47
 References:
1 Vining, C. B. An inconvenient truth about thermoelectrics. Nature Materials 8, 83-85
(2009).
2 H. Yan, N. S., N. Toshima. Thermal transporting properties of electrically conductive
polyaniline films as organic thermoelectric materials. Journal of Thermal Analysis
and Calorimetry 69, 6 (2002).
3 Jiang, F.-X. et al. Thermoelectric Performance of Poly(3,4-ethylenedioxythiophene):
Poly(styrenesulfonate). Chinese Physics Letters 25, 2202 (2008).
4 Park, Y. W., Han, W. K., Choi, C. H. & Shirakawa, H. Metallic nature of heavily
doped polyacetylene derivatives: Thermopower. Physical Review B 30, 5847-5851
(1984).
5 Mateeva, N., Niculescu, H., Schlenoff, J. & Testardi, L. R. Correlation of Seebeck
coefficient and electric conductivity in polyaniline and polypyrrole. Journal of
Applied Physics 83, 3111-3117 (1998).
6 Kemp, N. T. et al. Thermoelectric power and conductivity of different types of
polypyrrole. Journal of Polymer Science Part B: Polymer Physics 37, 953-960 (1999).
7 Lévesque, I. et al. Synthesis and Thermoelectric Properties of Polycarbazole,
Polyindolocarbazole, and Polydiindolocarbazole Derivatives. Chemistry of Materials
19, 2128-2138, doi:10.1021/cm070063h (2007).
8 Kaiser, A. B. Thermoelectric power and conductivity of heterogeneous conducting
polymers. Physical Review B 40, 2806-2813 (1989).
9 See, K. C. et al. Water-Processable Polymer−Nanocrystal Hybrids for
Thermoelectrics. Nano Letters 10, 4664-4667, doi:10.1021/nl102880k (2010).
10 Kim, D., Kim, Y., Choi, K., Grunlan, J. C. & Yu, C. Improved Thermoelectric
Behavior of Nanotube-Filled Polymer Composites with Poly(3,4-
ethylenedioxythiophene) Poly(styrenesulfonate). ACS Nano 4, 513-523,
doi:10.1021/nn9013577 (2009).
11 Seebeck, T. J. Magnetic polarization of metals and minerals. Abhandlungen der
Deutschen Akademie der Wissenschaften zu Berlin, 289-346 (1822).
12 Kasap, S. O. Thermoelectric effects in metals: thermocouples. (2001).
13 Pollock, D. D. Thermoelectricity: Theory, Thermometry, Tool. (Astm Intl, 1985).
14 Mahan, G. D. in Solid State Physics Vol. Volume 51 (eds Ehrenreich Henry &
Spaepen Frans) 81-157 (Academic Press, 1997).
15 Blatt, F. J. a. S., Peter A and Foiles, Carl L. and Denis, Greg. Thermoelectric power of
metals. (Plenum Press, 1976).
16 Roeser, W. F. Thermoelectric Thermometry. Journal of Applied Physics 11, 388-407
(1940).
17 Ioffe, A. F. Semiconductor Thermoelements and Thermoelectric Cooling.
(Infosearch, 1957).
18 Jordan, F. W. The Thomson and Peltier effects [1]. Nature 86, 380 (1911).
19 Goldsmid, H. J. Thermoelectric refrigeration. (Plenum Press, 1964).
20 Starling, S. G. Demonstration of Peltier and Thomson effects [11]. Nature 85, 512
(1911).
21 Sherman, B., Heikes, R. R. & Ure, J. R. W. Calculation of Efficiency of
Thermoelectric Devices. Journal of Applied Physics 31, 1-16 (1960).
22 Callen, H. B. Principle of minimum entropy production. Physical Review 105, 360-
365 (1957).

48
23 Mazur, S. R. d. G. a. P. non-equilibrium thermodynamics (Dover Publications Inc.,
1985).
24 Onsager, L. Reciprocal relations in irreversible processes. I. Physical Review 37, 405-
426 (1931).
25 Arnas, O. A. & Miller, D. L. ON AN IRREVERSIBLE THERMODYNAMIC
ANALYSIS OF THERMOELECTRIC DEVICES. 36-40 (1978).
26 Domenicali, C. A. Irreversible Thermodynamics of Thermoelectricity. Reviews of
Modern Physics 26, 237-275 (1954).
27 Armen M. Gulian, G. F. Z. Nonequilibrium Electrons and Phonons in
Superconductors (Selected Topics in Superconductivity). (Springer, 1999).
28 C. Kittel, H. K. Thermal Physics. (W. H. Freeman; Second Edition edition 1980).
29 Callaway, J. Model for Lattice Thermal Conductivity at Low Temperatures. Physical
Review 113, 1046-1051 (1959).
30 Ziman, J. M. Electrons and Phonons: The Theory of Transport Phenomena in Solids
(Oxford: Clarendon Press, 2001).
31 Reif, F. Fundamentals of Statistical and Thermal Physics. (Waveland Pr Inc 2008).
32 Snyder, G. J., Toberer, Eric S. Complex thermoelectric materials. Nature Materials 7,
105-114 (2008).
33 Tritt, T. M. Advances in Thermoelectric Materials: Pt. I (Semiconductors and
Semimetals). 326 (Academic Press Inc, 2000).
34 Dughaish, Z. H. Lead telluride as a thermoelectric material for thermoelectric power
generation. Physica B: Condensed Matter 322, 205-223, doi:10.1016/s0921-
4526(02)01187-0 (2002).
35 Kanatzidis, M. G. in Semiconductors and Semimetals Vol. Volume 69 (ed M. Tritt
Terry) 51-100 (Elsevier, 2001).
36 Tritt, T. M. Thermoelectric Phenomena, Materials, and Applications. Annual Review
of Materials Research 41, 433-448, doi:10.1146/annurev-matsci-062910-100453
(2011).
37 M., K. S. V. a. S. F. Thermal Conductivity of Diamond Composites. Materials 2,
2467-2495 (2009).
38 Bentien, A. et al. Thermal conductivity of thermoelectric clathrates. Physical Review
B 69, 045107 (2004).
39 R. Venkatasubramanian, E. S., T. Colpitts and B. O'Quinn. Thin-film thermoelectric
devices with high room-temperature figures of merit. Nature 413, 597-602 (2001).
40 Min, G. & Rowe, D. M. Cooling performance of integrated thermoelectric
microcooler. Solid-State Electronics 43, 923-929, doi:10.1016/s0038-1101(99)00045-
3 (1999).
41 Poudel, B. et al. High-Thermoelectric Performance of Nanostructured Bismuth
Antimony Telluride Bulk Alloys. Science 320, 634-638 (2008).
42 Fleurial, J.-P. in Short Course of Thermoelectrics (ITS, Japan, 1993).
43 Yim, W. M. & Rosi, F. D. Compound tellurides and their alloys for peltier cooling—
A review. Solid-State Electronics 15, 1121-1140, doi:10.1016/0038-1101(72)90172-4
(1972).
44 Kuznetsov, V. L., Kuznetsova, L. A., Kaliazin, A. E. & Rowe, D. M. Preparation and
thermoelectric properties of A[sub 8][sup II]B[sub 16][sup III]B[sub 30][sup IV]
clathrate compounds. Journal of Applied Physics 87, 7871-7875 (2000).
45 Poon, G. J. in Semiconductors and Semimetals Vol. Volume 70 (ed M. Tritt Terry)
37-75 (Elsevier, 2001).
46 Fleurial J-P, B. A., Caillat T, Morelli D.T, Meisner G.P. in 15th International
Conference on Thermoelectrics. 91-95.

49
47 Luo, W. et al. Rapid synthesis of high thermoelectric performance higher manganese
silicide with in-situ formed nano-phase of MnSi. Intermetallics 19, 404-408,
doi:10.1016/j.intermet.2010.11.008 (2011).
48 Shikano, M. & Funahashi, R. Electrical and thermal properties of single-crystalline
(Ca[sub 2]CoO[sub 3])[sub 0.7]CoO[sub 2] with a Ca[sub 3]Co[sub 4]O[sub 9]
structure. Applied Physics Letters 82, 1851-1853 (2003).
49 Sun, Y. et al. Organic Thermoelectric Materials and Devices Based on p- and n-Type
Poly(metal 1,1,2,2-ethenetetrathiolate)s. Advanced Materials 24, 932-937 (2012).
50 Toshima, N. Conductive polymers as a new type of thermoelectric material.
Macromolecular Symposia 186, 81-86 (2002).
51 Riffat, S. B. & Ma, X. Thermoelectrics: a review of present and potential applications.
Applied Thermal Engineering 23, 913-935, doi:10.1016/s1359-4311(03)00012-7
(2003).
52 Basel I. Ismail, W. H. A. Thermoelectric Power Generation Using Waste-Heat Energy
as an
Alternative Green Technology. Recent Patents on Electrical Engineering 2, 27-39 (2009).
53 Yu, G. G. & Logvinov, G. N. Physics of thermoelectric cooling. Semiconductor
Science and Technology 20, R57 (2005).
54 Snyder, G. J. in Thermoelectrics Handbook 9-1-9-26 (CRC Press, 2005).
55 Müller, E., Zabrocki, K., Goupil, C., Snyder, G. & Seifert, W. in Materials,
Preparation, and Characterization in Thermoelectrics 1-36 (CRC Press, 2012).
56 Rowe, D. M. & Min, G. Evaluation of thermoelectric modules for power generation.
Journal of Power Sources 73, 193-198, doi:10.1016/s0378-7753(97)02801-2 (1998).
57 Bredas, J. L. & Street, G. B. Polarons, bipolarons, and solitons in conducting
polymers. Accounts of Chemical Research 18, 309-315, doi:10.1021/ar00118a005
(1985).
58 Tessler, N. (1999).
59 Su, W. P., Schrieffer, J. R. & Heeger, A. J. Solitons in Polyacetylene. Physical Review
Letters 42, 1698-1701 (1979).
60 Heeger, A. J., Kivelson, S., Schrieffer, J. R. & Su, W. P. Solitons in conducting
polymers. Reviews of Modern Physics 60, 781-850 (1988).
61 Brédas, J. L., Wudl, F. & Heeger, A. J. Polarons and bipolarons in doped
polythiophene: A theoretical investigation. Solid State Communications 63, 577-580,
doi:10.1016/0038-1098(87)90856-8 (1987).
62 Fahlman, M. & Salaneck, W. R. Surfaces and interfaces in polymer-based electronics.
Surface Science 500, 904-922, doi:10.1016/s0039-6028(01)01554-0 (2002).
63 Reiss, H. Theoretical analysis of protonic acid doping of the emeraldine form of
polyaniline. The Journal of Physical Chemistry 92, 3657-3662,
doi:10.1021/j100323a064 (1988).
64 MacDiarmid, A. G. “Synthetic Metals”: A Novel Role for Organic Polymers (Nobel
Lecture). Angewandte Chemie International Edition 40, 2581-2590 (2001).
65 MacDiarmid, A. G. et al. The Concept of `Doping' of Conducting Polymers: The Role
of Reduction Potentials [and Discussion]. Philosophical Transactions of the Royal
Society of London. Series A, Mathematical and Physical Sciences 314, 3-15 (1985).
66 Chandrasekhar, P. Conducting Polymers, Fundamentals and Applications: A
Practical Approach. (Springer, 1999).
67 Heeger, A. J. Semiconducting and Metallic Polymers: The Fourth Generation of
Polymeric Materials (Nobel Lecture). Angewandte Chemie International Edition 40,
2591-2611, doi:10.1002/1521-3773(20010716)40:14<2591::aid-anie2591>3.0.co;2-0
(2001).

50
68 Edman, L., Summers, M. A., Buratto, S. K. & Heeger, A. J. Polymer light-emitting
electrochemical cells: Doping, luminescence, and mobility. Physical Review B 70,
115212 (2004).
69 Nilsson, D. et al. Bi-stable and Dynamic Current Modulation in Electrochemical
Organic Transistors. Advanced Materials 14, 51-54, doi:10.1002/1521-
4095(20020104)14:1<51::aid-adma51>3.0.co;2-# (2002).
70 Zhuo, J.-M. et al. Direct Spectroscopic Evidence for a Photodoping Mechanism in
Polythiophene and Poly(bithiophene-alt-thienothiophene) Organic Semiconductor
Thin Films Involving Oxygen and Sorbed Moisture. Advanced Materials 21, 4747-
4752, doi:10.1002/adma.200901120 (2009).
71 Hamadani, B. H., Corley, D. A., Ciszek, J. W., Tour, J. M. & Natelson, D. Controlling
Charge Injection in Organic Field-Effect Transistors Using Self-Assembled
Monolayers. Nano Letters 6, 1303-1306, doi:10.1021/nl060731i (2006).
72 Chiang, J.-C. & MacDiarmid, A. G. ‘Polyaniline’: Protonic acid doping of the
emeraldine form to the metallic regime. Synthetic Metals 13, 193-205,
doi:10.1016/0379-6779(86)90070-6 (1986).
73 Kaiser, A. B. Systematic Conductivity Behavior in Conducting Polymers: Effects of
Heterogeneous Disorder. Advanced Materials 13, 927-941, doi:10.1002/1521-
4095(200107)13:12/13<927::aid-adma927>3.0.co;2-b (2001).
74 Ohad, L. & David, S. Macroscopic disorder and the metal - insulator transition in
conducting polymers. Journal of Physics: Condensed Matter 9, L599 (1997).
75 Prigodin, V. N. & Epstein, A. J. Nature of insulator–metal transition and novel
mechanism of charge transport in the metallic state of highly doped electronic
polymers. Synthetic Metals 125, 43-53, doi:10.1016/s0379-6779(01)00510-0 (2001).
76 Kohlman, R. S. et al. Limits for Metallic Conductivity in Conducting Polymers.
Physical Review Letters 78, 3915-3918 (1997).
77 Kim, N. et al. Role of Interchain Coupling in the Metallic State of Conducting
Polymers. Physical Review Letters 109, 106405 (2012).
78 Kline, R. & McGehee, M. Morphology and Charge Transport in Conjugated
Polymers. Journal of Macromolecular Science, Part C: Polymer Reviews 46, 27-45,
doi:10.1080/15321790500471194 (2006).
79 Kramer, B. M., A. Localization: theory and experiment. Reports on Progress in
Physics 56, 1469-1564 (1993).
80 Mott, N. F. Conduction in glasses containing transition metal ions. Journal of Non-
Crystalline Solids 1, 1-17, doi:10.1016/0022-3093(68)90002-1 (1968).
81 Anderson, P. W. Absence of Diffusion in Certain Random Lattices. Physical Review
109, 1492-1505 (1958).
82 K. P. Ghatak, A. K. C., S. Ghosh, and A. N. Chakravarti. Physica Status Solidi B-
Basic Research 99, K55-K60 (1980).
83 Prigodin, V. N. & Epstein, A. J. Quantum hopping in metallic polymers. Physica B:
Condensed Matter 338, 310-317, doi:10.1016/j.physb.2003.08.011 (2003).
84 Alan, J. H. The Critical Regime of the Metal-Insulator Transition in Conducting
Polymers: Experimental Studies. Physica Scripta 2002, 30 (2002).
85 Kaiser, A. B., Flanagan, G. U., Stewart, D. M. & Beaglehole, D. Heterogeneous
model for conduction in conducting polymers and carbon nanotubes. Synthetic Metals
117, 67-73, doi:10.1016/s0379-6779(00)00540-3 (2001).
86 Yoon, C. O. et al. Hopping transport in doped conducting polymers in the insulating
regime near the metal-insulator boundary: polypyrrole, polyaniline and
polyalkylthiophenes. Synthetic Metals 75, 229-239, doi:10.1016/0379-
6779(96)80013-0 (1995).

51
87 Zuppiroli, L., Bussac, M. N., Paschen, S., Chauvet, O. & Forro, L. Hopping in
disordered conducting polymers. Physical Review B 50, 5196-5203 (1994).
88 Maji, S., Mukhopadhyay, S., Gangopadhyay, R. & De, A. Smooth crossover from
variable-range hopping with Coulomb gap to nearest-neighbor interchain hopping in
conducting polymers. Physical Review B 75, 073202 (2007).
89 Roth, S., Bleier, H. & Pukacki, W. Charge transport in conducting polymers. Faraday
Discussions of the Chemical Society 88, 223-233 (1989).
90 Epstein, A. J., Lee, W. P. & Prigodin, V. N. Low-dimensional variable range hopping
in conducting polymers. Synthetic Metals 117, 9-13, doi:10.1016/s0379-
6779(00)00531-2 (2001).
91 Mott, N. F. Electronic processes in non-crystalline materials. (Clarendon Press,
1971).
92 Sheng, P., Abeles, B. & Arie, Y. Hopping Conductivity in Granular Metals. Physical
Review Letters 31, 44-47 (1973).
93 Elschner, A. & Kirchmeyer, S. in Organic Photovoltaics 211-242 (Wiley-VCH
Verlag GmbH & Co. KGaA, 2009).
94 Levermore, P. A. et al. High efficiency organic light-emitting diodes with PEDOT-
based conducting polymer anodes. Journal of Materials Chemistry 18, 4414-4420
(2008).
95 Latessa, G., Brunetti, F., Reale, A., Saggio, G. & Di Carlo, A. Piezoresistive
behaviour of flexible PEDOT:PSS based sensors. Sensors and Actuators B: Chemical
139, 304-309, doi:10.1016/j.snb.2009.03.063 (2009).
96 Okuzaki, H., Suzuki, H. & Ito, T. Electrically driven PEDOT/PSS actuators. Synthetic
Metals 159, 2233-2236, doi:10.1016/j.synthmet.2009.07.054 (2009).
97 Nilsson, D. An organic electrochemical transistor for printed sensors and logic,
Univ., (2005).
98 Bolin, M. H. et al. Nano-fiber scaffold electrodes based on PEDOT for cell
stimulation. Sensors and Actuators B: Chemical 142, 451-456,
doi:10.1016/j.snb.2009.04.062 (2009).
99 Aleshin, A. N., Kiebooms, R. & Heeger, A. J. Metallic conductivity of highly doped
poly(3,4-ethylenedioxythiophene). Synthetic Metals 101, 369-370, doi:10.1016/s0379-
6779(98)00758-9 (1999).
100 Gustafsson, J. C., Liedberg, B. & Inganäs, O. In situ spectroscopic investigations of
electrochromism and ion transport in a poly (3,4-ethylenedioxythiophene) electrode in
a solid state electrochemical cell. Solid State Ionics 69, 145-152, doi:10.1016/0167-
2738(94)90403-0 (1994).
101 T.A. Skotheim, J. R. Handbook of conducting polymers. (CRC Press, 2007).
102 Crispin, X. et al. The Origin of the High Conductivity of Poly(3,4-
ethylenedioxythiophene)−Poly(styrenesulfonate) (PEDOT−PSS) Plastic Electrodes.
Chemistry of Materials 18, 4354-4360, doi:10.1021/cm061032+ (2006).
103 Na, S.-I. et al. Evolution of nanomorphology and anisotropic conductivity in solvent-
modified PEDOT:PSS films for polymeric anodes of polymer solar cells. Journal of
Materials Chemistry 19, 9045-9053 (2009).
104 Reyes-Reyes, M., Cruz-Cruz, I. & López-Sandoval, R. n. Enhancement of the
Electrical Conductivity in PEDOT:PSS Films by the Addition of Dimethyl Sulfate.
The Journal of Physical Chemistry C 114, 20220-20224, doi:10.1021/jp107386x
(2010).
105 Xia, Y. & Ouyang, J. Significant Conductivity Enhancement of Conductive Poly(3,4-
ethylenedioxythiophene): Poly(styrenesulfonate) Films through a Treatment with

52
 Organic Carboxylic Acids and Inorganic Acids. ACS Applied Materials & Interfaces
2, 474-483, doi:10.1021/am900708x (2010).
106 Winther-Jensen, B., Chen, J., West, K. & Wallace, G. Vapor Phase Polymerization of
Pyrrole and Thiophene Using Iron(III) Sulfonates as Oxidizing Agents.
Macromolecules 37, 5930-5935, doi:10.1021/ma049365k (2004).
107 Pauw, L. J. v. d. A Method of Measuring Specific Resistivity and Hall Effect of Discs
of Arbitrary Shapes. Phillips Research Reports 13, 1-9 (1958).
108 Bentley, R. E. Handbook of Temperature Measurement Vol. 3: The Theory and
Practice of Thermoelectric Thermometry: The Theory and Practice of Thermoelectric
Thermometry. Vol. 3 (Springer, 1998).
109 Nishida, I. A. in CRC Handbook of Thermoelectrics (CRC Press, 1995).
110 Cahill, D. G. & Pohl, R. O. Thermal conductivity of amorphous solids above the
plateau. Physical Review B 35, 4067-4073 (1987).
111 Koh, Y. K. et al. Comparison of the 3 omega method and time-domain
thermoreflectance for measurements of the cross-plane thermal conductivity of
epitaxial semiconductors. Journal of Applied Physics 105, 054303-054307 (2009).

53