Thin-films for Thermoelectric Applications 1 Introduction

The need for production of electricity by mankind has intensified and is likely to
M. F. Silva, J. F. Ribeiro, J. P. Carmo, L. M. Goncalves, and J. H. Correia increase. Much of the actions of human beings inevitably require a power source,
University of Minho - Dept. Industrial Electronics whether in food, transportation or work. The energy sources from fossil materials shall
4800-058 Azurem, Guimaraes, PORTUGAL disappear, and therefore, the solution in generating electricity lies from renewable
{fsilva, jribeiro, jcarmo, lgoncalves, [Link]}@[Link] sources. Renewable energy solutions exist to date, and the one that brings a greater
potential is the thermal/geothermal. Not always the source is available in solutions for
Abstract renewable energies such as sunlight, wind or water. The increase in global temperature
of the Earth produces a great potential for generating electricity by heat released or
The introduction of nanotechnology and new horizons enabled previously unattainable waste heat. The phenomenon of thermoelectricity may be the solution to be adopted in
by thermoelectric devices. The nano-scale phenomena began to be exploited through renewable energy thermal/geothermal applications. One of the challenges that mobile
techniques of thin-film depositions to increase the efficiency of thermoelectric films. devices face is the need to obtain a reliable source of power, beyond the limitations
This chapter reviews the fundamentals of the phenomenon of thermoelectricity and its imposed by the batteries. The solution lies in devices capable of recovering energy from
evolution since it was discovered in 1822. This chapter also makes a review about the the environment surrounding the device or the user. The human body radiates heat
thermoelectric devices the macro to nano devices, where it is described the most used continuously and the devices that are in direct contact can recovery the waste heat, i.e.,
techniques of physical vapor depositions to deposit thermoelectric thin-films. Then, it is the thermoelectric generators (TEG). Until now TEGs have low efficiency, a factor that
presented a custom made deposition chamber for depositing thermoelectric thin-films has been changing in recent years with the use of thin-film deposition techniques and
by the thermal co-evaporation technique, where the construction issues and the exploration of phenomena at the nano-scale. It thus becomes important to build
specifications are discussed. All the steps for obtaining a thermoelectric generator in systems capable of producing thin films for flexible microsystems for production of
flexible substrate with the custom deposition chamber (to incorporate in thermoelectric electricity.
microsystem) are described. The aim of thermoelectric microsystem relays to introduce
in a human energy harvesting applications to power wireless sensor networks (WSN) or 1.1 Thermoelectricity
biomedical devices. The scanning probe measuring systems for characterization of the
thermoelectric thin-films are also described in this chapter. Finally, a few number of In 1822 Thomas Seebeck discovered that if applied a temperature gradient between the
prototypes of thermoelectric thin-films (made of bismuth and antimony tellurides, junction of two different materials, occurs an voltage like in a temperature sensor, and
Bi2Te3 and Sb2Te3, respectively) were obtained by co-evaporation (using the custom consequently by an electric current in a closed circuit as a thermoelectric generator (in
made deposition chamber) and characterized for quality assessment. All of the issues Fig. 1) [1]. The Seebeck effect () is given by the difference of temperature (ΔT) which
involved in the co-evaporation and characterization are object of analysis in this causes a voltage difference (ΔV).
chapter.
∆!
= , [µVK %& ] (1)
∆"

Key words: Thin-films, thermoelectricity, energy harvesting.

Fig. 1. Seebeck experiment; (a) Temperature sensor; (b) Thermoelectric generator.

Only 12 years later Jean Peltier found that in the junction of two materis – bismuth (Bi) -
and Antimony (Sb) – thermic energy is released or absorbed depending on the electric * = + , [K %& ] (3)
(
current [2]. Although Peltier used the Seebeck effect in his experiments, he could not
estimate the foundations of their observations or relate to the effect Seebeck's findings. Where (-) represents the electrical conductivity inverse to the electrical resistivity ())
Only in 1838 Heinrich Lenz was able to explain the nature of the Peltier effect with an and given by.
simple experiment (Fig. 2) [3,4]. He put a drop of water on top of the junction of 1
- = , [Ωm%& ] (4)
bismuth (Bi) and antimony (Sb). He managed to freeze the water drop and changing the )
direction of electric current between junctions, it melted. As the figure of merit can vary with temperature it was later appointed by (*").
+
*" = " (5)
)(

1.2 Thermoelectric devices

Until around 1930 the thermoelectric materials had a Seebeck coefficient in the order of
10 µVK-1 or below and with efficiency of 1% therefore a weak source of electrical
energy. In 1947 Mária Telkes gave another boost by using synthetic thermoelectric
semiconductors with a Seebeck coefficient of 100 µVK-1 and efficiency of 5 % [11].
During the beginning of 1900 Abram Ioffe studied the theory of thermoelectric
materials with semiconductor published in 1949 [12]. He encouraged his PhD student
Yuri Maslakovets to truly develop the first thermoelectric generator (TEG) efficiently
than 5% through the semiconductor material lead sulfide (PbS). His PhD thesis [13] was
during World War II a secret document because their study allowed the development of
a TEG that fed a Soviet military radio station for 400 hours and produced 12 W with a
Fig. 2 - Lenz’s experiment by Peltier effect; (a) Cooling; (b) Heating. temperature gradient of ΔT > 300 ºC. The Ioffe’s group formulated the principles for
thermoelectric devices that are used up to today. In 1950 Ioffe and another student
Thus Lenz made the first thermoelectric device. In 1851 William Thomson - better Lazar Stilbans presents the first thermoelectric cooling device (TER) with the best
known as Lord Kelvin and the absolute temperature scale - proposed in 1848 [5] to thermoelectric material so far by joining bismuth, antimony and tellurium (BiSbTe)
establish the Peltier coefficient (') may be related to the Seebeck coefficient () using with a *= 2 x 10-3 K-1 [14-16]. In 1952 Stilbans presents for the first time the bismuth
the following formula. telluride (Bi2Te3) which later came to be used as material in the n-type thermoelectric
structures and published only in 1956 with Ioffe due the Soviet secrecy [17]. In 1954
H. Julian Goldsmid and R. W. Douglas presented the first thermoelectric device capable
' = " , [V] (2) of cooling below 0 ºC from room temperature [18]. Until then, the compound of
bismuth telluride (Bi2Te3) had the highest figure of merit (*) with many potential
applications for power generation and cooling but their efficiency is reduced at
Where " represents the temperature of in junction of the materials [6]. Thomson temperatures above 200 ºC. It was then studied a silicon germanium compound (SiGe)
predicted the existence of a third effect. The Thomson effect defines the cooling and which although not present a higher figure of merit (*) but have operating temperatures
heating of an electrical conductor in the presence of electric current when exposed to a much higher in the order of 730 ºC to 1000 ºC [7]. The first studies of the thermoelectric
gradient temperature. It is generally despised for being well below the Peltier effect [7]. materials at high temperatures were conducted and published in 1955 and 1958 [19,20]
In 1885 John William Strutt also known as the 3rd Baron Rayleigth (Nobel Prize in by I. M. Templeton and his group. R. B. Roberts reformulated them and published in
physics 1904) used the phenomenon of thermoelectric power generation but with 1977 [21].
incorrect calculations and regarding the efficiency of thermoelectric generator [8].
Studies on the satisfactory thermoelectric generation and cooling were achieved by
Edmund Altenkirch in 1909 and 1911 [9,10]. Altenkirch concluded that good
1.2.1 Macro-devices
thermoelectric materials needed to have a high Seebeck coefficient () but with low
The great breakthrough in semiconductor technology has allowed in the 60 decade the
thermal conductivity (() to retain heat in the junction, and low electrical resistivity ())
development of modern technologies of communication and data acquisition which
to minimize the Joule heating. These properties would then be designated to
were meeting with the characteristics of thermoelectric generators. Equipment’s such as
characterize the thermoelectric materials by assigning the name of figure of merit (*).
radio transmission towers in remote locations need a trustworthy and reliable energy
supply. Without the thermoelectric generators was not possible due to the limitations of
batteries. The solar cells were a very new technology and inefficient, motor generators
did not last long or the construction of power lines was too expensive. All these factors
contributed to the use of thermoelectric power generation and as an independent source
for seismographs, weather stations, and radio transmission towers in isolated places like
Mount St. Helens (Alaska - USA) and gas platforms [7]. Thermoelectric generators
were used in space in the 60’s to 90’s missions like Apollo, Voyager, Galileo and
Ulysses. The radioisotope thermoelectric generators (RTG) were used in space and in
isolated environments on Earth. The heat obtained from the decay of plutonium (Pu)
and strontium (Sr) where are used in nuclear auxiliary generators (SNAP) and the SiGe
as thermoelectric material to withstand the high operating temperatures [7,22,23]. Since
then important patents have been registered for the manufacture of thermoelectric Fig. 4 - BSST thermoelectric project to introduce in a BMW 5 series.
devices [24-26] that culminated in a modern thermoelectric device (Fig. 3). The
thermoelectric device is constituted by semiconductor thermoelectric materials The human body radiates heat continuously which promotes many thermoelectric
(alternating between p-type and n-type), electrically connected in series and thermally in applications, ie, TEG in wristwatches, like Seiko Thermic and Citizen Eco-Drive
parallel by metal contacts. Two ceramic plates are electrical insulators but good thermal Thermo [30,31].
conductive and serve as substrate for the thermoelectric structures. The devices are
silent, do not require maintenance and are very robust.

Fig. 5 - Thermoelectric wristwatches; (a) Seiko Thermic; (b) Citizen Eco-drive Thermo.

1.2.2 Micro-devices
Fig. 3 - A modern thermoelectric device.
The first attempt to miniaturize a thermoelectric generator was carried out by
D. M. Rowe [32] and patented in 1988 [33]. With deposition techniques of thin-films
Thermoelectric devices have been used in numerous areas. With efforts to reduce CO2
thermoelectric took the interest of other researchers and it was possible to develop
emissions by the automotive industry the research company IAV GmbH presents for the
small-scale thermoelectric devices. The Micropelt was among the first companies to
first time in 1986 the Eco-Polo with thermoelectric generators in the exhaust pipe. In
commercialize thermoelectric micro devices. It is used a semiconductor material
2004 the BSST develop to BMW a similar system capable of producing 1 kW [27-29].
(Bi2Te3) n-type and p-type deposited in a separate silicon wafer. Later the two substrates
are bonded resulting in the micro device with 540 junctions and capable of generating
approximately 24 W/cm2 with a temperature gradient of 120 ºC [34-36].
Fig. 6 - Micro thermoelectric device - Micropelt MPG-D751.

Fig. 8 - Thermoelectric projects form do IMEC; (a) Pulse oximeter powered by a TEG with power
The Nextreme Thermal Solutions Inc. sells micro thermoelectric devices with consumption of 62 µW at 22ºC; (b) Body-powered ECG headband. [42].
technology developed by the RTI laboratories (Research Triangle Institute).
D. D. L. Wijngaards presents in his PhD thesis a micro thermoelectric membranes
developed through a process compatible with CMOS fabrication. It was used the
polycrystalline SiGe (polySiGe) and polycrystalline Si (polySi) has the thermoelectric
materials and deposited by CVD (Chemical Vapor Deposition). The figure of merit
obtained for the n-type has 0.203 mK-1 and 0.123 mK-1 for the p-type with maximum
with maximum gradient temperature of 2 ºC [43]. Also in the PhD thesis of L. da Silva
is shown a cooling thermoelectric device (Fig. 9(a)). L. da Silva used the co-evaporation
technique to deposit thin-films thermoelectric materials of Bi2Te3 and Sb2Te3 with the
desired stoichiometry. The thermoelectric materials are characterized during the
deposition. It was achieved a stable temperature of the device of 63 °C from ambient
temperature and resistance of 51 Ω at 23 mA of electric current [44].
Fig. 7 - Thermoelectric devices from Nextreme; (a) OptoCooler UPF4 with cooling capacity of
112 mW/cm2; (b) eTEG UPF40 in comparison with a small dimensions common thermoelectric
device; (c) eTEG UPF40 structure, announced 4 W/cm2 with gradient temperature of 120 ºC [37].

Since then numerous projects of miniaturization of thermoelectric devices have been

developed by F. Völklein and V. Leonov from IMEC (Interuniversity Microelectronics
Centre) (Fig. 8). This group explored the fabrication process of thermoelectric sensors
with MEMS techniques (Micro-Electro-Mechanical Systems) [38,39] which are used in
human harvesting applications. [40,41].

Fig. 9 - (a) L. da Silva micro thermoelectric device with 60 junctions and thermoelectric structures
of 7 µm in dimension [44]; (b) L. M. Goncalves planar thermoelectric device in flexible substrate
with 8 junctions [45].

L. M. Goncalves presents in his PhD thesis the first planar micro thermoelectric cooling
device (Fig. 9(b)). The device has fabricated in flexible polyimide substrate with 25 µm
of thickness. The thermoelectric materials were deposited by thermal co-evaporation
and the Bi2Te3 as n-type and the Sb2Te3 as p-type. He obtained *" = 0.85 for Bi2Te3
and e *" = 0.3 for S2Te3 with Seebeck coefficient between 150 to 250 µVK-1 [45].

1.2.3 Nano-devices
The compound bismuth telluride (Bi2Te3) and antimony telluride (Sb2Te3) presents a
figure of merit higher than the compound polysilicon-germanium (polySiGe) but lower
operation temperature. During nearly four decades the figure of merit never passes the
unit value. Then the researchers focused on increasing the figure of merit using thin-
film depositions and the nano-scale phenomena. In 1993 L. D. Hicks and
M. S. Dresselhaus published the theoretical models about electron confinement through
quantum wells and increasing the figure of merit. This way the Seebeck coefficient is
increased and reduced the thermal conductivity of the thermoelectric material [46].
R. Venkatasubramanian from RTI laboratories presents in 2001 an semiconductor
thermoelectric material p-type (Bi2Te3/Sb2Te3) in superlattices with *" = 2.4. This
value inspired the researches and has completely revolutionary up to date.
R. Venkatasubramanian marked them as one of the most important personalities in the
thermoelectric field. He published well the approaches to increase the figure of merit:
by quantum wells, superlattices and thermionic effects in heterostructures [47]. Recently
he announced a thermoelectric generator with efficiency of 19 % (Fig. 10 (b)).

Fig. 11 - Power consumption scale by energy sources for thermoelectric generators [7].

Also a motivation of building a low cost chamber for deposit thermoelectric thin-films
is worth. This deposition system is custom fabricated for the thermoelectric thin-films
by the thermal co-evaporation technique. The necessity of build such a chamber lies on
the contamination of the thermoelectric materials in the actual deposition PVD
chambers. A small size chamber for depositing thermoelectric thin-films is represented
in Fig. 12. The chamber needs to have performance like the complete deposition
systems but with low costs. Also the thermal co-evaporation technique needs a custom
Fig. 10 - R. Venkatasubramanian thermoelectric projects; (a) Project presented in 2001; configuration to obtain the right stoichiometric of the thermoelectric thin-films. The
(b) Thermoelectric generator presented in 2009 with efficiency of 19 % (RTI courtesy) [48]. chamber was projected with a cylindrical special glass shape allowing the visualization
of the process by the operator. To obtain a uniformed thin-film the chamber was
1.3 Motivation projected by a rotated and heated substrate holder.

The scientific research in thermoelectric took another interest since 1993 [46] with the
introduction of nanotechnology and new horizons enabling previously unattainable by
thermoelectric devices. The nano-scale phenomena began to be exploited through
techniques of thin-film depositions to increase the efficiency of thermoelectric films. As
presented in Fig. 11 the thermoelectric generators have been applied in several areas.
The waste heat is the source with the largest range in the power range (W) that the
thermoelectric devices can acquire. Space applications have great potential for
application because they produce large amounts of electrical energy but presents serious
risks by using radioactive materials [49]. The micro thermoelectric generators have a
great potential of application in biomedical devices like the pacemaker [50] or powering
wireless sensor networks (WSN) [51-55].

Fig. 12 – Design of a deposition chamber to deposit thermoelectric thin-films.

2 Thermoelectricity
When the junction of two materials is covered by an electric current thermal energy is
released or absorbed at this junction. The thermal energy released or absorbed is
2.1 Thermoelectricity theory
proportional to the electrical current introduced at the junction. This effect is quantified
by the Peltier coefficient (') through the relationship with the Seebeck coefficient ()
The thermoelectricity consists in the conversion of temperature gradients to electric
and the temperature (") at the junction [7].
energy and it is quantified by the Seebeck coefficient () in 1822. When a material
(metal or semiconductor) with a negative Seebeck coefficient is heated at one end and
' = " , [V] (2)
cooled in the other, the electrons present in the material tend to move from the hot to the
cold section (Fig. 13). The hot section is in excess of ions (positively charged) while the
In terms of electrical conduction the materials can be conductors, insulators,
cold is to lower energy levels. The reverse is obtained for materials with a positive
semiconductors and superconductors. All materials have two energy bands: the valence
Seebeck coefficient () (Fig. 14) [7,45].
band and conduction band (Fig. 15). There is an energy associated with each of these
two bands, the valence energy (67 ) to the valence band and the conduction energy (69 )
to the conduction band. The energy gap (6:;< ) is the energy between the valence and
conduction bands. A metal is semiconductor if:

6:;< = 69 − 67 , [eV] (6)

An insulator has a very high energy gap with makes very difficult the transfer of
electrons from the valence band to the conduction band (Fig. 15(a)) [56]. As the energy
gap is smaller in a semiconductor then the electrons can move from the valence band to
the conduction band as a result of temperature is different from 0 K (Fig. 15(b)). In
semiconductors the electrons that pass from the valence band to the conduction leave a
gap in the valence band. In the conductors is not clear a separation between the valence
and the conduction band (Fig. 15(c)) [56].
Fig. 13 – Seebeck effect by demonstration the electron movement from the hot section to the cold
section.

It is reached a load balancing of the material when the electrical field obtained by the
load prevents electrons movement. The Seebeck effect is the voltage obtained by the
temperature difference.

Fig. 15 – Structure of the energy bands; (a) Insulator; (b) Semiconductor; (c) Conductor.

In a semiconductor the temperature rising is enough for electrons to gain energy to leave
the valence band to the conduction. It is known the pair electron-gap formed by a free
electron in the conduction band and a hole in the valence band [56]. In the case of
non-doped semiconductors (intrinsic) to move the electron from the valence band to the
conduction band is necessary provide energy (6?@ ) greater than the energy between the
two bands (6?@ > 6:;< ). In an intrinsic semiconductor the number of electrons (B) that
pass into the conduction band is equal to the number of gaps formed in the valence band
( C ), in other words B = C = B? . Where B? is the intrinsic concentration in a
semiconductor. The intrinsic concentration depends on the energy gap (6:;< ). The
electron and gap concentrations are given by:
Fig. 14 – Junction of two materials with negative Seebeck coefficient (on top) and positive (botton).
 I
EF "  $ ./01 The Fermi level for the electrons in the extrinsic semiconductor is given by [58]:
B = C = 2D H ( ∗
! # )%
∗
'*+ ,− 6 (7)
2'ℏ+ 234 5
QR
.BO = .C + 34 5 IJ
Where 34 is the Boltzman constant, ℏ = ℎ:29 is the Planck constant, !∗ and #∗ are
, [eV] (11)
ST

the effective masses of electrons and gaps in grams (g) [56]. The effective mass of the
electrons is given through [57,58]: Where .C is the Fermi level for intrinsic semiconductors, UV is the concentration of
donor impurities. Since the Fermi level for the gaps in the extrinsic semiconductor sis
ℏ given by [58]:
∗
! = ;< >?
, [g] (8)
;@ <
QW
.B0 = .C − 34 5 IJ , [eV] (12)
ST
And the effective mass of gaps is given by [57,58]:

∗ ℏ Where UX is the concentration of acceptors impurities. In Fig. 17 is represented the

# = ;< >A
, [g] (9) junction between a p-type semiconductor with a metal and a n-type semiconductor. As
;@ < can be seen in the n-type semiconductors the Fermi level lies near the conduction band
and in the p-type is located near the valence band (.BO > .B0 ). The Fermi level in the
For the intrinsic semiconductors the Fermi level is given by [57]: metal (.BZ ) is located between the p and n-type semiconductors [45].

DEFG $
!)
∗
.BC = + 34 5 IJ( # / ∗
, [eV] (10)
 %

$ DEFG
At the ambient temperature 34 5 ln( #∗ / !∗ ) ≪ , the Fermi level in an
% 
intermediate point of the two bands (.BC ≈ ./01 /2) [56]. The doped semiconductors
(extrinsic) have filled the valence band of five or three electrons. The doping is the
addition of impurities within the semiconductor. A doped semiconductor form group V
elements of the periodic table is called n-type semiconductor. If the semiconductor is
doped with elements from group III is called p-type semiconductor. As presented at Fig.
16 in extrinsic semiconductors the Fermi level decreases and the energy bands required
for electrons to move to the conduction band to the valence are smaller. In n-type
semiconductor the electron energy necessary to move to the conduction band is .N − .O
and in the p-type semiconductor the energy is .0 − .P . Fig. 17 – Junction between p-type semiconductor with metal and n-type semiconductor.

In the region of contact of a metal with a semiconductor there is an excess of electrons

and is called the region of accumulation. In Fig. 18 is represented the junction between
a metal and an n-type semiconductor when traversed by an electric current (i) [59].

Fig. 16 – Energy bands; (a) n-type semiconductor; (b) p-type semiconductor.

Fig. 18 – Junction between a metal and n-type semiconductor; (a) Absorbed heat in the junction;
(b) Released heat in the junction. Fig. 20 – Conventional Peltier device.

The electrons in the displacement of a region of lower energy to a region of higher When the electric current travels the junction between the metal and the n-type
energy absorb heat from the environment ([) and in the area of contact region cools the semiconductor (metal-n) is released thermic energy by the Peltier effect and heats the
environment by the Peltier effect (Fig. 18(a)). Reversing the direction of the electric bottom part of the device. On the top part the junctions between n-metal and metal-p is
current there is in the opposite direction of electrons and release heat to the environment absorbed heat. In the base of the device, when the electric current flows the junction
([) heating the junction (Fig. 18(b)) [59]. p-metal releases heat [45]. The junctions in the top part of the device absorbed heat by
the Peltier effect and the junctions at the bottom part release heat [45]. There is an
electrical and thermic resistance between the metal and the thermoelectric structures
called the contact resistance. This contact region is created essentially by diffusion of a
material junction with another material. The contact resistance (]^ ) is given by [45]:

_`
]^ = , [Ω] (13)
a

]^ is proportional to the contact resistivity (b^ ) and is inversely proportional to the

contact area or section (c) [45]. The internal resistance (]C ) of a thermoelectric device is
the sum of several resistances given by [7,45]:

]C = ]S + ]1 + 4]N + 2]e , [Ω] (14)

Fig. 19 – Junction between a metal and p-type semiconductor; (a) Absorbed heat in the junction;
(b) Released heat in the junction.
Where ]S and ]1 represents the internal resistance of the n and p-type thermoelectric
In Fig. 19 is represented the junction between a p-type semiconductor with a metal structures. ]N represents the contact resistance in the junction between the
when traversed by an electric current (i). When the electric current travel the junction thermoelectric materials and the metal. ]e represents the internal resistance of the metal
from the metal to the p-type semiconductor, the charges moves in the same direction in the junction [45]. In a thermoelectric device as a generator bases in the Seebeck effect
and the heat is absorbed from the environment ([). Heat is released in the junction to (Fig. 21). Each junction of p and n-type material (p-n junction) generates a voltage
the environment when the electric current flows from the p-type semiconductor to the proportional to the difference in temperature between the hot and cold side.
metal [59]. The same principle occurs in the thermoelectric device by the Peltier cooling
which is made by thermoelectric structures connected in series by metal junctions (Fig.
20). The electric current enters from one the terminal and flows through all the
structures and their metal contacts then exit at the other end. The n-type thermoelectric
structures have a negative Seebeck coefficient (\) and the p-type have a positive.
 \
rˆ = , [WK } m}Ž ] (18)
b

2.2 Thermoelectric materials

The figure of merit of the thermoelectric materials has been optimized over the years
(Fig. 22) by advances in research and fabrication or by the introduction of the
nanotechnology [46]. The thermoelectric materials have been deposited by several
deposition techniques like: flash evaporation, hot wall epitaxy, pulsed laser deposition,
sputtering, metal organic chemical vapor deposition, among others. However these
techniques require a long period of preparation of the materials or are relatively
expensive or complicated [60].

Fig. 21 – Thermoelectric device as a thermoelectric generator.

The voltage produced in the p-n junction is smaller (in order of µV/ºC). This voltage
can be multiplied by each junction in series (n) and given by [7,45]:

fghi = Jj\1 − \S k. ∆5 , [V] (15)

Where \p and \q are the Seebeck coefficients (\) for the p and n structures. The ∆5 is
the gradient temperature between the hot and cold side. The power that a thermoelectric
generator can produce depends of his internal resistance (]C ). This power is maximum
(rsXt ) when the internal resistance (]C ) is equal to the load resistance (]u ) and given
by [7,45]:

< Fig. 22 – The evolution of the figure of merit (‘) over the years.
vwxy < {Sj|G }|~ k.∆i€
rsXt = = , [W] (16)
%zT %jz~ zG %z? z‚ k
The superlattices structures promote the electrons mobility and gaps allowing flow of
electric current. But hampers the movement of the phonons responsible of the heat
The performance of the thermoelectric structures is measured by the value of the figure transfer. In the PhD thesis of D. D. L. Wijngaards is presented the study of the figure of
of merit (ƒ). The figure of merit is not constant with the temperature and is also merit for the majority thermoelectric materials by the operation temperature (Fig. 23)
designated by (ƒ5). The performance of a thermoelectric device is quantified by the [43]. The red zone in Fig. 23 represents the most common stoichiometric variations for
coefficient of performance COP. This coefficient measures the quantity of heat the compounds of tellurium (Te), bismuth (Bi), antimony (Sb) and selenium (Se). These
absorbed by the ambient ([N ) in function the quantity of electric energy consumption by compounds present a higher figure of merit and for that reason is chosen for the
the device ([! ) [7,45]. majority of the thermoelectric devices.

1
[N j\1 − \S k5N † − ]C †  − 3(5# − 5N )
„ r= = 2 (17)
[! ]C †  + j\1 − \S k(5# − 5N )†

Where 5# and 5N are the temperatures in the hot and cold sides. † is the electric current
and ]C is the internal resistance of the thermoelectric generator. 3 is the thermal
conductivity of the thermoelectric generator. The performance of the thermoelectric
thin-films is measured by the power factor given by:
 Fig. 24 – Model of a thermoelectric generator [43].

2.4 Thermoelectric energy scavenging microsystem

The construction of a custom deposition system allows the development of a
thermoelectric microsystem in a flexible substrate. The design of the microsystem is
Fig. 23 – Figure of merit vs. operation temperature of the majority thermoelectric materials [43].
represented in Fig. 25. The microsystem is constituted by several thermoelectric
generators and a solid-state battery in flexible substrate and connected to a step-up
DC-DC voltage converter. This microsystem is design to be applied in a human energy
2.3 Thermoelectric converters modeling harvesting applications to power wireless sensor networks WSN and biomedical
devices.
In Fig. 24 is presented the thermoelectric device as a generator. The resistance ’ “
represents the internal resistance of the device when is traveled through an electric
current ”• . On the left side is represented by the Seebeck effect of the voltage generated
by the p-n junctions of the device (–— − –˜ )(‘™ − ‘š ). The voltage generated depends
on the temperature at the junction of the hot side ‘™ and the temperature of the cold side
‘š [43,45]. The Peltier effect is present due to the electric current that passes between
the junctions. On the right side of the block the Peltier effect of the cold junction is
represented by the current source (–— − –˜ )”• ‘š and for the hot junction (–— - )  .
The thermal capacity equivalent to the cold junction is represented by !" and for the
hot junction by ! [43,45]. The losses by convection and radiation to the ambient
temperature ambient # are represented for the cold junction by $!" and for the hot
junction by $! . The resistances $!,% represents the internal resistances of the
thermoelectric structures [43,45]. The Joule effect for the thermoelectric structures is
represented by &' . The heat fluxes due to the Joule effect for the metal contacts are
represented by &'" . Finally are added to the system thermal loads to ensure the gradient
temperature (&(" e &( ) [43,45].

Fig. 25 – Design of an energy harvesting micro system in flexible substrate.

2.5 Voltage converters

The levels of voltage generated by most of the micro thermoelectric generators are not
sufficient to power electronic devices. For the widespread use of power micro systems
need to be able to power common devices. The solution for this problem lies in the
step-up DC-DC voltage converters. A model for a voltage converter to introduce in a
power micro system is presented in Fig. 26.

Fig. 27 – Four point probes resistivity measurement system; (a) Top view; (b) Side view.
Fig. 26 – Schematic circuit of a step-up DC-DC voltage converter [61].
The method consists in the application of a known electric current in two probes and
The voltage conversion through the step-up DC-DC converter is done by the help of the measured the voltage in the other two probes. The method is repeated in the four
capacitor *+- and the coil .+- . The electric current at the exit of the thermoelectric configurations represented in Fig. 28 and Fig. 29. By the van der Pauw method is
generator charges the capacitor *+- while the switch /0 is systematically closed and obtained the sheet resistance 7; for a thin-film independent of the shape. It is obtained
opened at high frequency. In the way to obtain a necessary voltage at the exit the signal by the horizontal resistance (7= ) and vertical resistance (7> ). The horizontal resistance
of the switch need in very low duty-cycle to avoid the discharge of the capacitor *+- . 7= value is given by:
When the switch /0 is open the energy storage in the coil .+- forces the discharge of
the capacitor *+- by the diode 1. Finally the electric current charges the high charge @ CDE CGH
7= = B + J, [Ω] (19)
capacitor *23456 and the current is rectified to the system or designed 7849: . A FGH FDE

3 Deposition and characterization technology

3.1 Measurement system of the thermoelectric properties
The electrical resistivity of the thermoelectric thin-films is measured by the van der
Pauw four point probe method, and is independent of the shape of the thin-film [62].
With this method is even possible to obtain the carrier density, the mobility and also the
concentration by the Hall effect through a magnetic field [63]. The system for measure
the resistivity by the van der Pauw method is presented in Fig. 27.

Fig. 28 – Horizontal configuration; (a) First configuration; (b) Second configuration.

The vertical resistance 7> value is obtained by:

@ CDG CEH
7> = B + J, [Ω] (20)
A FEH FDG
 To obtain the Seebeck coefficient (M) is necessary a gradient temperature. The sample is
placed between two plates with different temperatures. The hot plate has inside a cooper
filament which serves as heater when a constant and known voltage is applied. Each
plate has inside a resistance temperature detector (RTD) reference PT100 to obtain the
plates temperature. The temperature sensor has a range from 0 ºC to 100 ºC. The
difference in temperature is obtained by:

7 = 7Z (1 + \@ ^ − \A ^ A ) , [Ω] (23)

Where:

7Z = 100 Ω;
Fig. 29 – Vertical configuration; (a) Third configuration; (b) Fourth configuration. \@ = 3,90802 X℃O@ ;
\A = 0,5802 i℃O@ .
According van der Pauw the sheet resistance is related with the horizontal resistance 7=
and vertical resistance 7> as follows:
After obtaining the temperature in the two plates, voltage is measured in the same
plates. Then the Seebeck coefficient (α) is calculated.
Q Q
BOP R J BOP T J (21)
N QS +N QS =1
3.2 Thin-films deposition technology
By knowing the thin-film thickness (ℎ) is possible obtained the electrical resistivity by: All materials are composed of atoms in constant motion, which attract each other at
close range. In solid state the atoms are very close and form spatially ordered structures
V = 7; × ℎ , [ΩX] (22) called lattices [64,65]. The atoms have strong force which leads to an equilibrium state.
The temperature increase induces the removal of atoms together and causing the solid to
be liquid if there are still cohesive forces between atoms. Further increasing temperature
The error is in the order of 2r/D. the atoms acquire enough kinetic energy to leave the surface of the liquid and passing
The measurement of the Seebeck coefficient (M) is done by the system presented in Fig. into a gaseous state. With the increasing of temperature the number of atoms in the
30. gaseous state also increases. The atoms in a gaseous state are the vapor of the substance.
All the materials have a vapor pressure that is greater the higher the temperature. In the
gaseous state the forces between atoms are much weaker and are more important
interactions through collisions. The collisions depend on the number of atoms per unit
volume or numeric density and particles energy. The energy is related to the
temperature of the gas where the pressure is a very important factor [64]. To obtain a
thin-film deposition is necessary to control the pressure inside a deposition chamber.
The deposition of a material occurs by heating the material to be deposited. Where it
changes its phase from solid to liquid and then to a gaseous state. Sometimes occurs a
sublimation of a material which consists in the transition from solid state to gaseous
state without passing through the liquid state [65]. In the deposition the material
particles moves in all directions and are deposited on the substrate. Normally the
substrate is parallel to the charge (deposition material) which it is the shortest way that
the particles have to go. The position of the substrate induces a uniform thickness of the
material deposited on its surface. And to obtain a desire deposition of a material is
primary necessary to consider the tension pressure and is temperature on a certain
pressure. In Table 1 are presented the proprieties of most common deposited materials.
Fig. 30 – System for measure the Seebeck coefficient (top view).
 Table 1 – Common materials proprieties for deposition [66,67].
Temperature (ºC) Deposition
Fusion
Density @ Pressure (mbar) technique
Material temp.
(g/cm3) Thermal
(ºC) 10-8 10-6 10 -4
e-beam
evap.
Aluminum (Al) 2.7 660 677 821 1010 Excellent Bar
279 345 425
Antimony (Sb) 6.62 631 Weak Boat
Sublimes
Bismuth (Bi) 9.8 271 330 410 520 Excellent Boat
Boat,
Cooper (Cu) 8.94 1083 727 857 1017 Excellent
wire
837 977 1157 Wire,
Cromium (Cr) 7.2 1890 Good
Sublimes Boat
Silicon dioxide 1610 -
2.2–2.7 - - 1025 Excellent -
(SiO2) 1710
Lithium (Li) 0.53 180 227 307 407 Good Boat
Wire,
Nickel (Ni) 8.91 1453 927 1072 1262 Weak
Boat
Boat,
Gold (Au) 19.32 1962 807 947 1132 Excellent Fig. 31 – Operating pressure of different types of vacuum pumps.
Filament
Platinum (Pt) 21.45 1769 1292 1492 1747 Excellent -
Generally, the deposition of materials can be made by physical vapor deposition PVD
Wire,
Silver (Ag) 10.49 961 847 958 1105 Excellent or by chemical vapor deposition CVD. The most common physical deposition
Boat
techniques are the thermal evaporation with the variations of flash evaporation and co-
Silicon (Si) 2.33 1410 992 1147 1337 Weak Boat evaporation, e-beam (electron beam) and sputtering, among others.
Tellurium (Te) 6.25 452 157 207 277 Weak Boat
Titanium (Ti) 4.5 1668 1067 1235 1453 Excellent Boat

To obtain a pure deposition chamber is necessary to extract the gas inside with vacuum
pumps. The first vacuum pump was invented by Otto von Guericke in 1650 and
published in 1672 [68]. All actual vacuum pumps are based on positive displacement
which involves the mechanical movement of a piston, vanes, diaphragm or a suction
system to extract the gas inside the chamber. The vacuum pumps have three stages: a
capture of a certain gas volume, the compression, and the expulsion the same gas [69].

Fig. 32 – Typical PVD deposition system.

In Fig. 32 is represented the typical PVD deposition system with a deposition chamber
and vacuum pump connected in series to control the pressure inside. The charge is
inside a crucible and when the deposition starts also is deposited in the mass sensor. The
mass sensor typically is located near the substrate. Is constituted by a piezoelectric
crystal which during the deposition is vibrating and is possible to obtain the deposition
rate and thin-film thickness [64,65,70]. The thermal evaporation technique is
represented in Fig. 33. This technique consists in heating the charge by means a
resistance called crucible. The charge is located between the resistance and the heating power source used in this technique uses voltage in the range of 4 to 20 kV to heat the
is due the flow of electric current. This method is simple and cheap, but presents the filament. The heated filament promotes the emission of accelerated electrons by
inconvenience of projecting impurities or other contaminants present in the crucible. thermionic effect. Typically the filament is coated by an oxide to promote the electron
releasing in vacuum. A magnetic field is generated on top of the filament to steer the
electron beam to the crucible [69]. Two problems are associated to this technique: the
shifting density of the beam and the dispersion of the beam. These problems need to be
resolved by the operator. The beam density can be controlled by changing the electric
current of the power source. In the case of the beam is not steered to the middle of the
crucible the operator change the crucible position [69].

Fig. 33 – Thermal evaporation technique.

In Fig. 34 is represented he flash evaporation technique and as the name indicates is a

technique where the evaporation of the material to be deposited is processed very
quickly and at high temperatures. The charge is placed in a reservoir on top of the
crucible. The crucible is heated by a power source and the charge falls by a gutter into
the crucible. By a mechanical vibration in the gutter is controlled the material falling.
This technique as the advantage of being able to deposit in small holes in great depth
because de atoms have more energy than the thermal evaporation technique. All the
charge evaporates which leads to a composition of the thin-film equal to the charge Fig. 35 – Electron beam (e-beam) technique.
[65].
In Fig. 36 is represented the sputtering technique which allows the deposition of
materials without significant change the charge (target). Contrary to the evaporation
techniques the sputtering presents conformal thin-films in regions of degrees. This
factor is due to the large area of the target which is not seen in the e-bam or thermal
evaporation. The sputtering technique could be by direct current DC or by radio
frequency RF. The DC sputtering is typically for conductive materials i.e., metals. In
the case of RF sputtering is focused for nonconductive materials i.e., oxides. In the
sputtering technique is necessary the entry on gases to promote the plasma. In the
plasma creation is used inert gas like argon (Ar). In the reactive sputtering is used a gas
which enters in the composition of the thin-film. As represented in Fig. 37 the
magnetron is constituted by magnets placed on top to create a magnetic field centered to
the middle of the target.

Fig. 34 – Flash evaporation technique.

The depositions by thermal evaporation techniques have the disadvantages of projecting

impurities and others contaminants present in the filaments or boats. As well is
difficultly to deposit materials with high melting temperatures at a controlled rate
deposition [69]. In Fig. 35 is represented the electron beam technique (e-beam). With
the e-beam technique is to deposit a large range of materials and control the deposition
rate. The charge is placed in a crucible inside the evaporator. The evaporator is made of
cooper (Cu) with a very good thermal conductivity and is refrigerated. Through this
technique is obtained a much pure thin-film due the evaporator refrigeration which
avoids the contamination by impurities from the crucible or evaporator [65,69]. The
Fig. 36 – Sputtering technique.

The plasma is generated when the gas molecules ionize and charges are added or
removed – like electrons – this process is called ionization. The accelerated ions beam
collides with the negatively charged target. When the ions collide release atoms from
the target which are deposited in the substrate [64]. In this process is produced a lot of
Fig. 38 – Thin-film PVD deposition system.
heating in the target which is necessary a refrigeration system to maintain the target
integrity. The plasma color depends on the used gas. On RF sputtering for oxides is
In Fig. 39 are presented the techniques inside the PVD chamber shown in Fig. 38.
normally used the frequency of 13,56 MHz which produces relatively lower deposition
rates. The disadvantage of RF sputtering faces the DC sputtering is on the use of
dielectric targets with low thermal conductivity which leads to break easily due the
mechanical stress generated by cooling/heating [65].

Fig. 39 – Techniques inside the thin-film PVD chamber (top view).

3.3 Thermoelectric thin-films by co-evaporation

To deposit compounds, i.e., thermoelectric materials, it’s necessary deposit different
materials at the same time in order to obtain the desired stoichiometry. In Fig. 40 is
represented the thermal co-evaporation technique to archive compounds thin-films.
Fig. 37 – Sputtering technique (in detail). Each charge have different deposition rate and is necessary a power source, a mass
sensor crystal and controller for each material. By this way is possible to control the
In Fig. 38 is presented a complete thin-film PVD deposition system. desired stoichiometry for the compound. In this technique is necessary fabricate a
separation shield between the charges and has two main reasons: control the meeting
point of the two charges in the substrate, and allowing each mass sensor only read
values of one material. In Fig. 42 is presented the co-evaporation technique system inside the PVD chamber
shown in Fig. 41.

Fig. 40 – Thermal co-evaporation technique (cut view).

In Fig. 41 is presented a complete thin-film PVD thermal evaporation system for

thermoelectric materials.

Fig. 42 – Thermal co-evaporation technique system (top view).

3.4 Device Patterning techniques

The patterning techniques to fabricate thermoelectric structures focuses in
photolithography and wet-etching techniques, using HNO3/HCl based etchants. The
tentative deposition of thin-films by direct evaporation of the bulk materials - Bi2Te3
and Sb2Te3 - proved to be impossible due to the large differences in vapour pressure of
bismuth (Bi), Antimony (Sb) and Tellurium (Te), resulting in a compositional gradient
along the film thickness [71]. Other techniques have been reported in the literature for
the deposition of telluride thin films. Co-sputtering, electrochemical deposition, metal-
organic chemical vapour deposition or flash evaporation, are some examples.
Concerning the use of patterning techniques on thermoelectric films, few approaches
have been reported in the literature. Silva et al. [72] uses lift-off, with SU-8 photoresist,
to create vertical columns of thermoelectric materials. But due to the maximum working
temperature of SU-8 photoresist (≈180 ºC), thermoelectric properties of Bi2Te3 and
Sb2Te3 films incorporated in the devices are worse than those obtained in bulk
materials, since 200-300 ºC of substrate temperature is required to fabricate high quality
thermoelectric films. Böttner et al. [73] uses dry etching to pattern thermoelectric
devices. Shafai [74] reports the patterning of Bi2Te3 films by wet etching but no details
are given of the process [75].

Fig. 41 – PVD thermal co-evaporation system.

 4 Thermoelectric thin-films by co-evaporation
The proprieties of thermoelectric materials vary depending on the deposition technique
chosen or the stoichiometry of the compound. In Table 3 are presented the variations of
the thermoelectric materials properties by various authors and centered in compounds
based in bismuth (Bi), antimony (Sb) and tellurium (Te). To obtain a thermoelectric
compound is necessary knowing the materials proprieties. In Table 4 are presented the
materials properties of the most common deposited materials. Generally the depositions
of thermoelectric materials are in a heated substrate holder. The Bi2Te3 compound was
deposited in a heated substrate at 250 ºC with a pressure of 6.1 E-7 mbar. The
deposition rate for bismuth is obtained at 0.5 Å/s with 110 A of electric current and the
rate for telluride is archived at 1.5 Å/s with 50 A. The Sb2Te3 compound was deposited
with the substrate at 150 ºC and 1.2 E-6 mbar of pressure. The deposition rate of
antimony is obtained at 1.1 Å/s with 70 A and rate for telluride is archived at 2.2 Å/s
with 55 A.

Fig. 43 – Fabrication steps of a thermoelectric cooler by patterning techniques [75]. Table 3 – Thermoelectric materials.
Seebeck Figure of Thermal
In Fig. 43 shows the fabrication process of thermoelectric cooler. The p-type Sb2Te3 Resistivity Deposition
Material coeff. merit condutivity Ref.
thin-film is deposited by thermal co-evaporation followed by a thin layer (100nm) of (Ωm) technique
nickel (a).The thin layers of nickel are used to avoid diffusion of thermoelectric material (µVK-1) (ZT) -1 -1
(Wm K )
into the next deposited layers. Photoresist (PR) is spun and p-type elements are Bi2Te3 13 -228 - -
patterned by photolithography (b),(c). Nickel is etched in a chromium etchant (Transene Co-ev. [76]
Sb2Te3 10.4 171 - -
1020), thermoelectric film is patterned by wet-etching in HNO3:HCl bath (d) and PR is Bi2Te3 28.3 -228 0.28 2
removed. N-type film is then deposited by co-evaporation, followed by a 100 nm nickel Co-ev. [77]
layer (e). PR is applied and patterned by photolithography for n-type element definition Sb2Te3 25.5 126 0.09 2.1
(f),(g). Nickel is etched in a chromium etchant (Transene 1020), n-type film is etched in Bi2Te3 9 -210 - -
MOCVD [78]
HNO3 (h) and PR removed (i). Contacts are deposited, starting with a 100 nm layer of Sb2Te3 3.5 110 - -
nickel, followed by 1 μm of aluminium (j). PR is spun and contacts patterned by (Bi2Te3)0.9
photolithography (k). Nickel is etched in a chromium etchant (Transene 1020), and 15 -200 0.5 1.6
(Bi2Se3)0.1
aluminium with a standard aluminium etchant (Transene type A). PR is removed (l). A Flash [79]
(Bi2Te3)0.2
protective layer of Si3N4 can also be deposited by low-temperature hot-wire chemical 12 240 0.96 1.5
vapour deposition (HW-CVD) and patterned if required, depending on the application. (Sb2Te3)0.7
In Table 2 is represented the most relevant etch rates. Bi2Te3 21 -155 - -
Co-sputt. [35]
(Bi,Sb)2Te3 23 227 - -
Table 2 – Summary of etch rates for patterning techniques [75].
Bi2Te3 10 -60 - -
Material Bi2- Electpl. [80]
- - - -
x,Sbx,Te3
Etchant Bi2Te3 Sb2Te3 Aluminum Nickel

Al – Transene type A 8 Å /sec 5 Å /sec 10-80Å/sec < 0.1 Å/sec Bi2Te3 12.6 -248 0.98 1.5*
Co-ev. [45]
Cr - Transene 1020 ≈ 20 Å /sec <1 Å/sec - 10-40Å/sec Sb2Te3 12.6 188 0.56 1.5*
3HNO3:1HCl (dil 70% H20) 2000 Å /sec 800 Å /sec < 2 Å/sec < 0.2 Å/sec * Estimated values.
HNO3 (dil 70% H20) 2500 Å /sec 50 Å /sec < 0.1 Å/sec < 0.1 Å/sec
 Table 4 – Seebeck coeficiente and electrical proprieties of the most common materials [7,81,82].
Figure
Seebeck
of merit
Electrical Thermal coeff.
Resistivity (ZT)
Material conductivity conducti. (µVK-1)
(x 10-8Ωm) (x 10-3)
(x 10-7Ωm-1) (Wm-1K-1)
300 K
Aluminum (Al) 2.74 3.65 237 -1.7 0.13
Cooper (Cu) 1.72 5.88 401 1.83 0.15
Lead (Pb) 21 0.48 35.3 -1.047 0.04
Chromium (Cr) 12.9 0.78 94 17.3 7.4
Molybdenum
5.3 1.89 138 5.57 1.27
(Mo)
Nickel (Ni) 7 1.43 90.9 -19.5 17.93
Gold (Au) 2.2 4,55 318 1.94 0.16 Fig. 45 – Power factor of Bi2Te3 thin-films as function of Te composition, obtained by EDX [83].
Palladium (Pd) 10.5 0.95 71.8 -9.99 3.97
Platinum (Pt) 10.4 0.96 71.6 -5.28 1.12 4.2 P-type thermoelectric thin-films made of bismuth telluride
Silver (Ag) 1.59 6.21 429 1.51 0.1
The study of characterization power factor of p-type thermoelectric thin-films is
Rhodium (Rh) 4.8 2.08 58.2 0.4 0.02 presented in Fig. 46.
Tungsten (W) 5.3 1.89 173 1.07 0.04
Vanadium (V) 19.9 0.5 30.7 1 0.05

4.1 N-type thermoelectric thin-films made of bismuth telluride

As early reported the temperature of the substrate during the deposition is a very
important issue to obtain the desire stoichiometry for the compound. This temperature
also influences the power factor of the thermoelectric thin-film. In Fig. 44 is presented a
study of the n-type thermoelectric thin-film power factor as function of substrate
temperature during the deposition. In Fig. 45 is presented power factor of the n-type
thermoelectric thin-films as function of Te composition.

Fig. 46 – Power factor of Sb2Te3 thin-films as function of substrate temperature and as function of
Te composition, obtained by EDX [84].

4.3 Thermoelectric devices

The developed thermoelectric devices were obtained through shadow masks presented
in Fig. 47. The technique of shadow masks are simple and well used in thin-film
depositions were the precision distance is not critical. The distance of p-type and n-type
deposited thin-films are 500 µm. The contact area of the thermoelectric structures is
1 mm2. The aluminum metal contacts have 7 mm2 of area.
Fig. 44 - Power factor of Bi2Te3 thin-films as function of substrate temperature [83].
 functionally graded material. A scanning Seebeck microprobe is a device for measuring
the Seebeck coefficient (  ) on a sample surface spatially resolved to achieve
information on the homogeneity or distribution of the components. The Seebeck
coefficient () is a measure of the electrically active components in a material. Different
components in a single unit become visible by measuring the local () with a scanning
thermoprobe. This applies accordingly for the electrical conductivity ( ) and therefore
the behaviour of the material in a certain temperature gradient becomes predictable. A
scanning Seebeck microprobe is presented in Fig. 49 and combined with the
measurement of the electric potential along the surface of semiconducting or metallic
material [85].

Fig. 47 – Shadow masks to obtain the thermoelectric device.

The developed thermoelectric devices are presented in Fig. 48.

Fig. 49 – Schematic of the Seebeck microprobe. The temperature of the sample and the probe tip as
well as the Seebeck voltage can be measured. The probe tip is positioned by linear stages [85].

A heated probe tip is placed onto the surface of the sample under investigation,
measuring the Seebeck coefficient (). Using a specially designed sample holder, an
AC current can be applied to the sample, allowing for the detection of the voltage drop
between one current contact and the travelling probe tip. This voltage is proportional to
the electrical conductivity ( ) at the tip position. With this technique a spatially
resolved imaging of the Seebeck coefficient () as well as the electrical conductivity
( ) can be performed. Furthermore the electrical contact resistance between different
materials becomes visible, e.g., in segmented thermoelectric or other devices. So far it
was reported about a measurement equipment for the spatial resolution of the Seebeck
coefficient () [86,87]. With this equipment it is possible to detect inhomogeneities,
different phases, even different doping levels or anisotropies [66], that is hardly possible
by other surface analysis methods like SEM, EDX etc. For many materials also the
homogeneity of the electrical resistivity plays an important role in their performance,
especially for good quality of semiconductors.

Fig. 48 – Developed thermoelectric devices in polyimide foil of 50 µm of thickness.

4.4 Why Scanning Probe?

The local resolution of the Seebeck coefficient () is a measure for different electrically
active components in materials. This becomes important especially investigating
 Fig. 51 – Schematic of the potential probe. An electrical AC current is applied to the sample and a
probe tip scans the surface of the sample and measures the electric potential in each point resulting
in the electrical resistivity [85].

Similar to the Seebeck surface scan the electric potential can be measured. Therefore a
sample holder was constructed not only to support the sample mechanically, but also to
Fig. 50 – Thin-film surface image of a Seebeck coeficiente peer pixel by scanning Seebeck apply an electric current to the sample. A probe tip for the voltage pick-up was installed
microprobe [45].
(Fig. 51). The tip´s movement by linear stages allows for scanning the sample, and the
change of the electric potential can be measured along the sample. The specific
The behavior of the Potential-Seebeck-Microprobe PSM consists in a heated probe tip
resistivity can be calculated for each measurement point according to Ohm´s law with
that is positioned onto the surface of a sample. The probe is connected with a
the measured voltage *, the current < applied to the sample and with the resistance > ,
thermocouple (in this case type T, Cu-CuNi) measuring the temperature !" . The sample
the specific resistivity ?, the lenght @ and the cross section A of the sample [88].
is in good electrical and thermal contact to a heat sink and also connected with a
thermocouple measuring !# . The heat flow from the probe tip to the sample causes a
local temperature gradient in the vicinity of the tip. Combining the Cu-Cu and the @
?=> , [ΩcmC ] (27)
CuNi-CuNi wires of the thermocouples a voltage $# and $" is measured yielding the A
Seebeck coefficient % of the sample at the position of the probe tip according to
equations [85]: With help of this tool not only the electric resistivity can be measured, but also the
ohmic contact resistance between different materials, i.e., in a stacked thermoelectric
device or contacts [85]. In Fig. 52 is presented the instrument of scanning probe
*+ = (-. − -01 ). (45 − 4+ ), [V] (24) microprobe called SPM-Potential-Seebeck Microprobe developed by D. Platzek to
spatial resolution of the Seebeck coefficient and the electrical conductivity in
thermoelectric thin-films. This product is sold by Panco GmbH [89].
And:

*5 = (-. − -0178 ). (45 − 4+ ), [V] (25)

Yielding in:

*+
-. = . (-01 − -0178 ) + -01 , [µVK ;5 ] (26)
*5 − *+

Where &' and &'() are the Seebeck coefficients of Cu and CuNi, respectively.
Mounting the probe to a three dimensional micro-positioning system (Fig. 49) allows
the determination of the thermopower of each single sample position for a certain
temperature range [87], in the easiest case at room temperature. The result is a two
dimensional image of the Seebeck coefficient (Fig. 50) [85].
 [14] L. S. Stilbans, “The Investigation and New Applications of Semiconductor
Thermocouples”, PhD Thesis, A. F. Ioffe Physica1-Technical Institute, 1951.
[15] L. S. Stilbans, “The Investigation and Some Applications of Semiconductor
Thermoelements”, DSc Thesis, A. F. Ioffe Physical-Technical Institute, 1960.
[16] M. V. Vedernikov and E. K Jordanishvili, “A. F. Ioffe and Origin of Modern
Semiconductor Thermoelectric Energy Conversion”, 17th International
Conference on Thermoelectrics, pp. 37-42, 1998.
[17] A. F. Ioffe, et al., “On Increasing the Efficiency of Semiconducting
Thermocouples”, The Reports of Academy of Sciences USSR, Vol. 106, 1956.
[18] H. J. Goldsmid and R. W. Douglas, “The Use of Semiconductors in
Thermoelectric Refrigeration”, British Journal of Applied Physics, Vol. 5, Issue
l l, pp. 386-390, 1954.
[19] W. B. Pearson and W. B. Templetion, “Thermo-electricity at Low Temperatures.
III. The Absolute Scale of Thermo-electric Power: a critical discussion of the
Present Scale at Low Temperatures and Preliminary Mesurements Towards its
Fig. 52 – Scanning probe instrument for spatial resolution of the Seebeck coefficient and the Redetermination”, Proc. Royal Society A, Vol. 231, pp. 534-544, 1955.
electrical conductivity – SPM-Potential-Seebeck Microprobe (Courtesy of D. Platzek).
[20] J. W. Christian, et al., “Thermo-electricity at Low Temperatures. VI. A
Redetermination of the Absolute Scale of Thermo-electric Power of Lead”, Proc.
Royal Society A, Vol. 245, pp. 213-221, 1958.
References [21] R. B. Roberts, “Absolute scale of thermoelectricity”, Nature, Vol. 265,
pp. 226-227, 1977.
[1] TH. J. Seebeck, “Magnetische Polarisation der Metalle und Erze durch [22] C. B. Vining, “Thermopower to the people”, Nature, Vol. 423, pp. 391-392,
Temperatur-Differenz”, Abhandlungen der Deutschen Akademie der 2003.
Wissenschaften zu Berlin, pp. 265-373, 1822. [23] C. B. Vining, “Semiconductors are cool”, Nature, Vol. 413, pp. 577-578, 2001.
[2] M. Peltier, “Nouvelles Expériences sur la Caloricité des Courans Électrique”, [24] T. L. Charland, et. al., “Process for the preparation of thermoelectric elements”,
Annales de Phys. et de Chimie., pp. 180-193, 1834. US Patent No. 3086068, 1963.
[3] H. F. E. Lenz, “Über die Gesetze die Elektricitätsleitung in Drähten von [25] R. C. Linnon, “Method of manufacturing a thermoelectric device”, US Patent
verschieden Lange u Dicke”, Mém. Acad. St Pétersb. Sér. VI, Vol. III, 1838. No. 4054478, 1977.
[4] A. F. Ioffe, “Semiconductor Thermoelements and Thermoelectric Cooling”, [26] K. Yamada, et al.,“Thermoelectric module unit” US Patent No. 6233944 B1,
1st Edition, Infosearch, 1957. 2001.
[5] W. Thomson, “On an Absolute Thermometric Scale founded on Carnot's Theory [27] IAV Gmbh, “Thermoelectrics in Motion Electric Power from Waste Heat – A
of the Motive Power of Heat, and calculated from Regnault's Observations”, Chance for the Automotive Industry”, 1st IAV Conference on Thermoelectrics –
Philosophical Magazine, 1848. A Chance for the Automotive Industry?, 2008.
[6] W. Thomson, “On a Mechanical Theory of Thermo-Electric Currents”, [28] J. LaGrandeur, et al., “Vehicle Fuel Economy Improvement through
Proceedings of the Royal Society of Edinburgh, Vol. III, No. 40, pp. 91-98, Thermoelectric Waste Heat Recovery”, Proceedings of the 11th Diesel Engine
1851. Emissions Reduction (DEER) Conference, 2005.
[7] D. M. Rowe, “Handbook of Thermoelectrics”, 1st Edition, CRC Press, 1995. [29] G. J. Snyder, “Small Thermoelectric Generators”, The Electrochemical Society
[8] Lord Rayleigh, “On The Thermodynamic Efficiency of the Thermopile”, Interface, pp. 54-56, 2008.
Philosophical Magazine Series 5, 1941-5990, Vol. XX., Issue 125, pp. 361-363, [30] J. A. Paradiso, and T. Starner, “Energy Scavenging for Mobile and Wireless
1885. Electronics”, PERVASIVE Computing, 2005.
[9] E. Altenkirch, “Uber den Nutzeffeckt der Thermosaule”, Physikalische [31] M. Kishi, et al., “Fabrication of a miniature thermoelectric module with
Zeitschrifr, Vol 10, pp. 560-580, 1909. elements composed of sintered Bi-Te compounds”, 16th International
[10] E. Altenkirch, “Electrothermische Kalteerzeugung und Reversible Electrische Conference on Thermoelectrics, 1997.
Heizung”, Physikalische Zeitschrifr, Vol. 12, pp. 920-924, 1911. [32] D. M. Rowe, “Review Thermoelectric Waste Heat Recovery as A Renewable
[11] M. Telkes, “The Efficiency of Thermoelectric Generators”, Journal of Applied Energy Source”, International Journal of Innovations in Energy Systems and
Physics, Vol. 18, pp. 1116-1127, 1947. Power, Vol. 1, No. 1, 2006.
[12] A. F. Ioffe, “Energeticheskic Osnovy Termoelektricheskikh Baterei iz [33] D. M. Rowe, “Miniature Thermoelectric Convertors”, U. K. Patent
Poluprovoduikov”, Academy of Sciences of the USSR, 1949. No. 8714698, 1988.
[13] Yu. P. Maslakovets, “Heat Energy Conversion in Electrical One by
Semiconductor Thermocouples”, DSc Thesis, A. F. Ioffe Physical-Technical
Institute, 1949.
[34] H. Böttner, “Micropelt® Miniaturised Thermoelectric Devices: Small Size, High [54] J. P. Carmo, et al., “A 2.4-GHz CMOS Short-Range Wireless-Sensor-Network
Cooling Power Densities, Short Response Time”, 24th International Conference Interface for Automotive Applications”, IEEE Transactions on Industrial
on Thermoelectrics, 2005. Electronics, Vol. 57, No. 5, pp. 1764-1771, 2010.
[35] H. Böttner, et al., “New thermoelectric components using microsystem [55] J. Liu and J. Yao, “Wireless RF identification system based on SAW”, IEEE
technologies”, Journal of Microelectromechanical System, Vol. 13 Issue 3, pp. Transactions on Industrial Electronics, Vol. 55, No. 2, pp. 958-961, 2008.
414-420, 2004. [56] J. H. Correia and J. P. Carmo, “Introdução às Microtecnologias no Silício”,
[36] [Link] 1st Edition, Lidel, 2010.
[37] [Link] [57] R. Lobo, “Nanotecnologia e Nanofísica”, 1st Edition, Escolar Editora, 2010.
[38] V. Leonov, et al., “Thermoelectric Converters of Human Warmth for Self- [58] J. Singh, “Semiconductors Devices: an Introduction”, McGraw Hill, 1994.
Powered Wireless Sensor Nodes”, IEEE Sensors Journal, Vol. 7, No. 5, 2007. [59] S. O. Kasap, “Principles of Electrical Engineering Materials and Devices”,
[39] F. Völklein, et al., “Thermoelectric microsensors and microactuators (MEMS) McGraw Hill, 2000.
fabricated by thin-film technology and micromachining”, 18th Internacional [60] H. Zou, et al., “Peltier effect in a co-evaporated Sb2Te3(P)-Bi2Te3(N) thin film
Conference on Thermoelectrics, 1999. thermocouple”, Thin Solid Films, Vol. 408, Issues 1-2, pp. 270-274, 2002.
[40] Z. Wang, et al., “Realization of a wearable miniaturized thermoelectric [61] J. P. Carmo, et al., "Thermoelectric Microconverter for Energy Harvesting
generator for human body applications”, Sensors and Actuators A: Physical, Systems", IEEE Transactions on Industrial Electronics, Vol. 57, No. 3,
Vol. 156, pp. 95–102, 2009. pp. 861-867, 2010.
[41] J. Weber, et al., “Coin-size coiled-up polymer foil thermoelectric power [62] A. A. R. Elshabini, F. D. Barlow III, “Thin Film Technology Handbook”, 1st
generator for wearable electronics”, Sensors and Actuators A: Physical, Vol. Edition, McGraw-Hill, 1998.
132, Issue 1, pp. 325-330, 2006. [63] L. J. van de Pauw, “A method of measuring specific resistivity and hall effect of
[42] V. Leonov, and R. J. M. Vullers, “Wearable electronics self-powered by using discs of arbitrary shape”, Philips Research Reports, Vol. 13, No. 1, pp. 1-9,
human body heat: The state of the art and the perspective”, Journal of 1958.
Renewable and Sustainable Energy, Vol. 1, Issue 6, 2009. [64] S. A. Campbell, “The Science and Engineering of Microelectronic Fabrication”,
[43] D. D. L. Wijngaards, “Lateral on-chip integrated Peltier elements based on 2nd Edition, Oxford University Press, 2001.
polycrystalline silicon germanium”, PhD Thesis, Technische Universiteit Delft, [65] D. M. Mattox, “Handbook of physical vapor deposition (PDV) processing: film
2003. formation, adhesion, surface preparation and contamination control”, Noyes
[44] L. da Silva, “Integrated Micro Thermoelectric Cooler: Theory, Fabrication and Publications, 1998.
Characterization”, PhD Thesis, University of Michigan, 2005. [66] [Link] (acedido em
[45] L. M. Goncalves, “Microssistema termoeléctrico baseado em teluretos de 20/08/2010)
bismuto e antimónio”, PhD Thesis, University of Minho, 2008. [67] D. A. Glocker and S. I. Shah, “Handbook of Thin Film Process Technology”,
[46] L. D. Hicks, and M. S. Dresselhaus, “Effect of quantum-well structures on the Institute of Physics Publishing, 1995.
thermoelectric figure of merit”, Physical Review B, Vol. 47, Issue 19, [68] Ottonis de Guericke, “Experimenta Nova (ut vocantur) Magdeburgica De Vacuo
pp. 12727-12731, 1993. Spatio”, Amsterdam J. Jansson à Waesberge, 1672.
[47] R. Venkatasubramanian, et al., “Thin-film thermoelectric devices with high [69] M. Ohring, “Materials Science of Thin Films”, 2nd Edition, Academic Press,
room-temperature figure of merit”, Nature, Vol. 413, pp. 597-602, 2001. 2001.
[48] [Link] [70] A. E. Riad and F. D. Barlow III, “Thin Film Technology handbook”, McGraw-
[49] R. Ewell, et. al, “SP-100 Thermoelectric Cell Testing at JPL”, NASA Jet Hill, 1998.
Propulsion Laboratory. [71] Luciana W. da Silva and Massoud Kaviany, “Miniaturized Thermoelectric
[50] R. Venkatasubramanian, et. al, “Energy Harvesting for Electronics with Cooler”, ASME’02, New Orleans, USA, Nov 17-22, 2002.
Thermoelectric Devices using Nanoscale Materials”, IEEE Electron Devices [72] Luciana Wasnievski da Silva, Massoud Kaviany, Fabrication and Measured
Meeting, pp. 367-370, 2007. Performance of a First-Generation Microthermoelectric Cooler, J. of
[51] J. Heo, et al., “EARQ: Energy Aware Routing for Real-Time and Reliable Microelectromechanical Systems, v. 14, 5 (2005) pp. 1110,.
Communication in Wireless Industrial Sensor Networks”, IEEE Transactions on [73] Harald Böttner, et. Al., New Thermoelectric Components Using Microsystem
Industrial Electronics, Vol. 5, No. 1, pp. 3-11, 2009. Technologies, Journal of Microelectromechanical Systems, 13 (2004) pp. 414-
[52] J. Lee, et al., “Group Connectivity Model for Industrial Wireless Sensor 420.
Networks”, IEEE Transactions on Industrial Electronics, Vol. 57, No. 5, pp. [74] Cyrus Shafai, Fabrication of a micro-Peltier device, National Library of Canada
1835-1844, 2010. (1998) pp. 76,
[53] Bin Lu and V. C. Gung, “Online and Remote Motor Energy Monitoring and [75] L. M. Goncalves, et al., “Thermoelectric Micro Converters for Cooling and
Fault Diagnostics Using Wireless Sensor Networks”, IEEE Transactions on Energy Scavenging Systems”, Journal of Micromechanics and
Industrial Electronics, Vol. 56, No. 11, pp. 4651-4659, 2009. Microengineering, Vol. 18, No. 6, 2008.
[76] H. Zou, “Preparation and Characterization of bismuth Telluride and antimony
telluride thermoelectric thin films”, Phd Thesis, Cardiff UK, 2003.
[77] L. da Silva, “Micro Thermoelectric Cooler Fabrication: Growth and
Characterization of Patterned Sb2Te3 and Bi2Te3 Films” 22th International
Conference on Thermoelectrics, 2003.
[78] A. Giani, et al., “Growth of Bi2Te3 and Sb2Te3 thin films by MOCVD”,
Materials Science and Engineering: B, Vol. 64, Issue 1, pp. 19-24, 1999.
[79] A. Foucaran, et al., “Flash evaporated layers of (Bi2Te3–Bi2Se3)(N) and
(Bi2Te3-Sb2Te3)(P)” Materials Science and Engineering B, Vol. 52, Issues 2-3,
pp. 154-161, 1998.
[80] J. R. Lin, et al., “Thermoelectric Microdevice Fabrication Process and
Evaluation at the Jet Propulsion Laboratory (JPL)”, 21th International
Conference on Thermoelectrics, 2002.
[81] C. Kittel, “Introduction to Solid State Physics”, John Wiley, 1986.
[82] [Link]
[83] L.M. Goncalves, et al., “Optimization of thermoelectric properties on Bi2Te3
thin films deposited by thermal co-evaporation”, Thin Solid Films, Vol. 518,
pp. 2816-2821, 2010.
[84] L.M. Goncalves, et al., “Thermal co-evaporation of Sb2Te3 thin-films optimized
for thermoelectric applications”, Thin Solid Films, Vol. 519, pp. 4152-4157,
2011.
[85] D. Platzek, et al., “Potential-Seebeck-Microprobe (PSM): Measuring the Spatial
Resolution of the Seebeck Coefficient and the Electric Potential”, Proc. of 24th
International Conference on Thermoelectrics (ICT 2005), pp. 13-16, 2005.
[86] P. Reinshaus, et al., Proc. 2nd Europ. Symposium on Thermoelectrics-Materials,
Processing Techniques, and Applications, Dresden, Germany, pp. 90, 1994.
[87] D. Platzek, et al., “An Automated Microprobe for Temperature Dependent
Spatial Scanning of the Seebeck Coefficient“, Proc. 22nd International
Conference on Thermoelectrics (ICT2003), pp. 528-531, 2004.
[88] D. Platzek, et al., “Anisotropy of the Seebeck Coefficient Detected by the
Seebeck Scanning Microprobe“, Proc. 8th European Workshop on
Thermoelectrics, 2004.
[89] [Link]