NSF Workshop:
Emerging Opportunities of Nanoscience to Energy Conversion and Storage
Section 3: Nanoscience and Thermoelectric Energy
by Alexis R. Abramson
3.1. The Problem Addressed
Introduction
Thermoelectric devices are
especially attractive because they do not contain moving parts, they are
environmentally benign, and they may be easily incorporated into technologies
for heat removal or for energy conversion. Thermoelectric devices, however,
currently have limited use in the marketplace, which is largely a result of
their low efficiencies. One big question is whether nanoscience may offer the
potential for improvement of these efficiencies.
Conventional thermoelectric
devices used for either heating/cooling or energy conversion are typically
comprised of semiconductor alloys and are fabricated from pairs of elements
made of dissimilar thermoelectric materials (n-type and p-type). The
thermoelectric elements are connected electrically in series and thermally in
parallel, and in principle the same device could be used as both power
generator and heat pump. Figure 1 shows a schematic of a thermoelectric device.
Thermoelectric refrigeration proceeds via the Peltier effect, whereby passing a
current across a thermoelectric junction produces cooling. Similarly, imposing
a temperature gradient takes advantage of the Seebeck effect, and current flow
is generated, producing electrical power. The dimensionless thermoelectric
figure of merit, ZT, a measure of a
material’s thermoelectric efficiency, is given by 
, where S is
the Seebeck coefficient, T is
temperature, and s and k are
the electrical and thermal conductivity, respectively [[1]].
Even though clever manipulation of material properties may provide an
opportunity for improving room temperature ZT, there has been limited progress to develop novel
high performance thermoelectric materials. Nonetheless, promising research has
demonstrated that thermoelectric performance can be significantly enhanced in
low-dimensionality structures [[2]–[9]]; although there is a limited
understanding of the phenomena involved. Therefore, by replacing conventional
semiconductors/semimetals with high performance thermoelectric nanomaterials, a revolution in energy conversion and
generation and heat removal applications will undoubtedly ensue. In view of
this lack of understanding and its scientific and technological importance,
there exists a great opportunity for new discoveries using nanomaterials in
thermoelectrics.
Metrics for Comparison
Highly
efficient thermoelectric energy conversion requires large ZT values, i.e. large electrical conductivity (in order
to reduce the self-heating due to electrical current passing through the
device), high Seebeck coefficient (for large voltage in power generation and
large Peltier coefficient in cooling), and low thermal conductivity (to allow
for a large temperature difference and therefore large voltage in power
generation or to reduce the heat leakage between the hot and the cold side of
the device if used as a refrigerator). As shown in Figure 2, the typical
performance and efficiency of standard thermoelectric devices are inferior to
other comparable energy conversion technologies [[10]]. Typical household refrigerators
operate at performance levels about three times higher than typical
thermoelectric refrigerators, and other comparable power generators all
demonstrate much higher efficiencies as compared with a conventional
thermoelectric power generation device. It is important to note, however, that
even with increased efficiency, a power generation thermoelectric device
requires a temperature differential to exist for adequate electrical current
generation, and this may not be achievable in various applications.
Nonetheless, with an increase in efficiency/performance, these environmentally
benign, easy to maintain and inexpensive thermoelectric devices could certainly
be employed for various heating/cooling and power generation applications where
thermal energy harvesting is a possibility.
Theoretical Limits
In
conventional solids, it is difficult to manipulate the three parameters: S, s
and k, in order to obtain large ZT values. In fact, a modification to any of these
properties often adversely affects the others, such that Z does not change significantly [[11]]. Therefore, the room temperature
figure of merit of the best commercially available thermoelectric materials is
approximately one, while a ZT of
approximately four is necessary for thermoelectrics to replace competing energy
conversion technologies in various applications. Note that the thermoelectric
device efficiency approaches a fundamental thermodynamic limit (the Carnot
limit), and values of ZT greater
than approximately ten are not realistically achievable.
Nanostructures
and nanostructured materials hold the promise of more efficient thermoelectric
energy conversion. In nanostructures occur novel phenomena, properties, and
functions which were not possible to achieve in bulk materials. The reasons for
the appearance of the enhanced behavior are rooted in the way nanostructures
affect the transport of electrical charge and heat at nanoscale. Hicks and
Dresselhauss [4, 11, [12]] demonstrated that for suitable
nanostructures, quantum confinement of the electrical charge may enhance the
Seebeck coefficient and electrical conductivity even in existing materials
which do not exhibit good thermoelectric performance in bulk form. The quantum
confinement can be achieved, for example, via reduction of one dimension (e.g.
in a thin film structure), or of two dimensions (e.g. in a nanowire). Since the
quantum confinement effects become stronger by reducing the dimensionality,
large thermoelectric figures of merit have been calculated for nanowire systems
[5]. As
shown in Figure 3, more than an order of magnitude enhancement in the figure of
merit was predicted in BiTe nanowires [[13]].
Another
major contribution to enhancing the thermoelectric figure of merit in
nanostructures comes from the change in the thermal transport properties at
nanoscale. For example,
experimental investigations of heat transport in superlattices [[14] – [21]] show that although the layers are high
quality single crystals, the effective thermal conductivity may be much lower
than that estimated from the bulk values of the constituent materials, and even
smaller than the thermal conductivity values of the equivalent composition
alloys. It is believed that scattering of energy carriers at surfaces and
interfaces plays the most important role [[22]].
This effect is beneficial for thermoelectric energy conversion, and may further
enhance the predicted figures of merit predictions shown in Figure 3.
Therefore, it can be readily observed that a large
improvement of the thermoelectric efficiency can be obtained in principle by
combining both heat and electron transport effects in nanostructures. Order of
magnitude enhancements that can greatly benefit thermoelectric energy
conversion efficiencies have been predicted. More detailed descriptions of
recent developments in thermoelectric materials are reviewed in references [10, 13, [23]].
3.2. Conventional Technologies
Methods and architectures
Conventional thermoelectric
devices consist of a sequence of thermoelectric materials which are connected
electrically in series and thermally in parallel. As discussed previously,
efficiency is determined by the figure of merit ZT of the constituent n-type and p-type thermoelectric materials. Conventional thermoelectric materials
are typically bulk crystalline semiconductors, two examples being BiTe and
SbTe. It should be noted that a unipolar thermoelectric generator (i.e. a
single n-type or p-type region with metal contacts on either side) can
have this same maximum efficiency.
Limitations of current
methods
Limitations for current
thermoelectric generator architectures fall into two categories: limitations on
heat power captured and limitations on energy conversion efficiency. Regarding
the former, thermoelectric materials are typically single-crystal
semiconductors and thus cannot be deposited on arbitrary surface geometries in
order to maximize capturing the highest thermal gradient. A flexible
thermoelectric generator, on the other hand, could be conformally wrapped
around heat sources for enhanced heat transfer (and thus higher output power),
although this type of device is not typically manufactured. Moreover,
conventional methods for thermoelectric material deposition cannot achieve
large-area growth (~1m2) with low cost as required for energy
harvesting from large-area heat sources such as engines. Regarding the latter
group of limitations, thermoelectric generator efficiencies are limited by the ZT of the thermoelectric materials themselves.
Materials which have high electrical mobility due to crystal quality tend to
also have high thermal conductivity. Doping a material with low thermal
conductivity and low mobility to increase its electrical conductivity greatly
reduces its Seebeck coefficient. These practical tradeoffs in conventional
thermoelectric materials have led to thermoelectric generator efficiencies
which have remained at < 5% for the last several decades. It is only
recently with the advent of nanostructured thermoelectric materials that some
of these tradeoffs have been circumvented to create materials with much higher ZT that can more than double thermoelectric generator
efficiency.
3. Nanotechnology Approaches
How Nano Can Help
High thermoelectric efficiency
requires materials with large Seebeck coefficient and large electrical
conductivity, but small thermal conductivity. While these properties are
strongly coupled in bulk materials, material
structures in which at least one dimension is on the nanometer scale
dramatically alter the carrier transport properties due the confinement of the
carriers inside the material and provide possibilities to increase the figure
of merit [[24]]
In general, low dimensional semiconductors are a benefit to thermoelectrics due
to (1) increased Seebeck coefficient due the altered densities of states near
the Fermi energy, (2) carrier pocket engineering (3) increased phonon
scattering at boundaries (relative to electron scattering) for reduced thermal
conductivity and (4) the opportunity to physically separate the dopant
impurities from the carrier channel.
The possibility of obtaining a superior figure of merit in
low-dimensional materials has sparked a large interest in investigating
thermoelectric effects in nanostructures. Figure 4 shows the figure of merit in
the best nanostructured systems to data as compared with the bulk state of the
art materials. Thermoelectric materials
with enhanced performance have recently been realized in nanometer-scale
heterogeneous semiconductor material systems. In both the PbTe-based quantum dot superlattice system
demonstrated by Lincoln Labs [[25], [26]] and the Bi2Te3-Sb2Te3
two dimensional (2-D) superlattice system demonstrated by RTI [[27]], the carriers are weakly confined.
The enhancements are attributed to selective phonon scattering at the
interfaces; in fact, in both these thin-film systems, the power 
factor (numerator of the
expression for Z,
) is not significantly enhanced and is sometimes even lower
than the corresponding bulk material. Much larger enhancements in
thermoelectric performance are predicted in quantum wires and quantum dot
structures since this type of confinement results in sharp peaks in the density
of states and therefore should result in the highest increase in Seebeck
coefficient [8, [28], [29]]
Assemblies of nanometer-sized spheres or rods should offer the optimum
combination of enhanced Seebeck coefficient and phonon scattering. Miniband
formation in quantum dot solids or modulated nanowires offers additional
opportunities for engineering highly efficient materials. The primary
mechanisms for increased thermoelectric efficiency come from reduced
dimensionality of the material. However in order to access this “intrinsic”
increased efficiency, the nanostructures need to be incorporated into a
macroscopic structure while retaining the quantum confinement.
Key Problems Blocking Further Progress
While nanoscale heterostructures
(superlattices) have proven to increase the thermoelectric efficiency, much of
this research is directed toward thin-film (two-dimensional) microchip-based
technologies and other small scale niche areas. We feel that this increased thermoelectric efficiency is not
unique to these thin-film systems and it will be possible to achieve
large-scale, efficient direct thermal to electrical energy conversion using
other, perhaps more versatile, nanoscale materials. Here, we outline some of the key areas which need to be
addressed.
·
Large scale synthesis of nanomaterials
Not to minimize the outstanding
achievement of producing a material with ZT > 2 in thin-film systems [26, 27, [30]], it is hard to imagine that
materials manufactured using expensive processing of multilayer nanoscale films
in this way could be scaled up to replace the existing materials in conventional
thermoelectric devices. We believe that to address this challenge, other
nanoscale architectures and manufacturing avenues will have to be developed.
This includes large-scale chemical synthesis of nanomaterials with controllable
size, shape and surface chemistry. In addition to binary semiconducting
compounds of current interest (bismuth and lead chalcogenides), the carrier
concentration in these materials must be controlled so that the Fermi level can be adjusted to coincide with the
peak in the electronic density of states. Techniques for doping semiconductor
nanomaterials need to be developed. There are countless reports in the
chemistry literature on the synthesis of nanoparticles, nanorods and nanowires.
However there are actually few methods which allow even gram-scale synthesis of
material or that allow for systematic doping or stoichiometric control while
retaining the control over the particle morphology. Additionally, the potential
for using nanocomposites for thermoelectrics will require a better
understanding of interfacial science to ensure that properties are optimized
accordingly and enhancements can be taken advantage of in the presence of a
matrix.
·
Self Assembly
Equally important is the
methodology for the controlled assembly of the nanomaterials into larger
structures. We point out that most technologically important materials and
device structures involve heterogeneous materials systems; that is, material
structures contain more than one constituent material. Examples in
thermoelectrics, include Bi2Te3-Sb2Te3
alternating thin-film material [27]; PbTe-PbSeTe quantum-dot superlattices [26];
bismuth nanowires and their corresponding ceramic template [28]; and PbTe nanocomposites with Pb inclusions [[31]].
Robust methods are needed for the ordered assembly of two or three
component nanoparticle-based systems.
·
Theory
The theory of electronic transport in superlattice (2-D materials) and
quantum wires (1-D) materials is relatively well developed. While it is agreed
that 0-D materials should offer the largest enhancement in Seebeck coefficient
it is much less clear how to retain these enhancements while also achieving
acceptable electrical conductivities. Some theories exist, although these
usually address electron transport through a single quantum dot in a somewhat
idealized environment. These theories need to be extended to ensembles of
quantum dots with particular attention to transport across the interfaces. Specifically required are
self-consistent electrical and thermal transport theories for organic/inorganic
hybrid structures as well as for ordered and disordered quantum-confined
arrays. Theories for electron transport (including ballistic transport) in
periodic heterostructures which include coupling to phonons and electron and
phonon transport in nanostructures such as quantum wires/rods are also
required.
In conjunction with the theories, measurement techniques specific to
these nanoscale systems need to be developed. Currently, detailed
experimental and theoretical low-dimensional transport studies are performed on
planar geometries/structures fabricated by nanolithography. What is required
are techniques for self assembly of nanomaterials onto test structures designed
not only for electrical but thermal transport measurements as well.
Role of modeling/simulation
Developing thermoelectric
materials with high efficiency (i.e. high ZT) has, in the past, been a trial-and-error process in which the bulk
properties (thermal and electrical conductivities, Seebeck coefficient) of an
entire material system were characterized experimentally while varying material
composition. It has recently become possible, however, to create nanostructured
thermoelectric materials which have tailored properties. While this degree of
control has been demonstrated to have the potential for creating much higher ZT materials than would exist in bulk form, the
fabrication process for nanostructured thermoelectric materials remains mostly
a trial-and-error series of depositions with varying nanostructure geometries.
In order to predict and optimize
the thermoelectric performance of nanostructured materials for high ZT, modeling and simulation tools for these materials
must be developed. Several reasons exist to explain why progress in this area
has been slow: 1) It is difficult to create thermoelectric nanocomposites
(especially those based on quantum dots) with controlled geometries in order to
test theories, and 2) it is difficult to solve heat transport and electrical
transport self-consistently, especially in nanoscale systems where quantum
confinement and ballistic transport play a role.
Examples of modeling and
simulation work which would be beneficial for next-generation thermoelectric
materials include:
• Self-consistent
electrical and thermal transport in organic/inorganic hybrid structures
• Self-consistent
electrical and thermal transport in ordered/disordered quantum-confined arrays
• Hot-electron
transport in periodic heterostructures including coupling to phonons
• Ballistic
electron and phonon transport in nanostructures such as quantum wires/rods
Current status and future directions
There are two proven directions
to enhance the thermoelectric figure of merit in nanostructures: periodic
multilayer thin-film structures and nanowires. A schematic of typical sample
configurations is shown in Figure 5. In superlattices the enhancement of the
thermoelectric figure of merit can be obtained either along or perpendicular to
the film plane. If transport is perpendicular to the film plane, the material
system is chosen such that electrons can relatively easily transfer across the
interfaces (weak confinement), while the heat carriers (lattice vibrations) are
strongly scattered. This approach
requires materials systems with good electrical and Seebeck coefficient since
the power factor cannot be enhanced by quantum confinement effects. Using superlattices
as thermoelectric material has been investigated extensively since the multiple
interfaces (and the presence of additional quantum dots in some designs) impede
heat flow often without worsening electrical characteristics [3, 9, 12, 14 - 21, 25, 26, 30, [32]
– [41]].
In particular, this approach was demonstrated in a BiTe/BiSbTe material system,
where ZT = 2.4 was obtained [27] from a very effective reduction in thermal
conductivity. On the other hand if the structures are designed such that
transport is along the film planes, quantum confinement effects may now lead to
increases in the power factor. However this enhancement typically occurs only
in certain regions of the multilayer structures (the quantum well regions)
while other regions (barriers) do not contribute to thermoelectric transport.
This is a concern if the barriers are so thick to overcome the enhancements in the
quantum well regions. Nevertheless, a high thermoelectric figure of merit for
transport along the film plane was demonstrated in a PbTeSeTe/PbTe [26] quantum dot superlattice system. In this system,
alternative layers of quantum dots made from one material (PbTe) were separated
by layers of a different material (PbTeSeTe). Experiments showed enhancements
in the power factor and a large reduction of thermal conductivity along the
plane of the film in a PbTeSeTe/PbTe system. However, the reasons for the
enhancement in the power factor in PbTeSeTe/PbTe quantum dot superlatices seems
are not well understood. One possibility is that the quantum dots provide
selective scattering of electrons, which may lead to an increase of the Seebeck
coefficient [[42]]. On the other hand, theoretical
predictions [[43]] performed on a Ge/Si quantum dot
superlatice system indicate that enhancements are possible if quantum dots are
able to interact electronically to form minibands in the crystal. The findings
on quantum dot superlattices open a new research direction for quantum dot
materials with high ZT.
The
interest in nanowire systems stems from the large figures of merit predicted
for Bi,[5]
BiSb [6] and BiTe [13] nanowires. Nanowires have been synthesized by
various methods such as pressure injection, vapor deposition or
electrodeposition in amorphous alumina nanochanneled templates. Figure 6 shows
an electron microscope image of bismuth telluride nanorods synthesized using a
one-step templateless aqueous phase technique. Research in this area is recent
and ongoing [4 – 8, 28, [44]
– [48]]. For example, the enhancement in
the Seebeck coefficient in bismuth nanowires has been verified [28]. Furthermore, a prototype thermoelectric device
using nanocomposite materials comprised of Si nanowires and a polymeric
material [47] has been preliminarily demonstrated. However,
comprehensive experiments on one-dimensional nanostructures for thermoelectrics
are limited. Moreover, experimental characterization of all thermoelectric
properties in individual nanowires has proved extremely difficult, and direct
measurements of high ZT values in
nanowires have not been reported thus far.
While both two- and one-dimensional materials could benefit from
additional investigation to fully realize enhanced ZT, alternative nanomaterial configurations have
recently received some attention. Some of these alternative nanomaterials were
discussed and presented at the NSF Workshop on “Emerging Opportunities
of Nanoscience to Energy Conversion and Storage,” and include:
- regimented
quantum dot superlattices [43]
segmented nanowires [[49]]- three dimensional arrays of nanoparticles (potentially
self-assembled – see Figure 7) [[50],
[51],
[52]];
- one-dimensional chains of nanoparticles;
- oriented nanorods/nanowires [46, 47, 48]
- two-component nanoparticle assemblies;
- nanostructure/polymer (see Figure 8) or nanostructure/semiconductor
composites [47]
- reversible thermoelectric
nanomaterials [[53]]
- electronic device-level integration of micro/nanoscale
thermoelectrics for energy harvesting/cooling [[54],
[55],
[56]]
- bio-inspired thermoelectrics
Many of these directions are just
in the beginning stages, and experimental demonstrations of high ZT materials based on these approaches have not yet
materialized.
For many of the strategies to succeed, methods
are needed for assembly of organic/inorganic hybrid materials with controlled
spacing, size, and orientation of inorganic inclusions; fabrication of
organic/inorganic interfaces with chemically tailored properties (small
electrical resistance, high thermal resistance); reliably contacting large
arrays of quantum-confined structures and synthesis of organic materials with
high carrier mobility. All of these approaches will require much needed
additional fundamental understanding of micro/nanoscale electrical and thermal
phenomena and the development of improved experimental characterization
techniques. Novel nanoscale property nanoscale measurement approaches are under
development (e.g. see Figure 9).
Timetable and Technical Milestones
The fundamental metric for
evaluating thermoelectric materials is, of course, the figure of merit. An
enhancement in the figure of merit for a nanoparticle-based material system has
yet to be demonstrated. This represents the key milestone for this avenue of
research. For many of these nanomaterial systems, a figure of merit greater
than unity should be readily demonstrated within four years. Further
enhancement to the figure of merit well beyond unity should be achieved within
six years. Another technical milestone will be the ability to experimentally
measure the ZT of the actual
nanostructures themselves. Simple measurement techniques are expected to be
achieved within three years. This will further enable the goal towards
achieving superior thermoelectric nanomaterials.
It is important to realize that
while true momentum behind research in low-dimensional thermoelectrics began almost
a decade ago, a limited understanding of the complex phenomena underlying the
thermoelectric properties associated with low-dimensional structures has slowed
progress in this area. As more experimental and theoretical developments are
made, we will see numerous discoveries in the coming years, and the promise of
nanomaterials for thermoelectrics will be realized.