Subscriber access provided by University of Oklahoma
Article
High Thermoelectric Performance of a Heterogeneous PbTe Nanocomposite Hongchao Wang, Junphil Hwang, Matthew Loren Snedaker, Il-Ho Kim, Chanyoung Kang, Jungwon Kim, Galen D. Stucky, John Bowers, and Woochul Kim Chem. Mater., Just Accepted Manuscript • DOI: 10.1021/cm5042138 • Publication Date (Web): 13 Jan 2015 Downloaded from http://pubs.acs.org on January 22, 2015
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Chemistry of Materials is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 15
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Chemistry of Materials
High Thermoelectric Performance of a Heterogeneous PbTe Nanocomposite Hongchao Wang1, *, Junphil Hwang1,*, Matthew Loren Snedaker2, Il-ho Kim3, Chanyoung Kang1, Jungwon Kim1, Galen D. Stucky2,4, John Bowers4, Woochul Kim1, 1
School of Mechanical Engineering, Yonsei University, Seoul 120-749, Korea
2
Department of Chemistry & Biochemistry, University of California, Santa Barbara, Santa Barbara, California 93106, U.S.A.
3
Department of Materials Science and Engineering, Korea National University of Transportation, Chungju 380-702, Korea
4
Materials Department, University of California, Santa Barbara, Santa Barbara, California 93106, U.S.A.
Keywords: Thermoelectrics; Heterogeneous Material; Nanocomposite
Abstract: In this paper, we propose a heterogeneous material for bulk thermoelectrics. By varying the quenching rate of Na doped PbTe, followed by hot pressing, we synthesized heterogeneous nanocomposites, a mixture of nanodot nanocomposites and nanograined nanocomposites. It is well known that by putting excess amounts of Na (i.e., exceeding the solubility limit) into PbTe, nanodots with sizes as small as a few nanometers can be formed. Nanograined regions with an average grain size of ca. 10 nm is observed only in materials synthesized with extremely low quenching rate, which was achieved by using a quenching media of iced salt water and cold water. Dimensionless thermoelectric figures of merit, zT, of those heterogeneous nanocomposites exhibited a zT around 2.0 at 773 K, which is a 25% increase compared to zT of a homogeneous nanodot nanocomposite with the largest quenching rate in our experiment, i.e. furnace cooled. The power factor increase is 5% and the thermal conductivity reduction is 15%; thus, zT increase mainly comes from the thermal conductivity reduction. .
respectively. It is obvious that zT can be enhanced either by
INTRODUCTION
maximizing the power factor, S2/ρ, and/or by minimizing
Thermoelectric energy conversion can directly convert heat
thermal conductivity, κ. As shown in the expression, the zT is
into electricity or vice versa1-3. Thermoelectrics are scalable
composed of purely material properties, so the choice of right
and reliable but their application for waste heat recovery from
material in a designated temperature range ensures high
automotive, power plant, industrial waste heat, and body heat
conversion efficiency.
requires improvement in thermoelectric conversion efficiency.
As reviewed in several articles3-5, there has been active
The conversion efficiency of the thermoelectric device is
research on thermoelectrics after Hicks and Dresselhaus’s6,7
proportional to the dimensionless thermoelectric figure of
papers in 1993 where they proposed nanostructured materials
merit, zT = [S2/ρ(κl+κe)]T, where T is the absolute temperature,
for high zT. Indeed, these efforts led to an increase in the
S is the Seebeck coefficient, ρ is the electrical resistivity and κl
dimensionless thermoelectric figure of merit beyond ~ 2, and
and κe are the lattice and electronic thermal conductivities,
can be categorized into two different approaches: band
ACS Paragon Plus Environment 1/9
Chemistry of Materials
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 15
engineering for the power factor increase and nanostructuring
nanograined nanocomposite. It is known that thermal
for the thermal conductivity reduction. An example of power
conductivity of the heterogeneous material is less than that of
factor enhancement by band engineering was demonstrated by
the volume averaged thermal conductivity due to the
8
Heremans et al. , who showed that the Seebeck coefficient can
heterogeneous temperature gradient19-21. In addition to this, the
be increased due to the distortion of the electronic density of
mean distance between particles should be reduced because
states created by impurity atoms – thallium in their study – in
under a fixed nanodot concentration, nanodots have to reside
PbTe, yielding zT ~ 1.5 at 773 K. This study proved a concept
in a more confined region compared to the homogeneous
9
suggested by Mahan and Sofo , who suggested that the
nanocomposites; therefore, phonon scattering due to nanodots
Seebeck coefficient can be enhanced if there exists a local
increases so thermal conductivity decreases. Previously, Kang
increase in the density of states over a narrow energy range
et
10
al.22
compared
thermal
conductivities
of
nanodot
(i.e. a resonant level). Pei et al. proposed another example of
nanocomposites
power factor enhancement by band engineering through the
concluded that the nanograined nanocomposite could achieve
convergence of many valence (or conduction) valleys for a
lower thermal conductivity than the nanodot nanocomposite as
simultaneous increase in Seebeck coefficient and electrical
long as the nanograined nanocomposite has a sufficiently
conductivity. They achieved this by tuning the doping and
small grain size (i.e. a few nanometers in size). In our case, we
composition in bulk PbTe and showed zT value of 1.8 at about
found there are grains with a few nanometers in size and
850 K.
regions where nanodots reside, which led to a thermal
scatter phonons and thereby reduce the thermal conductivity and increase the thermoelectric figure of merit
nanograined
nanocomposites
and
conductivity reduction.
It is known that interfaces and boundaries of nanostructures 11-13
and
Lead telluride (PbTe) with a NaCl-type crystal structure is
. In bulk
one of the most promising thermoelectric materials in the
materials, two different types of nanocomposites were used for
temperature range of 600 to 900 K8,10,13-15,18,23-26. We
the zT increase – nanodot nanocomposites
14,15
and
previously reported zT ~1.7 at 773 K for Pb0.98Na0.02Te by
. Nanodot nanocomposites
optimizing the hot pressing conditions25. Later, we further
contain nanostructures embedded in a host material and
enhanced the zT to 2.0 at 773 K by quenching the PbTe melt
generally consist of single crystals. The figures of merit for Na
followed by the optimized hot pressing conditions23. In this
doped PbTe alloys were about 1.6, 1.5, 1.3 and 1.7 for
paper, we varied the quenching medium from liquid N2 (LN),
nanograined nanocomposites
insertion of Mg2Te
17
16
, CaTe and BaTe
18
, and strained
iced salt water (ISW), cold water (CW), air cooling (Air) to
endotaxial SrTe nanostructures , respectively. Recently,
furnace cooling (FC) to study the effects of quenching rate on
15
14
Biswas et al. reported a dimensionless figure of merit as 2.2
thermoelectric properties. We intended to form nanodots by
at 915 K in 2 mol% Na doped PbTe-SrTe alloys synthesized
putting excess amount of Na (i.e. exceeding the solubility
by spark-plasma-sintering (SPS). The increase in zT was
limit), into PbTe26. We found that the dimensionless figure of
attributed to the lower thermal conductivity by ‘all-scale
merit increased from 1.6 to 2.0 by increasing the quenching
hierarchical architectures’ – alloy atoms for scattering high
rate from FC to ISW, except for LN. An inhomogeneous
frequency phonons, nanodots for scattering mid- frequency
mixture of the nanodot nanocomposites and nanograined
phonons and mesoscale grains for scattering low frequency
nanocomposites were observed only in the ISW and CW
phonons. Nanograined nanocomposites are polycrystalline
samples. We attributed the low thermal conductivity of these
materials, with grain sizes on the order of 1–100 nm. Poudel et
materials as the main reason for such increase in zT.
al.
16
reported a peak zT of 1.4 at 100°C in a p-type
nanocrystalline BiSbTe bulk alloy. They attributed the zT increase to phonon scattering by 2- to 10-nm grains. In this study, we propose a heterogeneous nanocomposite – mixture of the nanodot nanocomposite and the nanograined nanocomposites – as a route for enhancing zT of a material. In this
case,
nanodots
only
form
inside
the
nanodot
nanocomposite and nanosized grains only exist in the
ACS Paragon Plus Environment 2/9
Page 3 of 15
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Chemistry of Materials 1 hour.
Electrical heaters supplied heat in the hot pressing
machine. The hot pressing condition used was based on an optimized condition for this material obtained in our previous work25. For comparison purpose, the grounded powders were also spark-plasma-sintered (SPSed) as well. The SPS conditions we used were similar to that of the reference paper by Biswas et al.14 (i.e. 50MPa at 773K for 10min). The pellet with a relative density of 99% or higher was used for simultaneous measurements of the thermoelectric properties. Cross-sectional
morphology
was
observed
using
a
JSM-6701F scanning electron microscope (SEM) by looking at the fractured surface of samples. Transmission election
Figure 1. A schematic showing material synthesis processes. The
microscopy (TEM) was carried out using a JEM-2100F
quenching media were liquid nitrogen (LN), iced salt water
microscope. Specimens for TEM were prepared by a focused
(ISW), cold water (CW), room air (Air) and air inside the furnace
ion beam (FIB) using a milling power below 5 keV as
(FC). The Qed ingot and QHed pellet are denoted as quenched
suggested by Baram and Kaplan28. An ULVAC ZEM-3 was
ingot and a hot pressed pellet of the Qed ingot, respectively.
used
to
measure
electrical
resistivities
and
Seebeck
coefficients simultaneously during heating from room
RESULTS AND DISCUSSION
temperature to 800 K. Thermal diffusivity was measured with
Samples of Pb0.98Na0.02Te were synthesized by melting,
a laser flash apparatus (Netzsch LFA 457). Heat capacity was
quenching and hot pressing (see Figure 1). High purity starting
estimated by the relationship, Cp (kB per atom)=3.07+4.7×10-4
elements of lead (Pb, 3N, Alfa Aesar), tellurium (Te, 4N, Alfa
× (T/K-300), which was obtained by fitting the experimental
Aesar) and sodium (Na, 99.95%, Alfa Aesar) were used for
data
the reaction. They were mixed in stoichiometric proportion
reported
by Blachnik
and
Igel29.
The
thermal
conductivity, κ, was extracted from the thermal diffusivity, λ,
and put into a carbon-coated quartz tube under N2 atmosphere
the specific heat capacity, Cp, and the density, d, based on the
in a glove box. The tube was flame-sealed under a pressure of
relationship κ=λCpd. Here, the density was measured by
approximately 10-4 Torr. The reaction mixtures were heated to
Archimedes’ method. The dimensionless figure of merit was
1073 K for 2 h followed by melting at 1273 K for 6 h. Then,
then calculated from the measured thermoelectric parameters.
the mixtures were quenched. To vary the quenching rate, we
We performed theoretical calculations focusing on the
used the different quenching media of liquid N2 (LN), iced salt
lattice thermal conductivity. The lattice thermal conductivities
water (ISW), cold water (CW), and room air (Air) as well as
were deduced based on the Wiedemann-Franz law using the
natural cooling in the reaction furnace (FC). The quenching
Lorenz number that we have calculated in our previous study23.
time, the time for the reaction mixture to reach room
In the paper, the Lorenz number was calculated based on the
temperature, was roughly estimated with the lumped
linearized Boltzmann transport equation (BTE) with an
capacitance approximation to be about 6, 15 and 30s for LN,
approximate relaxation time considering convergence of
ISW, CW, respectively as presented in Figure S2 in the
valence bands. The lattice thermal conductivity was calculated
supporting information. The quenching times for air
using Callaway’s model30. The details for simulating the
quenching and furnace cooling were measured approximately
Lorenz
to be about 0.5 h and 8 h, respectively. We found blankets of
number
and
11,12,22,23,30,31
elsewhere
vapors wrapping around the quartz tube especially in liquid
Callaway’s
model
are
available
, so it will not be repeated here. Effective
thermal conductivity should be calculated in a material
nitrogen quenching, which should inhibit heat transfer
containing regions of different thermal conductivities. Parrott
between the liquid nitrogen and the sample. Therefore, the
and Stuckes reviewed32 treatments on these cases. In this study,
actual quenching time should be slower than the estimated
we used the following relationship suggested by Klemens et
time 27. After the quenching, the ingots were then ground into
al.19-21 to calculate the effective thermal conductivity, κeff,
powder and subsequently hot-pressed at 100 MPa at 773 K for
ACS Paragon Plus Environment 3/9
Chemistry of Materials
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
κ eff = κ 0 −
1
κ0
∑ cos
2
θ κ ( q)
2
precipitated nanostructures are clearly observable. According , (1)
to previous reports14,33,34, the 2 % mole fraction of Na is
q
already beyond the solubility limit of Na in PbTe. Thus, excessive Na atoms tend to form nano-scale precipitates in the
where κ0 is the volume-averaged thermal conductivity, κ is the
form of Na2Te, or accumulate at the grain boundaries or defect
thermal conductivity as a function of a wave-vector q, and θ is
sites. We found the shape of the precipitates depends on the
the angle between each q and the direction of the average
zone axis. In the LN samples, for example, the precipitates are
temperature gradient. In our case, the nanodot nanocomposites
platelet shaped at the [100] zone axis. But at the [110] and
and the nanograined nanocomposites are distributed randomly,
[111], the shape is rounded or ellipsoidal. Therefore the shape
so we assume that they are isotropically dispersed. In this case,
of the precipitation is platelet-like but it looks different
the effective thermal conductivity can be expressed as
depending on the direction where it is being looking at, which
follows19-21:
κ eff
1 = κ0 − 3κ 0V0
Page 4 of 15
is consistent with observations by He et al.34. The bright field
∫ κ ( r ) − κ 0 dr , 2
(BF) and high angle annular dark field (HAADF) images in (2)
Figure S3 in the supporting information confirmed that the contrast in the TEM image is caused by compositional differences. Also, we deduced the lattice constant based on
where κ(r) is the thermal conductivity as a function of position
high resolution TEM images and found that the lattice
r and V0 is the volume of the domain of the integration.
constant in the nanodot region is smaller than that of the PbTe
Figure 2 shows micro- and nanostructures of the samples.
matrix, which is also consistent with findings by He et al. 34. It
Scanning electron microscope (SEM) images of as-quenched
was Na rich precipitation revealed in recent studies14,34,35.
ingots (i.e. Qed Ingot) are shown in Figure 2(a). Except for the furnace-cooled sample (FC), morphologies of all samples look similar; there are many tiny blobs throughout. The density of the tiny blobs is higher in the ISW and CW samples than in the others. Interestingly, these tiny blobs survived even after the hot pressing only in the ISW and CW samples as shown in Figure 2(b), where the SEM images of the hot pressed pellets of the quenched ingot (i.e. QHed pellets) are shown. Looking at these tiny blobs under the transmission electron microscope (TEM) in Figure 2 reveals that they consist of nanosized grains around 10 nm in size (i.e. the tiny blobs constitute the nanograined nanocomposite). As shown in Figure 2(b), the grain size of each samples increase when decreasing the quenching rate, except for the LN case. During the quenching process, we found the quenching rate of LN is slower than that of CW and ISW. In the liquid nitrogen quenching (LN), blankets of vapors generated during the quenching process blocked heat transfer between the liquid nitrogen and the molten sample, which should delay the quenching process leading increased grain sizes. In the Air and FC cases, grains greater than > 100 µm coexist with mid-sized (~10 µm) grains.
Figure 2. The micro- and nanostructures of samples. (a) SEM
For the cold water (CW) and ISW samples, variation in grain
images of as quenched ingot (Qed ingot). (b) SEM images of the
size is large; there are 20~60 µm sized grains and
