High Thermoelectric Performance of a Heterogeneous PbTe

Jan 13, 2015 - Miguel Angel Torres , Gustavo Garcia , Idoia Urrutibeascoa , Maria Antonieta Madre ... Andres Sotelo , Miguel Torres , Maria Madre , Ju...
3 downloads 0 Views 4MB Size
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