You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/344389640

From Dislocation to Nano‐Precipitation: Evolution to Low Thermal


Conductivity and High Thermoelectric Performance in n‐Type PbTe

Article  in  Advanced Functional Materials · December 2020


DOI: 10.1002/adfm.202005479

CITATIONS READS

14 97

4 authors, including:

Kunag-KuO Wang Hsin-jay Wu


National Sun Yat-sen University National Chiao Tung University
36 PUBLICATIONS   280 CITATIONS    61 PUBLICATIONS   889 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development and application of ultrasonic mechanical coating equipment for anti-corrosion treatment on aluminum and copper surface View project

All content following this page was uploaded by Hsin-jay Wu on 05 October 2020.

The user has requested enhancement of the downloaded file.


Full Paper
www.afm-journal.de

From Dislocation to Nano-Precipitation: Evolution to Low


Thermal Conductivity and High Thermoelectric
Performance in n-Type PbTe
Ping-Yuan Deng, Kuang-Kuo Wang, Jia-Yu Du, and Hsin-Jay Wu*

zT  = (S2σ/κ)T, gauges the thermal-to-


PbTe-based alloys have been widely used as mid-temperature thermoelectric electricity conversion efficiency of a TE
(TE) materials since the 1960s. Years of endeavor spurred the tremendous alloy, in which S is the thermopower, σ the
advances in their TE performance. The breakthroughs for n-type PbTe have electrical conductivity, κ the thermal con-
been somewhat less impressive, which limits the overall conversion efficiency ductivity, and T the absolute temperature.
of a PbTe-based TE device. In light of this obstacle, an n-type Ga-doped PbTe For decades, the conversion efficiency of
lead tellurides with dopants has steadily
via an alternative thermodynamic route that relies on the equilibrium phase
improved alongside the development
diagram and microstructural evolution is revisited. Herein, a plateau of zT = 1.2 of mid-temperature TE generators.[4–8]
is achieved in the best-performing Ga0.02Pb0.98Te in the temperature range of Through the carrier optimization[9–12]
550–673 K. Notably, an extremely high average zTave = 1.01 is obtained within and the all-scale microstructure engi-
300 − 673 K. The addition of gallium optimizes the carrier concentration and neering,[13–19] the zT values for both p-type
and n-type lead tellurides are enhanced.
boosts the power factor PF  =  S2ρ−1. Meanwhile, the κL of Ga-PbTe reveals a
Recently, the thermodynamic routes
significantly decreasing tendency owing to the defect evolution that changes emerged[20] as new paradigms to hunt
from dislocation loop to nano-precipitation with increasing Ga content. The for the low-κ and high-zT TE alloys, in
pathway for both the κL reduction and defect evolution can be probed by an which their TE performance is correlated
equilibrium phase diagram, which opens up a new avenue for locating high zT with the mutual solubilities of constituent
TE materials. elements, the microstructure, and the
phase stability. The above thermodynamic
approach (or the so-called phase-boundary
mapping[21,22]) can be easily realized by an
1. Introduction equilibrium phase diagram, which might be overlooked by the
TE community for the past few years.
To pursue environmental and economic sustainability, the Compared with the extraordinary zT values in the p-type
research trend seeks green and high-performance energy lead tellurides, such as the PbTe-SrTe with a peak zT  ≈ 1.8
resources, which aim to cut down the reliance on fossil fuel at 950 K,[23] the progress in zT enhancements of n-type lead
while retaining technological development. To this end, ther- tellurides seems to be sluggish, with only a few studies on
moelectric (TE) materials and devices, which can convert the (Ag, La)-PbTe, Mn-PbTe, and (I, Sb)-PbTe that achieved
ejected heat into electricity via the Seebeck effect,[1–3] attract the zT  >  1.5 at T  >  773 K.[24–26] Herein, we aim to enhance
research attention. Conventionally, the use of a figure-of-merit, the TE performance of n-type PbTe through doping with gal-
lium that features with the electron donor nature.[27,28] Before
thermal/electronic transport property optimization, the solu-
P.-Y. Deng, Prof. H.-J. Wu bility limit of gallium in PbTe, as well as the neighboring
Department of Materials Science and Engineering phase relations, shall be understood by using an equilibrium
National Chiao Tung University phase diagram, in order to forecast the maximum dopant
Hsinchu 30010, Taiwan concentration and the presence of secondary phases or even
E-mail: ssky0211@nctu.edu.tw
the precipitates. The addition of Ga is expected to bring addi-
Dr. K.-K. Wang
Department of Materials and Optoelectronic science
tional impurity energy levels falling way below the conduction
National Sun Yat-sen University band maximum (CBM),[29] which act as electron reservoirs
Kaohsiung 80424, Taiwan that constrain the extra donor electrons inside deep energy
J.-Y. Du well.[30] Even under a thermal activation, the deep impu-
Department of Chemical Engineering rity energy level helps to tune the electron carrier concentra-
National Tsing Hua University tion and pins the Fermi level to an optimal position for n-type
Hsinchu 30010, Taiwan
conduction.[11,31]
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adfm.202005479.
In addition, defect engineering enables the reduction in
lattice thermal conductivity κL and raise the zT value pro-
DOI: 10.1002/adfm.202005479 foundly.[32–34] The presence of crystal defects, including the

Adv. Funct. Mater. 2020, 2005479 2005479  (1 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

point defects, the dislocations, and the nano-precipitates, are the previous study.[41] The lightly doped x = 0.004 also reveals a
associated with the increased lattice strain fields, which could similar temperature dependence trend of its zT curve.
either enhance the phonon scattering or cause the lattice sof- As the Ga content (x) increases, the zT curves move upward
tening.[35,36] For the phonon scattering, the phonon propagates significantly. In the temperature range of 300–673 K, the
under a constant speed while the direction (vector) changes. In x  = 0.02 outperforms the rest of alloys, by not only yielding a
comparison, the lattice softening is accompanied only with a high peak zT  > 1.2 at 573 K but also showing an outstanding
decrease in the phonon speed, which is even more critical for average zT (zTave) ≈ 1.01. Compared with the previous Ga-
the κL reduction in the anharmonic materials.[37–39] For most of PbTe alloys,[31,41] our x  = 0.02 (Ga0.02Pb0.98Te) reveals a higher
TE materials, the two mechanisms virtually co-exist, and both zTave  = 1.01 within 300–673 K (Figure  1b). Further compared
account for the κL reduction.[40] with other n-type PbTe doped with Sb,[42] Ag,[24] Cu,[10] I/Sb,[26]
Most importantly, the present study provides the experi- and Ga[31,41] (Figure  1b and Table S1, Supporting Information),
mental evidence for the defect evolution in the n-type Ga-PbTe we found that our x  = 0.02 alloy possesses an extraordinary
alloys, whose carrier concentrations are simultaneously opti- zTave =  1.01,  which is by far one of the best-performing n-type
mized. The formation of the dislocation loop and Ga-rich nano- PbTe in the mid-temperature region (300–673 K). Even for the
precipitate realize the significant reduction in κ. In particular, less doped x = 0.01, the zTave is still as high as 0.84, suggesting
an equilibrium Ga-Pb-Te phase diagram probes the narrow that Ga could be an ideal dopant for utilizing the high-perfor-
compositional region where the defect evolution and κ reduc- mance n-type PbTe materials. The reproducibility for the high
tion takes place. As a consequence, the n-type Ga-PbTe alloy performance of x = 0.02 is verified by measuring three different
with a composition of Ga0.02Pb0.98Te achieves a peak zT ≈ 1.2 at samples with an identical nominal composition (Figure S1,
573 K, and yields a record-high zTave =  1.01 in the temperature Supporting Information). Additionally, thermal cycling between
range of 300–673 K. 300 and 673 K is conducted for one x = 0.02 sample (Figure S2,
Supporting Information) to confirm the thermal stability.
For comparison purposes, we prepared the other two
2. Results and Discussion series of Ga-PbTe alloys, the Gay(PbTe)1−y (y  = 0.004–0.012),
and the GazPbTe1−z (z  = 0.006–0.01), respectively. As shown
It is well known that the substitution of Pb by Ga in PbTe could in Figure  1a, the zT curves of Gay(PbTe)1−y move upward with
enhance the zT value.[31,41] The composition of Pb0.98Ga0.02Te the increasing y, and a moderate peak zT ≈ 0.68 is reached in
achieves a peak zT  = 1.34 at 766 K,[31] while a slightly Ga- y  = 0.012 at 650 K. However, a different tendency is revealed
rich Pb0.97Ga0.03Te possess an zT  = 1.3 at 823 K.[41] To explore for the GazPbTe1−z alloys whose zT curves go downwardly with
the mechanism yielding the high zT values, we revisit the increasing z. The lightly doped z  = 0.006 achieves the highest
GaxPb1−xTe alloys (x  = 0–0.06) using the Bridgman method. peak zT ≈ 0.72 at 625 K. In a previous case of I-PbTe,[43] the defi-
The temperature-dependent zT curve (Figure 1a) of undoped ciency of Te in a PbTe generally brings down the zT values as a
PbTe (x  = 0) declines with increasing temperature within result of an increase in κ. That explains the declining zT values
300–573 K. It then increases as the temperature is higher than in our GazPbTe1−z alloys.
600 K. This dramatical change in temperature dependence is due The success in boosting the zTave from 0.84 (x  = 0.01) to
to the co-existence of electron and hole carriers in the undoped 1.01 (x  = 0.02) within the mid-temperature region reveals that
PbTe that leads to the significant bipolar behavior at elevated tem- the n-type lead tellurides could have a better TE performance
perature. A similar behavior for the undoped PbTe is found in than what had been previously reported (Table S1, Supporting

Figure 1.  a) Temperature-dependent zT curves for the GaxPb1−xTe (x = 0–0.02), the Gay(PbTe)1−y (y = 0.004–0.012), and the GazPbTe1−z (z = 0.006, 0.01).
b) The averaged zT values calculated with 300–673 K, for various n-type PbTe with different dopants: the Sb-doped PbTe,[50] the Ag-doped PbTe,[44] the
Ga-doped PbTe,[31,41] the (I, Sb)-doped PbTe,[26] the Cu-doped PbTe,[6] and the (Ga, Ge)-doped PbTe.[55]

Adv. Funct. Mater. 2020, 2005479 2005479  (2 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

Figure 2.  a) The isothermal section of the ternary Pb-Te-Ga system at 673 K. b) The enlarged isothermal section superimposed with the color
contour of peak zT values at 673 K. The wavelength dispersive spectrum (WDS) maps for individual elements of Pb, Ga, and Te for c) alloy#33
(Ga-25.0at%Pb-55.0at%Te), and d) alloy#28 (Ga-10.0at%Pb-55.0at%Te) underwent a 30-day post-annealing at 673 K.

Information).[6,8,10,11,24–26,31,41–56] The breakthrough shall be Figure 3a shows the temperature dependence ρ(T) curves in
attributed to their microstructures and phases, in which a phase the temperature range of 300–673 K. For the x = 0 and x = 0.004
diagram could interpret the relationship. Figure 2a shows alloys, the ρ(T) curves transit from the metallic to semicon-
the isothermal section of ternary Pb-Te-Ga at 673 K, which is ductor conducting with increasing temperature. Meanwhile,
constructed by an experimental approach. The phase bounda- the S(T) curves for these two alloys (inset of Figure  3b) also
ries on three edges of Figure  2a are extracted from the binary reveal the transition from p-type to n-type conducting. With the
phase diagrams.[57–60] In order to determine the phase relation- increasing x and y, the ρ(T) curves move downward, implying
ships inside the ternary space, various Pb-Te-Ga ternary alloys that the incorporation of Ga increases the electron carrier
are thermally equilibrated by a post-annealing at 673 K for concentration nH. Table 1 summarizes the room-temperature
30 days. The equilibrium phase compositions for those ther- nH for three series of alloys. For GaxPb1−xTe, the nH gradually
mally equilibrated alloys (Table S2 and Figure S3, Supporting increases from nH =  3.03 × 1018 cm−3 (x = 0) to nH =  1.05 × 1019
Information) are put together to construct the phase diagram. (x  = 0.02), which explains the decreasing tendency in ρ. The
Taking two alloys, the alloy#33 (Ga-25.0at%Pb-55.0at%Te) Gay(PbTe)1−y also reveals a similar increasing nH with increasing
and alloy#28 (Ga-10.0at%Pb-55.0at%Te), as examples. The WDS y. Nevertheless, the increasing z in the Te-deficient GazPbTe1−z
elemental mapping results (Figure 2c,d) reveal the distribution oppositely leads to a decrease in nH, which corresponds to the
of Pb, Te, and Ga elements, inferring that the alloy #33 falls in rise in their ρ(T) curves (Figure  3a). The mobility μH for all
a PbTe + Ga6PbTe10 two-phase region while the alloy #28 locates the Ga-PbTe alloys is boosted compared with that of undoped
in a PbTe + Ga6PbTe10 three-phase region. The X-Ray diffrac- PbTe (x = 0), indicating that the incorporation of Ga plays a sig-
tion (XRD) patterns (Figure S4, Supporting Information) for nificantly positive role upon the electrical transport properties.
selective thermally equilibrated alloys are in good agreement Almost all the Ga-PbTe alloys reveal n-type conduction
with their metallographic observations (Figure S5, Supporting within 300–673 K, as confirmed by their negative S values
Information). It is worth noting that the PbTe phase in a binary (Figure  3b). For the GaxPb1−xTe and Gay(PbTe)1−y, the S(T)
Pb-Te phase diagram[58] covers a small yet non-stoichiometric curves declines with increasing Ga content, which are asso-
compositional homogeneity, which extends toward the Te-rich ciated with the increasing nH (Table  1). In contrast, the S(T)
side. With soluble Ga, the homogeneity region of PbTe still curves of GazPbTe1−z show a downward shifting with increasing
retains asymmetric, as shown in the enlarged isothermal sec- z, owing to the decreasing nH. The thermoelectric power factor
tion (Figure  2b). Notably, the nominal compositions for three (PF) = S2/ρ (Figure 3c), generally decreases with increasing tem-
series of alloys (the GaxPb1−xTe (x  = 0–0.06), the Gay(PbTe)1−y perature. Among them, the PF curve of x  = 0.02 outperforms
(y = 0.004–0.012), and the GazPbTe1−z (z = 0.006, 0.01)) fall inside the other alloys, which reaches a peak value of 4.4 mW m−1 K−2
the PbTe single-phase region, except for the x = 0.06 alloy. The at 300 K. This ultra-high PF value is 200% higher than that of
magnified isothermal section is further superimposed with the x  = 0 (PF = 2.0 mW m−1 K−2) at the same temperature. Fur-
color contour of peak zT values at 673 K. The high-zT zone for thermore, a calculated PF curve (denoted as optimal PF) using
Ga-PbTe alloys falls across the maximum solubility of Ga in the single Kane band (SKB) model is added in Figure 3c. One
Te-enriched PbTe. can conclude that the optimal PF curve describes a similar

Adv. Funct. Mater. 2020, 2005479 2005479  (3 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

Figure 3.  The temperature-dependent a) electrical resistivity (ρ(T)) curves, b) Seebeck coefficient (S(T)) curves, c) the PF(T) = S2σ curves for
the GaxPb1−xTe (x = 0–0.02), the Gay(PbTe)1−y (y = 0.004–0.012), and the GazPbTe1−z (z = 0.006, 0.01). d) The average power factor calculated within
300–673 K, for various n-type PbTe with different dopants: the Ag-doped PbTe,[44] the Sb-doped PbTe,[50] the Ga-doped PbTe,[31,41] the (Ga, Ge)-doped
PbTe,[55] the (I, Sb)-doped PbTe,[26] and the Cu-doped PbTe.[6]

Table 1.  Electrical resistivity, Seebeck coefficient, thermal conductivity, Hall carrier concentration, and mobility for the GaxPb1−xTe (x  = 0–0.02), the
Gay(PbTe)1−y (y = 0.004–0.012), and the GazPbTe1−z (z = 0.006, 0.01) measured at 300 K.

Sample Electrical resistivity(ρ) Seebeck coefficient(S) Thermal conductivity(κ) Carrier concentration(nH) Carrier mobility(μH)
[mΩ cm] [μV K−1] [W m−1 K−1] [1018 cm−3] [cm2 V−1 s−1]
x = 0 3.11 298.91 2.30 3.03 1202.50
x = 0.004 3.29 328.48 2.31 0.83 1033.00
x = 0.006 4.66 −187.53 2.33 2.31 1291.78
x = 0.008 2.45 −234.17 2.53 5.16 1347.20
x = 0.01 0.88 −190.90 2.68 8.27 1688.00
x = 0.02 0.41 −132.83 2.52 10.50 1916.00
x = 0.06 0.96 −131.64 4.39 31.39 264.80
y = 0.004 3.72 −218.37 1.97 2.49 1903.00
y = 0.008 1.77 −222.35 2.14 4.50 1395.00
y = 0.012 1.12 −211.10 3.65 8.52 1777.00
z = 0.006 1.18 −173.48 2.15 5.32 1252.00
z = 0.01 1.73 −204.96 2.69 4.26 1955.25

Adv. Funct. Mater. 2020, 2005479 2005479  (4 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

temperature dependence trend compared with that of the curve and experimental results is observed, presumably owing
x  = 0.01 and x  = 0.02 alloys. In addition to the high peak PF to the complex electronic band structure[31] and co-existence of
value, the x = 0.02 also presents an extraordinarily high average dual carriers in those lightly doped or overly doped Ga-PbTe
PF (PFave) of 3.37 mW m−1 K−2 in the temperature range of alloys. Nevertheless, our best-performing x  = 0.02 alloy falls
300–673 K (Figure  3d). This value is higher than most of in an optimal region of nH ≈ 1019 cm−3, whose electronic trans-
the previously reported n-type PbTe alloys, including the hot- port properties can be predicted by the SPB and SKB model. In
pressed Cu-PbTe (PFave  ≈ 3.01 mW m−1 K−2) and the sparkle- other words, the remarkable PF and PFave in our x = 0.02 alloy
plasma-sintering Ga-PbTe (PFave  ≈ 2.99 mW m−1 K−2), within are due to the optimal nH and enhanced μH (1916 cm2 V−1 s−1)
the same temperature range.[6,31] that combinatorially lead to high electrical conductivity (σ = ρ−1).
Although extrinsic doping offers an effective way of carrier Details of SKB and SPB models can be found in the Supporting
optimization and zT enhancement, it inevitably brings con- Information as well as in the previous work done by Su et al.[31]
cern on whether the band structure and mechanism of car- The thermal conductivity κ for all the Ga-PbTe alloys are
rier scattering are affected. For example, the substitution of Pb measured within 300–673 K (Figure 5a). As known, the κ com-
by a cation, such as the Bi,[60] Ti,[61] I,[62] in n-type PbTe would prises two primary contributors, the electronic thermal con-
lower the carrier mobility due to the deterioration of conduc- ductivity κe and the lattice thermal conductivity κL, respectively.
tion band. In contrast, the intercalation of copper in n-type Based on the Weidemann–Franz law, the κe can be expressed
PbTe (CuxPbTe1−x)[6] yields the best n-type performance, as a as κe = LTρ−1, whereas the Lorenz number L falls in the range
result of retaining the electronic band structure. Nevertheless, of 2.44  × 1018 to 1.5  × 1018 WΩ K−2 for degenerate to non-
the gallium, which is known to have electron donor nature, degenerate semiconductors. Herein, we use the SPB model to
could create a deep impurity level, which pins the Fermi level calculate the L number (Figure S6, Supporting Information),
on an optimal position even under the thermal activation.[31] to ensure the accuracy in the estimation of κL  =  κ  −  LTρ−1
The existence of this impurity level explains the temperature- (Figure  5b). At 300 K, the κ increases with increasing doping
insensitive and low-lying ρ(T) curves (Figure 3a), and therefore content for all the Ga-PbTe alloys, owing to the increasing
results in the extraordinary PF and PFave. κe (Figure  5d). As the temperature elevates, the κ(T) and
In order to validate the scattering mechanism in our κL(T) curves decline except for that of undoped x  = 0. For the
Ga-PbTe alloys, the Pisarenko curves using the single para- x = 0.01 and x = 0.02 alloys, their κ(T) curves move down sig-
bolic band (SPB) and the SKB model with the same effective nificantly with the increasing temperature, become the lowest
mass of 0.3 me are shown together in Figure 4a, in which the ones among all the alloys as T > 573 K, and reach an extremely
experimental data are further superimposed. At 300 K, the low κ = 1.25 W m−1 K−1 at 673 K. Notably, the κL(T) curve of the
experimentally determined S and nH from this work (denoted x  = 0.02 (Figure  5b) remains low lying within 300–673 K, and
by filled circles) and by others[24,31,41] are in good agreement achieves an ultra-low κL = 0.85 W m−1 K−1 at 673 K.
with that of SKB and SPB-estimated curves, especially when The specific heat Cp for all the Ga-PbTe alloys are measured
nH  >  5 × 1018 cm−3. Figure  4b summarizes the experimentally within 350–575 K (upper panel of Figure  5c), and the values
determined μH and nH for our Ga-PbTe alloys and the other locate in the similar range as that calculated by an equation:
n-type PbTe alloys.[24,31,41] A solid curve from the SKB model is Cp/kB per atom = 3.07 + (0.00047 × (T − 300)).[46,47] Good agree-
superimposed to guide the nH–μH relationship, which generally ment is achieved between the experimentally determined Cp
gives a satisfactory estimation for our Ga-PbTe alloys especially and calculated Cp,[46,47] suggesting that the ultra-low κ and
when nH is close to 1019 cm−3. The deviation between the SKB κL for x  = 0.02 alloys are convincible. The x  = 0.06 exhibits a

Figure 4.  a) Pisarenko relation of Ga-PbTe at 300 K, compared the predicted results (curves) with the effective mass of 0.3 me according to the single
parabolic model and SKB model. Note that the literature data (La-PbTe,[24] Ga-PbTe[31,41])fits with the curve of 0.3 me. b) The relation between Hall
mobility (μH) and carrier concentration (nH) at 300 K.

Adv. Funct. Mater. 2020, 2005479 2005479  (5 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

Figure 5.  Temperature-dependent a) thermal conductivity (κ(T)) curves, b) lattice thermal conductivity (κL(T)), c) specific heat capacity (Cp(T)), and
d) electrical thermal conductivity (κe(T)) curves for the GaxPb1−xTe (x  = 0–0.02), the Gay(PbTe)1−y (y  = 0.004–0.012), and the GazPbTe1−z (z = 0.006,
0.01). The inset in (b) showed enlarged section of (κL(T)) curves within the high-temperature region. The κ(T) calculated by the Debye–Callaway model
is superimposed in (b). The high-temperature lattice thermal conductivity (κL) of Ga0.02Pb0.98Te approach the amorphous limit of cubic structure
(≈0.4 W m−1 K−1).

high-rise κ(T) curve, presumably due to the co-existence of diffraction spot of (220) in Figure 6c, the DF image (Figure 6b)
PbTe and Ga6PbTe10, as suggested by the isothermal and further reveals the bright streaks and spots, which correspond to the
confirmed by experimental results (Figure S7, Supporting Infor- linear dislocation and dislocation loops in Figure 6a. The high-
mation). On the contrary, the x = 0.01 and x = 0.02 appear to be resolution image (Figure 6e) reveals a closer observation upon
single phase of PbTe (Figure S7, Supporting Information). the dislocation loop, which suggests a severe lattice distortion.
Nevertheless, the transmission electron microscope (TEM) In order to quantify the distortion, the lattice profiles
analyses discussed below uncover a different story. (Figure 6g) are analyzed across the edge and center of the dis-
Figure 6a–g shows the TEM analyses results for the x = 0.01 location loop, as denoted by green and blue lines in Figure 6e,
alloy. The bright-field (BF) image (Figure 6a) corresponds to the respectively. The peaks are overlapping and broadening when
diffraction vector g = 2 20 in the [111] zone axis, revealing that the the line-scan is across the distorted area (upper panel of
PbTe matrix is embedded with a large number of dislocations. Figure 6g), indicating that the lattice shrinks gradually along the
The diffraction pattern (Figure  6c) affirms the cubic structure 111 direction. As a comparison, the lower panel in Figure  6g
of the matrix phase. Nevertheless, the energy dispersive spec- suggests the less-distortion area, as the peaks are sharp and are
trum (EDS) (Figure  6d) suggests a small yet detachable solu- separated with nearly an equal space. Moreover, another sample
bility of Ga within the measured area. Moreover, the linear of x  = 0.01 is analyzed by TEM, and its BF image (Figure  6f)
and enclosed dislocations (aka dislocation loop) are uniformly also confirms the presence of dislocation loops. Good reproduc-
dispersed (Figure  6a), which originate from the lattice imper- ibility in the TEM analysis results can be achieved in different
fections. The dark-field (DF) image is suitable for disloca- samples of x = 0.01 alloy.
tion imaging, as the bright region in the DF mode represents With the increasing Ga content, the defects are likely to
the bending lattice near the dislocation cores. By taking the transform from 2D dislocations to 3D nano-precipitation.

Adv. Funct. Mater. 2020, 2005479 2005479  (6 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

Figure 6.  The TEM analyses of Ga0.01Pb0.99Te (x  = 0.01) in the [111] zone axis, and the corresponding a) BF image, b) the dark-field (DF) image in
the [ 220 ] two-beam condition, c) the diffraction pattern. d) The EDS of the whole specimen. e) The high-resolution TEM image in the [112 ] zone axis.
f) The BF image of another x = 0.01 sample in the [111] zone axis. g) Lattice profiles of both lines in the [ 1 1 1] direction of (e).

Figure 7a–l summarizes the TEM analyses results for the coherent, as shown by a high-resolution image (Figure  7g). In
x  = 0.02 alloy. The BF images (Figure  7a,d) are taken in two order to validate the formation of nano-precipitate, the high-
different zone axes of [001] and [021], respectively, in different angle annular dark-field (HAADF) images in the [001] zone
regions for the x = 0.02 specimen, which both reveal the square- axis are taken (Figure  7h,i), by using the scanning transmis-
shaped nano-precipitate with a characteristic size of 20–40 nm. sion electron microscope (STEM) mode. Consistently, the low-
The diffraction patterns (Figure 7b,e) also confirm the existence magnification STEM-HAADF images uncover the uniform
of cubic structure. Bright regions in the DF images (Figure 7c,f) dispersion of nano-precipitate. Since a HAADF image reflects the
further imply the presence of the strain field induced by the difference in mass density, the darker nano-precipitate, by all
nano-precipitate. In particular, the DF image (Figure  7f) is means, possesses a lighter atomic mass compared with that of
obtained using the satellite spot nearing the (400) reflection, matrix PbTe. Given that Ga is lighter than Pb and Te, the nano-
as highlighted in an inset of Figure  7f. That suggests the co- precipitation is likely to be a Ga-rich cubic phase. There are
existence of two cubic structures with very similar lattice con- two types of Ga ions, the Ga+(0.81 Å) and Ga3+(0.62 Å), which
stants. It is, therefore, speculated that the nano-precipitate also both possess a smaller ionic radius than Pb2+(1.2 Å). No matter
crystallizes in a cubic structure and possesses a similar lattice which Ga ion substitutes the Pb2+, it is expected that a lattice
constant compared with that of PbTe matrix. Another evidence mismatch takes places and therefore induces the formation of
for the structural similarity between the nano-precipitation and defects.[55,63] Nevertheless, it still acquires more experimental
matrix PbTe is revealed by the splitting of diffraction spots from evidence to determine whether the Ga ions would substitute
high-order reflections (444) and (224) (Figure 7k,l). the Pb2+ and which Ga ion (Ga+ or Ga3+) plays an essential role.
Upon closer observation, the interface between the nano- Meanwhile, the HAADF images confirm the presence of nano-
precipitate and PbTe matrix (region A in Figure  7a) is nearly scale mass fluctuations, contributing from two identical cubic

Adv. Funct. Mater. 2020, 2005479 2005479  (7 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

Figure 7.  The TEM analyses of Ga0.02Pb0.98Te including the a) BF image, b) diffraction pattern, and c) dark-field (DF) image in the [001] zone axis, d)
BF image, e) diffraction pattern, and f) DF image in the [021] zone axis, showing aligned nano-precipitate with rectangular-shaped region embedded in
the PbTe matrix. g) The high-resolution TEM image of the rectangular-shaped region in (a) in the [001] zone axis, indicating a well-connecting boundary
between precipitate and PbTe matrix. h,i) The STEM, HAADF images with two scales of magnification of the PbTe in the [001] zone axis, providing the
aspect to observe the mass contrast. j) The EDS of the whole specimen. k,l) Splitting spots appeared in the high-order diffracted plane, which were
444 and 224, respectively.

3 θ D /T
structures (matrix PbTe and nano-precipitate) with slightly dif- kB  kBT  x 4e x
ferent mass density, which benefits κ reduction.
κL =  
2π 2υ    ∫τ
0
tot (x )
(e x
− 1)
2 dx (1)

In short, the defect transits from the dislocation loop to


nano-precipitation with increasing Ga solubility in PbTe. In Equation (1), the kB, ν, ℏ, θD refers to the Boltzman con-
That demonstrates two scattering mechanisms, as shown stant, the average speed of sound, the reduced Planck constant,
in Figure 8a,b, respectively. If the dislocation loop forms as and the Debye temperature, respectively. The x  =  ℏω/kBT con-
a major defect, the speed of phonon would be slowed down tains the angular frequency ω. The τtotal is the total relaxation
while the direction of phonons (a vector) is unchanged when time that comprises the scattering of the normal (N process,
they travel through the distortion area. That also refers to the τN), the Umklapp phonon–phonon (U process, τU), the point
phonon softening, as shown in Figure  8a and found in our defect (PD) (τP),[65,66] the interface (I) (τI),[67] the dislocation core
x  = 0.01 alloy. Nevertheless, those soluble Ga (blue atom in (DC) (τDC),[68] and the dislocation strain field (DS) (τDS)[69] via
Figure  8a,b) could also induce the phonon scattering that the relation:
lowers the κL. The presence of nano-precipitate fulfills the typ-
ical phonon scattering mechanism, as illustrated in Figure 8b 1 1 1 1 1 1 1 (2)
= + + + + +
and demonstrated in our x  = 0.02 alloy. The color contour τ total τ N τ U τ P τ I τ DC τ DS
of κL at 673 K is superimposed on the isothermal section
(Figure 8c), and the superposition suggests that the decreasing Details of κL calculation based on the Debye approximation
tendency in κL could be mapped out by a phase diagram. The is included in the Supporting Information. Parameters for κL
x  = 0.01 and x  = 0.02 alloys, which contained different types calculation are listed in Table S3,[70,71] Supporting Information.
of defects, achieve the ultra-low κL  ≈ 0.99 (W m−1 K−1) and It should be noted that the density of dislocation and interface
κL ≈ 0.85 (W m−1 K−1) at 673 K, and κL ≈ 2.12 (W m−1 K−1) and is estimated mainly from the TEM analysis results (Figures  6
κL ≈ 1.19 (W m−1 K−1) at 300 K, showing a great reduction com- and 7). A similar κL calculation has also been performed for the
pared with the undoped PbTe. Moreover, a theoretical model p-type (Na, Eu)-PbTe.[67] Emphasis needs to be made that the
for κL based on the Debye approximation[64] offers a rational umklapp and normal (UN) process stands for the combined
way to compute the individual contribution of each type of contribution of U process and N process, while the interface I
defect on the κL reduction: scattering is mainly due to the nanostructuring.

Adv. Funct. Mater. 2020, 2005479 2005479  (8 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

Figure 8.  Schematic illustration of phonon scattering and softening due to a) dislocation loop and b) nano-precipitation, were generated in the samples
of x = 0.01 and x = 0.02, respectively. c) The mapping of lattice thermal conductivity (κL) at 673 K superimposed with enlarged isothermal section of
Ga-Pb-Te. d) The experimentally determined κL of pristine PbTe (x = 0), Ga0.01Pb0.99Te (x = 0.01), and Ga0.02Pb0.98Te (x = 0.02) superimposed with the
Debye–Callaway models that consider the normal scattering (N process), the Umklapp phonon–phonon scattering (U process), the PD scattering,
and the interface scattering (I), respectively.

Figure 8d summarizes the temperature-dependent κL curves Ga-PbTe alloys. Through phase diagram engineering, a specific
for x = 0, x = 0.01, and x = 0.02 alloys, superimposed with the compositional region reveals the decreasing trend in κL with
calculated κL curves based on Equations (1) and (2). The pres- increasing Ga content, in which the defects transits from the
ence of dislocation loops (i.e., point defect, x  = 0.01) dramati- dislocation loop to nano-precipitation. Experimental evidence
cally reduces the κL by 40–60% compared with x  = 0, while suggests that the coherency strains increase across the perfect
the formation of nano-precipitation (x  = 0.02) further leads to match of interfaces between those defects and the matrix PbTe,
another 40% and 10% drop in the κL at 300 and 673 K, respec- and hence the transporting phonons are severely scattered
tively. The experimentally determined κL curves of x = 0.01 and while the electrons pass without much interruption.
x  = 0.02 alloys fit very well with the model predictions that Astonishingly, the interplay of an ultra-high average
consider the UN + PD process and the UN + PD + I process, PF = S2ρ−1  ≈ 3.37 mW m−1 K−2, and low-lying κL curves result
respectively, inferring that in addition to the U and N process, in a high-rise plateau of zT ≈ 1.2 within 573–673 K, yielding an
the dislocation loop (point defect, PD) and the nano-precipitate extraordinary average zT  = 1.01 in n-type Ga0.02Pb0.98Te, in the
(nanostructuring interface, I) indeed enhance the phonon scat- temperature range of 300–673 K.
tering in these two alloys. Moreover, there is a noticeable devia-
tion between the experimentally determined κL and the UN
process in the x = 0 (undoped PbTe) as T > 400 K, reflecting the 4. Experimental Section
possibility of existence of bipolar thermal conductivity κb.
Synthesis: Thirty-six ternary alloys underwent a 30-day post-annealing
determined the Ga-Pb-Te isothermal section at 673 K. In a total of 1  g,
3. Conclusion the high-purity elements of Ga (99.999%, Alfa Aesar), Pb (99.95%,
Alfa Aesar), and Te (99.999%, Alfa Aesar) were weighed according
In summary, a successful paradigm of reduced κL via defeat to the predetermined nominal compositions as listed in Table S2,
evolution and carrier optimization is elaborated in n-type Supporting Information and were sealed in a quartz tube under vacuum

Adv. Funct. Mater. 2020, 2005479 2005479  (9 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

(3.0 × 10−3 Pa). The ampoules were homogenized at 1273 K for 5 h and Acknowledgements
were subsequently quenched in ice water to retain the fine solidification
microstructures. Subsequently, the sample ampoules were annealed at The authors acknowledged the financial support from the Young Scholar
673 K for 30 days and were quenched in the water again. Those thermally Fellowship Program by the Ministry of Science and Technology (MOST)
equilibrated alloys were subjected to metallographic observations and in Taiwan, under the Grant MOST 109-2636-E-009-016.
crystal structure analysis to identify the types and compositions of the
equilibrium phases.
As for the thermoelectric property assessments, the Bridgman
method was chosen to grow the fully dense Ga-PbTe alloys whose Conflict of Interest
relative density is larger than 99%. Three series of Ga-PbTe alloys, The authors declare no conflict of interest.
the GaxPb1−xTe (x  = 0 – 0.06), the Gay(PbTe)1−y (y  = 0.004–0.012), and
the GazPbTe1−z (z  = 0.006–0.01) were prepared through the Bridgman
method. The pure elements of Ga, Pb, and Te in a total of 8  g were
loaded in a carbon-coated quartz tube that was subsequently sealed Keywords
under vacuum (3.0 × 10−3 Pa). The sample ampoules were slowly heated
up to 1273 K in a furnace over 20 h, homogenized at this temperature for dislocation, nano-precipitation, n-type PbTe, thermal conductivity,
12 h, and were solidified by a water-quenching process. At the beginning thermodynamic approaches
of the Bridgman growth, the as-solidified ingots were placed in the top
and high-temperature region (T = 1213 K) of a Bridgman furnace where Received: June 30, 2020
the temperature was intentionally set to be higher than the materials’ Revised: July 21, 2020
melting points. Once the ingot became a melt, the sample ampoule Published online:
moved downward with a constant growth rate of 1.1 (K h−1) while rotated
axially until it reached to a low-temperature zone (T  = 1163 K) where
the melt solidified completely. The as-grown TE alloys were polished
and subjected for the following characterization and TE property [1] F. J. DiSalvo, Science 1999, 285, 703.
measurement. [2] L. E. Bell, Science 2008, 321, 1457.
Characterization: Microstructure analysis and phase identification [3] E. S. Toberer, G. J. Snyder, Nature 2008, 7, 105.
were conducted on both the thermally equilibrated and Bridgman-grown [4] K. Biswas, J. He, I. D. Blum, C. I. Wu, T. P. Hogan, D. N. Seidman,
alloys. Before the metallographic observation, the alloys were mounted V. P. Dravid, M. G. Kanatzidis, Nature 2012, 489, 414.
in the epoxy resin and polished with a series of SiC papers, ranging [5] Y. Z.  Pei, X.  Shi, A. D.  LaLonde, H.  Wang, L.  Chen, G. J.  Snyder,
from#1000 to #4000, and with Al2O3 powder with the particle size of Nature 2011, 473, 66.
0.1 and 0.05 µm, respectively. The microstructure was observed using [6] L.  You, J. Y.  Zhang, S. S.  Pan, Y.  Jiang, K.  Wang, J.  Yang, Y. Z.  Pei,
a field-emission scanning microscope (FESEM, JEOL 6330) equipped
Q.  Zhu, M. T.  Agne, G. J.  Snyder, Z. F.  Ren, W. Q.  Zhang, J.  Luo,
with a backscattered electron (BSE) detector, while the composition
Energy Environ. Sci. 2019, 12, 3089.
of the phase was obtained through a field-emission electron probe
[7] P. F. P.  Poudeu, J.  D’ Angelo, H.  Kong, A.  Downey, J. L.  Short,
microanalyzer (EPMA, JXA-8530F, JEOL). The crystal structure was
determined by the X-ray diffraction pattern acquired by a powder X-ray R.  Pcionek, T. P.  Hogan, C.  Uher, M. G.  Kanatzidis, J. Am. Chem.
diffractometer (XRD, Siemens D-5000) with a Cu Kα source at angles of Soc. 2006, 128, 14347.
20–90o. The high zT x  = 0.01 and x  = 0.02 alloys were further examined [8] Y.  Xiao, D.  Wang, B.  Qin, J.  Wang, G.  Wang, L. D.  Zhao, J. Am.
by a field-emission transmission electron microscopy (FETEM, FEI E.O Chem. Soc. 2018, 140, 13079.
Tecnai F20 G2) equipped with EDS. Prior to the TEM analysis, a focused [9] Y. Z.  Pei, Z. M.  Gibbs, A.  Gloskovskii, B.  Balke, W. G.  Zeier,
ion beam (NX2000, Hitachi, Japan) with the Ga+ as the ion source was G. J. Snyder, Adv. Energy Mater. 2014, 4, 1400486.
used for sample milling. In order to minimize the damage from the [10] Y.  Xiao, H. J.  Wu, W.  Li, M. J.  Yin, Y. L.  Pei, Y.  Zhang, L. W.  Fu,
ion-thinning process, a Pt layer of 3 µm thicknesses was sputtered Y. X. Chen, S. J. Pennycook, L. Huang, J. Q. He, L. D. Zhao, J. Am.
on the targeted region. With 200 KV acceleration electron voltage, Chem. Soc. 2017, 139, 18732.
the TEM analysis including the BF, the DF, the high angular area [11] Q.  Zhang, Q. C.  Song, X. Y.  Wang, J. Y.  Sun, Q.  Zhu, K.  Dahal,
dark-field image under the STEM mode, and the diffraction patterns X.  Lin, F.  Cao, J. W.  Zhou, S.  Chen, G.  Chen, J.  Mao, Z. F.  Ren,
under the selected-area diffractions (SAD) mode was performed. Energy Environ. Sci. 2018, 11, 933.
Thermoelectric Property Measurements: The Bridgman-growth alloy [12] B. Xiang, J. Q. Liu, J. Yan, M. G. Xia, Q. Zhang, L. X. Chen, J. Y. Li,
was sliced into a rod (7.0  mm in diameter and 15mm  in height) and a X. Y. Tan, Q. Y. Yan, Y. C. Wu, J. Mater. Chem. A 2019, 7, 18458.
pellet (7.0  mm in diameter and 1.5  mm in thickness) specimen, which [13] H. J.  Wu, L. D.  Zhao, F. S.  Zheng, D.  Wu, Y. L.  Pei, X.Tong,
are subjected for the power factor S2ρ−1 evaluation and the thermal M. G. Kanatzidis , J. Q. He, Nat. Commun. 2014, 5, 4515.
conductivity κ measurement in the temperature range of 300–673 K,
[14] A.  Bali, R. J.  Chetty, A.  Sharma, G.  Rogl, P.  Heinrich, S.  Suwas,
respectively. The electrical resistivity (ρ) and (S) within 300–673 K
D. K. Misra, P. Rogl, E. Bauer, R. C. Mallik, J. Appl. Phys. 2016, 120,
were measured by a commercial apparatus (ZEM-3, ULVAC, Japan)
175101.
in a helium-filled atmosphere. The κ was obtained from the following
equation: κ  =  D  × CP  × d, whereas the thermal diffusivity (D) was [15] J. Zhang, D. Wu, D. He, D. Feng, M. Yin, X. Qin, J. He, Adv. Mater.
measured by the laser flash method (LFA-467, NETZSCH, Germany) 2017, 29, 1703148.
within 300–673 K. The heat capacity (Cp) was obtained from the [16] J.  Luo, L.  You, J. Y.  Zhang, K.  Guo, H. T.  Zhu, L.  Gu, Z. Z.  Yang,
differential scanning calorimetry (DSC 3500 Sirius, NETZSCH, Germany) X.  Li, J.  Yang, W. Q.  Zhang, ACS Appl. Mater. Interfaces 2017, 9,
with a constant heat rate of 10 K min−1. Mass density (d) was obtained 8729.
via the Archimedes method. [17] D.  Ginting, C. C.  Lin, L.  Rathnam, J. H.  Yun, B. K.  Yu, S. J.  Kim,
J. S. Rhyee, J. Mater. Chem. A 2017, 5, 13535.
[18] Z. Z.  Luo, S.  Cai, S.  Hao, T. P.  Bailey, X.  Su, I.  Spanopoulos,
I. Hadar, G. Tan, Y. Luo, J. Xu, C. Uher, C. Wolverton, V. P. Dravid,
Supporting Information Q. Yan, M. G. Kanatzidis, J. Am. Chem. Soc. 2019, 141, 16169.
Supporting Information is available from the Wiley Online Library or [19] Y. X.  Wu, Z. W.  Chen, P. F.  Nan, F.  Xiong, S. Q.  Lin, X. Y.  Zhang,
from the author. Y. Chen, L. D. Chen, B. H. Ge, Y. Z. Pei, Joule 2019, 3, 1276.

Adv. Funct. Mater. 2020, 2005479 2005479  (10 of 11) © 2020 Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de

[20] P. C. Wei, C. N. Liao, H. J. Wu, D. W. Yang, J. He, G. V. Biesold-McGee, [45] B. Xiang, J. Q. Liu, J. Yan, M. G. Xia, Q. Zhang, L. X. Chen, J. Y. Li,
S.  Liang, W. T.  Yen, X. F.  Tang, J. W.  Yeh, Z. Q.  Lin, J. H.  He, X. Y. Tan, Q. Y. Yan, Y. C. Wu, J. Mater. Chem. A 2019, 7, 18458.
Adv. Mater. 2020, 32, 1906457. [46] S. N.  Girard, T. C.  Chasapis, J.  He, X.  Zhou, E.  Hatzikraniotis,
[21] S.  Ohno, K.  Imasato, S.  Anand, H.  Tamaki, S. D.  Kang, P.  Gorai, C.  Uher, K. M.  Paraskevopoulos, V. P.  Dravid, M. G.  Kanatzidis,
H. K. Sato, E. S. Toberer, T. Kanno, G. J. Snyder, Joule 2018, 2, 141. Energy Environ. Sci. 2012, 5, 8716.
[22] H. J. Wu, W. T. Yen, Acta Mater. 2018, 157, 33. [47] Y. Z. Pei, A. D. LaLonde, H. Wang, G. J. Snyder, Energy Environ. Sci.
[23] G. J.  Tan, F. Y.  Shi, S. Q.  Hao, L. D.  Zhao, H.  Chi, X. M.  Zhang, 2012, 5, 7963.
C. Uher, C. Wolverton, V. P. Dravid, M. G. Kanatzidis, Nat. Commun. [48] H.  Zhang, J.  Luo, H. T.  Zhu, J. K.  Liang, L. M.  Ruan, Q. L.  Liu,
2016, 7, 12167. J. B. Li, G. Y. Liu, Acta Mater. 2012, 60, 7241.
[24] Y. Z.  Pei, J.  Lensch-Falk, E. S.  Toberer, D. L.  Medlin, G. J.  Snyder, [49] A. Bali, H. Wang, G. J. Snyder, R. C. Mallik, J. Appl. Phys. 2014, 116,
Adv. Funct. Mater. 2011, 21, 241. 033707.
[25] Y.  Xiao, H. J.  Wu, J.  Cui, D. Y.  Wang, L. W.  Fu, Y.  Zhang, Y.  Chen, [50] G. J. Tan, C. C. Stoumpos, S. Wang, T. P. Bailey, L. D. Zhao, C. Uher,
J. Q. He, S. J. Pennycook, L. D. Zhao, Energy Environ. Sci. 2018, 11, M. G. Kanatzidis, Adv. Energy Mater. 2017, 7, 1700099.
2486. [51] P. K. Rawat, B. Paul, P. Banerji, ACS Appl. Mater. Interfaces 2014, 6,
[26] L. W. Fu, M. J. Yin, D. Wu, W. Li, D. Feng, L. Huang, J. Q. He, Energy 3995.
Environ. Sci. 2017, 10, 2030. [52] P.  Jood, M.  Ohta, M.  Kunii, X. K.  Hu, H.  Nishiate, A.  Yamamoto,
[27] Z. Feit, D. Eger, A. Zemel, Phys. Rev. B 1985, 31, 3903. M. G. Kanatzidis, J. Mater. Chem. C 2015, 3, 10401.
[28] E. P.  Skipetrov, E. A.  Zvereva, N. N.  Dmitriev, A. V.  Golubev, [53] E. K. Chere, Q. Zhanga, K. McEnaneyb, M. L. Yao, F. Cao, J. Y. Sun,
V. E. Slyn’ko, Mold. J. Phys. Sci. 2006, 5, 32. S. Chen, C. Opeil, G. Chen, Z. F. Ren, Nano Energy 2015, 13, 355.
[29] E. P.  Skipetrov, E. A.  Zvereva, L. A.  Skipetrova, V. V.  Belousov, [54] D.  Ginting, C. C.  Lin, L.  Rathnam, J. H.  Yun, B. K.  Yu, S. J.  Kim,
A. M. Mousalitin, J. Cryst. Growth 2000, 210, 292. J. S. Rhyee, J. Mater. Chem. A 2017, 5, 13535.
[30] K. Lischka, Phys. Status Solidi B 1986, 133, 17. [55] Z. Z. Luo, S. T. Cai, S. Q. Hao, T. P. Bailey, X. L. Su, I. Spanopoulos,
[31] X. L.  Su, S. Q.  Hao, T. P.  Bailey, S.  Wang, I.  Hadar, I.  Hadar, G. J.  Tan, Y. B.  Luo, J. W.  Xu, C.  Uher, C.  Wolverton,
G. J. Tan, T. B. Song, Q. J. Zhang, C. Uher, C. Wolverton, X. F. Tang, V. P.  Dravid, Q. Y.  Yan, M. G.  Kanatzidis, J. Am. Chem. Soc. 2019,
M. G. Kanatzidis, Adv. Energy Mater. 2018, 8, 1800659. 141, 16169.
[32] G. J. Tan, W. G. Zeier, F. Y. Shi, P. L. Wang, G. J. Snyder, V. P. Dravid, [56] Y. W. Xiao, Y. X. Wu, P. F. Nan, H. L. Dong, Z. W. Chen, Z. Q. Chen,
M. G. Kanatzidis, Chem. Mater. 2015, 27, 7801. H. K. Gu, B. H. Ge, W. Li, Y. Z. Pei, Chem 2020, 6, 523.
[33] S. I. Kim, K. H. Lee, H. A. Mun, H. S. Kim, S. W. Hwang, J. W. Roh, [57] N.  Bouad, M. C.  Record, J. C.  Tedenac, R. M.  Marin-Ayral, J. Solid
D. J.  Yang, W. H.  Shin, X. S.  Li, Y. H.  Lee, G. J.  Snyder, S. W.  Kim, State Chem. 2004, 177, 221.
Science 2015, 348, 109. [58] S.  Bajaj, G. S.  Pomrehn, J. W.  Doak, W.  Gierlotka, H. J.  Wu,
[34] G. J.  Tan, X. M.  Zhang, S. Q.  Hao, H.  Chi, T. P.  Bailey, X. L.  Su, S. W. Chen, C. Wolverton, W. A. Goddard, G. J. Snyder, Acta Mater.
C.  Uher, V. P.  Dravid, C.  Wolverton, M. G.  Kanatzidis, ACS Appl. 2015, 92, 72.
Mater. Interfaces 2019, 11, 9197. [59] A. Turchanin, W. Freyland, Phys. Chem. Chem. Phys. 2003, 5, 5285.
[35] R. Hanus, M. T. Agne, A. J. E. Rettie, Z. Chen, G. J. Tan, D. Y. Chung, [60] Z. Chen, D. C. Li, S. P. Deng, Y. Tang, L. Q. Sun, W. T. Liu, L. X. Shen,
M. G. Kanatzidis, Y. Z. Pei, P. W. Voorhees, G. J. Snyder, Adv. Mater. P. Z. Yang, S. K. Deng, Phys. B 2018, 538, 154.
2019, 31, 1900108. [61] G.  Komisarchik, D.  Fuks, Y.  Gelbstein, J. Appl. Phys. 2016, 120,
[36] S. Y. Back, H. Y. Cho, Y. K. Kim, S. Y. Byeon, H. G. Jin, K. Koumoto, 055104.
J. S. Rhyee, AIP Adv. 2018, 8, 115227. [62] J. Male, M. T. Agne, A. Goyal, S. Anand, I. T. Witting, V. Stevanović,
[37] E. S. Božin, C. D. Malliakas, P. Souvatzis, T. Proffen, N. A. Spaldin, G. J. Snyder, Mater. Horiz. 2019, 6, 1444.
M. G. Kanatzidis, S. J. L. Billinge, Science 2010, 330, 1660. [63] E. P.  Skipetrov, E. A.  Zvereva, B. B.  Kovalev, A. M.  Mousalitin, J.
[38] O.  Delaire, J.  Ma, K.  Marty, A. F.  May, M. A.  McGuire, M. H.  Du, Phys. Condens. Matter. 2004, 16, S235.
D. J. Singh, A. Podlesnyak, G. Ehlers, M. D. Lumsden, B. C. Sales, [64] J. Callaway, H. C. Vonbaeyer, Phys. Rev. 1960, 120, 1149.
Nat. Mater. 2011, 10, 614. [65] W.  Kim, J.  Zide, A.  Gossard, D.  Klenov, S.  Stemmer, A.  Shakouri,
[39] S. Y. Lee, K. Esfarjani, T. F. Luo, J. W. Zhou, Z. T. Tian, G. Chen, Nat. A. Majumdar, Phys. Rev. Lett. 2006, 96, 045901.
Commun. 2014, 5, 3525. [66] P. G. Klemens, Proc. Phys. Soc. 1955, A68, 1113.
[40] G. J.  Tan, S. Q.  Hao, R. C.  Hanus, X. M.  Zhang, S. S.  Anand, [67] Z. W.  Chen, Z. Z.  Jian, W.  Li, Y. J.  Chang, B. H.  Ge, R.  Hanus,
T. P. Bailey, A. J. E. Rettie, X. L. Su, C. Uher, V. P. Dravid, G. J. Snyder, J.  Yang, Y.  Chen, M. X.  Huang, G. J.  Snyder, Y. Z.  Pei, Adv. Mater.
C. Wolverton, M. G. Kanatzidis, ACS Energy Lett. 2018, 3, 705. 2017, 29, 1606768.
[41] Z. S. Wang, G. Y. Wang, R. F. Wang, X. Y. Zhou, Z. Y. Chen, C. Yin, [68] A.  Bali, R.  Chetty, A.  Sharma, G.  Rogl, P.  Heinrich, S.  Suwas,
M. J. Tang, Q. Hu, J. Tang, R. Ang, ACS Appl. Mater. Interfaces 2018, D. K. Misra, P. Rogl, E. Bauer, R. C. Mallik, J. Appl. Phys. 2016, 120,
10, 22401. 175101.
[42] J. R.  Sootsman, H.  Kong, C.  Uher, J. J.  D’Angelo, C. I.  Wu, [69] C. S.  Oh, D. N.  Lee, CALPHAD: Comput. Coupling Phase Diagrams
T. P.  Hogan, T.  Caillat, M. G.  Kanatzidis, Angew. Chem., Int. Ed. Thermochem. 1992, 16, 317.
2008, 47, 8618. [70] Y. I. Ravich, B. A. Efimova, I. A. Smirnov, Semiconducting Lead Chal-
[43] A. D.  LaLonde, Y. Z.  Pei, G. J.  Snyder, Energy Environ. Sci. 2011, 4, cogenides, Plenum Press, New York 1970.
2090. [71] G. T.  Alekseeva, B.  Efimova, L. M.  Ostrovsk, O. S.  Serebrya,
[44] Y. Z. Pei, A. F. May, G. J. Snyder, Adv. Energy Mater. 2011, 1, 291. M. Tsypin, Sov. Phys. Semicond. 1971, 4, 1122.

Adv. Funct. Mater. 2020, 2005479 2005479  (11 of 11) © 2020 Wiley-VCH GmbH

View publication stats

You might also like