Engineering the thermoelectric transport in half-Heusler materials through

a bottom-up nanostructure synthesis

Huaizhou Zhao1*, Binglei Cao1, 2, Shanming Li1, Ning Liu1, Jiawen Shen3, Shan Li1, Jikang

Jian2, Lin Gu1, Yanzhong Pei3, G. Jeffrey Snyder4, Zhifeng Ren5 and Xiaolong Chen1

1
Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of

Sciences, Beijing 100190, China

2
Guangdong University of Technology, Guangzhou 510006, China

3
School of Materials Science and Engineering, Tongji University, Shanghai 201804, China

4
Department of Materials Science and Engineering, Northwestern University, Evanston, Illinois 60208,

USA.

5
Department of Physics and TcSUH, University of Houston, Houston, TX 77204, USA

Corresponding author: Hzhao@iphy.ac.cn

Abstract: Half-Heusler (HH) alloys Zr(Ti,Hf)NiSn, Zr(Ti,Hf)CoSb, and Nb(V)FeSb are

among the best promising thermoelectric (TE) materials applicable for the middle-to-high
temperature power generation. Despite of the large thermoelectric power factor and decent
figure-of-merit ZT (~1) in HHs, their broad applications and enhancement on TE performance
are limited by the high intrinsic lattice thermal conductivity (κL) due to insufficiencies of

This is the author manuscript accepted for publication and has undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1002/aenm.201700446.

This article is protected by copyright. All rights reserved.

phonon scattering mechanisms, and the fewer powerful strategies associated with the
micro-structural engineering for HH materials. We herein report a bottom-up nanostructure
synthesis approach for these HH materials based on the displacement reaction between metal
chlorides/bromides and Magnesium (or Lithium), followed by vacuum-assisted spark plasma
sintering process. The samples are featured with dense dislocation arrays at the grain
boundaries, leading to a minimum κL of ~1 W m-1 K-1 at 900 K and one of the highest ZT
(~1) and predicted η (~11 %) for n-type Hf0.25Zr0.75NiSn0.97Sb0.03. Further manipulation on the
dislocation defects at the grain boundaries of p-type Nb0.8Ti0.2FeSb leads to enhanced
maximum power factor of 47×10-4 W m-1 K-2 and the predicted η of ~7.5 %. Moreover,
vanadium substitution in FeNb0.56V0.24Ti0.2Sb significantly promotes the η to ~ 11 %. This
strategy can be extended to some other notable TE materials and a broad range of advanced
alloys and compounds for improved properties.

1. Introduction

A variety of structural defects from atomic to micron scales in especially nano-structured

alloys or compounds such as high strength lightweight titanium alloy Ti185,1 aluminum or
magnesium based alloys NH-7075 and NH-5083,2 and particularly thermoelectric materials,3
play vital roles in improving the performance. Thermoelectric (TE) materials can realize
direct conversions between heat and electricity, and the maximum energy conversion
efficiency (ηmax) is primarily governed by the thermoelectric figure-of-merit ZT (ZT =
(S2σ/κ)T, where S, σ, κ, and T are the Seebeck coefficient, electrical conductivity (inverse of
electrical resistivity σ = 1/ρ, thermal conductivity, and absolute temperature, respectively).

Recent studies have witnessed the fast growth of TE materials. Among the notable TE
material systems, such as Bi2Te3,4, 5 PbTe,6 AgSbTe2,7 Zn4Sb3,8 Zintl phase9 and SiGe,10,11 as

This article is protected by copyright. All rights reserved.

well as the recently emerged ionic conductor Cu2Se,12 filled skutterudites,13 α-MgAgSb,14
SnSe,15 BiCuSeO,16 Mg2Si1-xSnx 17, 18 etc., half-Heuslers (HHs) materials are the most
promising ones in middle-to-high temperature (600-900 K) for power generation.
Semiconducting HHs have the appropriate band gap size for applications from middle to high
temperature19 (for instance 0.51 eV for ZrNiSn,20 and 0.54 eV for NbFeSb 21. In the context

of thermoelectric quality factor ( ),22 HH materials exhibit superior

electrical transport properties ( For instance, the maximum power factors for n type
Hf0.5Zr0.5NiSn0.99Sb0.01 and p-type FeNb0.95Ti0.05Sb reach 50×10-4 and 106×10-4 W m-1 K-2,23,
24
respectively, which outperforms the PbTe based materials6 and Na doped SnSe.25
Moreover, HH materials possess high thermal stability and superior mechanical strength over
most other TE materials.26, 27 Therefore, HH materials have become the focus of increasing
interest in thermoelectric research.

HHs have a potential to reach higher ZTs due to the very high power factor if their intrinsic
high thermal conductivity (~10 W m-1 K-1 at 300 K)28 dominated by phonons can be
significantly reduced. Phonon scattering can be estimated by the Callaway model based on
relaxation time approximation,29 the existing phonon scattering mechanisms in the best HH
materials is comprised of Umklapp process, grain boundary, point defect and
electron-phonon scatterings. However, despite of the reduction in κL obtained by all above
mechanisms,30, 31 there is still a large room for further κL reduction given the amorphous limit
of κmin is ~ 1 W m-1 K-1 for typical HH TE materials calculated by Cahill’s formulation.32, 33
To achieve this goal, it seems that new phonon scattering mechanism is needed for HH
materials. It is important to note that the existing scattering mechanisms in HH materials
target mostly at the high frequency phonons through Umklapp and electron-phonon
scatterings, the substantial portion of phonons in the low to middle frequency (≤ 30THz)
range have been overlooked. Therefore, a new mechanism, such as dislocation arrays at the
grain boundaries,4 might be efficient in scattering low to middle frequency phonons.

This article is protected by copyright. All rights reserved.

 Fabricating dense dislocation arrays at the grain boundaries for HH materials as well as in
other alloys is a great challenge due to the co-existence of multiple thermodynamic phases
constituted by more than 3 elements during the non-equilibrium dynamic synthesis process,
which is believed to be far more complicated than that in Bi0.5Sb1.5Te3 system4 and in the
Pb1-xSb2x/3Se solid solutions.34 In this paper, we proposed a bottom-up nanostructure
synthesis approach for HH materials based on the displacement reaction between metal
chlorides/bromides and magnesium (or lithium), followed by vacuum-assisted spark plasma
sintering process. The new HH samples are featured with dense dislocation arrays at the grain
boundaries, and these dislocation defects can be manipulated to adjust the thermal and
electrical transports through the sintering process during SPS, leading to significantly
lowered κL and one of the highest ZT (~1) for n type Hf0.25Zr0.75NiSn0.97Sb0.03, as well as
enhanced thermoelectric performance for p-type Nb0.8Ti0.2 FeSb and FeNb0.56V0.24Ti0.2Sb.
This strategy can be extended to other good TE materials for improved properties.

2. Results and discussion

Schematic diagram of the synthesis of a series of bulk HH samples is illustrated in figure 1.

As shown in the figure, based on a simple displacement reaction between metal
chlorides/bromides and the metal magnesium (or lithium), stoichiometric ratios of metal
chloride or bromide powder and Mg/Li metal granules are put into the stainless steel ball
milling jar, followed by 2-5 hours high-energy ball milling process. During the ball milling
process, a mixture of multiple metal nanoparticals and MgX2/LiX were formed, as can be
identified by the X-ray diffraction (XRD) results shown in figure S1(a). However,
removing the MgX2/LiX byproducts from the metal nanoparticals may be difficult, which is
why this method was not heavily reported. Interestingly, we discovered that the MgX2/LiX
byproducts can be expelled by vacuum-assisted spark plasma sintering (SPS) process, which

This article is protected by copyright. All rights reserved.

can be ascribed to the lower melting point and high vapor pressures of MgX2/LiX. Indeed
uniformly distributed MgX2/LiX powder can be observed all around the surface of the
graphite dies and the vacuum chambers after the spark plasma sintering process. The obtained
HH bulk samples (as shown in the photo images in lower left of Fig. 1) exhibit the same
appearance and phase purities, and enhanced thermoelectric properties (particularly for
p-type) as those made by the conventional metal alloying method, which will be discussed in
the following context.

Figure 1. Schematic illustration of the synthesis process of half-Heusler materials based on displacement

chemical reactions and spark plasma sintering process.

Seven typical HH thermoelectric materials were synthesized by the new approach, and
for each composition, three samples have been made with the sintering time of 5 minutes, 20
minutes, and 4 hours, respectively. The obtained bulk samples were firstly characterized by
powder XRD, typically for the samples being sintered for 20 minutes, as shown in Fig. 2a.

This article is protected by copyright. All rights reserved.

The XRD for 5 minutes sintered samples were shown in figure S1(b). All patterns
corresponding to six compositions in Fig. 2a and figure S1(b) can be indexed to the MgAsAg
prototype structure (space group, ̅ ) with the representative formula of XYZ.35 The
structures is composed of three interpenetrating fcc sublattices, each of which is occupied by
the X, Y and Z atoms, as shown in the inset of Fig. 2a. With respect to ZrCoSb and

Zr0.44Hf0.44Ti0.12CoSb0.85Sn0.15, two minor peaks at 2 ~ 28.6o and 31.5o can be observed, as

can be seen from the enlarged XRD pattern between 25-35o in figure S1(c), which is believed
to originate from the Zr/Hf oxides during the preparation. In addition, the impurities in above
two samples cannot be easily removed, which would induce poor electrical transport
properties for these two compositions. Meanwhile, no visible peaks of the impure phases can
be observed for other four compositions. Thus in this work we focused on p-type
Nb0.8Ti0.2 FeSb (NTFC-20min) and n-type Hf0.25Zr0.75NiSn0.97Sb0.03 (HZNSS-20min) samples
being sintered for 20 minutes during SPS to elucidate the underneath relations between their
rich varieties of structural defects and the TE properties, particularly on the phonon transport.

Meanwhile, in order to check the loss of materials due to liquid expelling in the SPS
process, we use the inductively coupled plasma mass spectrometry (ICP-MS) to confirm the
accurate elemental compositions. The results indicate that the actual compositions of the
(NTFC-20min) and (HZNSS-20min) bulk samples are consistent with the stoichiometric ratio
of the starting materials within the standard deviation of 1-3% (Table S1). Furthermore,
energy dispersive X-ray spectroscopy (EDX) mapping analysis indicate that the
compositional elements distributed uniformly, as shown in Fig. 2b and 2c. In Fig. 2b, Fig. 2c,
and Fig. S2, from the scanning electron microscopy (SEM) images of the cross sections, we
can see that the grains are relatively compact, while the grain size are around 150-400 nm for
NTFC-20min, and 500 nm in average for HZNSS-20min. These grain sizes and morphologies
are similar to those conventionally sintered samples reported in the literature,30 the details of

This article is protected by copyright. All rights reserved.

the microstructure in the lattice and atomic level are investigated using transmission electron
microscopy (TEM) technique.

Figure 2. (a) Room temperature powder X-ray diffraction patterns for seven half-Heusler samples. The inset

shows the crystal structure model of half-Heusler and the corresponding space group (S.G.); (b) SEM images and

EDX mappings of the compositional element distribution for Nb0.8Ti0.2FeSb (NTFC-20min); and (c)

Hf0.25Zr0.75NiSn0.97Sb0.03 (HZNSS-20min).

The morphology of the grain boundary structures in NTFC-20min and HZNSS-20min is

displayed in the TEM images as shown in Figs. 3a-3l. Generally, in comparison to the
nano-structured HH TE materials synthesized by conventional process,23,28 two kinds of

This article is protected by copyright. All rights reserved.

dislocations with significantly high densities were found at the grain boundaries of these two
samples. Similar but not identical to the grain boundary structures as observed in
Bi0.5Sb1.5Te3 synthesized by the liquid-phase compacting process4, TEM images in Figs. 3b,
h, and c, e, j, k reveal the existence of twist-type grain boundaries and Mioré patterns (up to
15 nm wide), respectively for both samples. Since these two structures are indication of
highly crystalline with clean grain boundaries for both samples, the presence of dislocation
arrays are required to lower the energy at the grain boundaries for these samples.36 These
dislocation arrays are described in details individually in the following.

Fig. 3a and 3b display the low-magnification TEM image of NTFC-20min and the
enlarged view of boxed region in (a), respectively. Dislocation arrays with the periodic misfit
spacing of 2.5 nm were observed in high resolution TEM image (Fig. 3c), which can be
viewed as a result of the compensation effect for the misorientation between two adjacent
grains on the (010) atomic planes along the dislocation arrays marked in red in Fig. 3d (as the
inverse fast Fourier-transformed IFFT image for 3c). In addition, the Burgers vector of each
dislocation in Fig. 3d is identified as BD = <010>. Another kind of dislocation array is shown
in the transitional Mioré patterns in HRTEM image of Fig. 3e, which has the periodic misfit
spacing of 5 nm. The IFFT and FFT images of Fig.3e are shown in Fig. 3f, and the upper
inset of Fig. 3f, respectively. The dislocation arrays in Fig. 3f are shown in red circles, and
the boxed region shows the enlarged view of the dislocation. The zone axis is identified as
[100] for both adjacent grains in Fig. 3f and the atomic planes are (010). The dislocation
arrays featured by transitional Mioré patterns are observed as the major dislocation at the
grain boundaries for NTFC-20min, which may reflect the higher crystal symmetry of HH
materials ( ̅ ) than that of the Bi0.5Sb1.5Te3 structure ( ̅ ). As expected, the dense
dislocation arrays are also observed in the HZNSS-20min sample, as shown in Fig. 3h, 3j,
and 3k, while 3h and 3j show the enlarged HRTEM images for the boxed areas of Fig. 3g and
3i, respectively. Transitional Mioré patterns are observed in Fig. 3j and the enlarged image in

This article is protected by copyright. All rights reserved.

Fig. 3k. The IFFT and the FFT images of Fig. 3k are shown in Fig. 3l and the left inset in 3l,
respectively, while the right inset in Fig.3l is the enlarged IFFT image for the boxed
dislocation area in Fig. 3l. The zone axes of atomic planes in Fig. 3l is identified as [211], and
the Burgers vector of each dislocation in Fig. 3l is < ̅ >. Overall, the dislocation arrays
observed for NTFC-20min and HZNSS-20min samples have close spacing of 2.5~6 nm

between the cores, which corresponds to a dislocation density of ~11011 cm-2.

To the best of our knowledge, bulk HH TE materials exhibiting dense dislocation arrays
at the grain boundaries have not been reported. 23, 30 The formation mechanism of the
dislocations in NTFC-20min, HZNSS-20min, and others samples should be similar to those
observed in materials possessing enhanced mechanical strength,37 magnetism properties,38
and TE performance.39 During the SPS process, the MgX2/LiX byproducts produced in the
high energy ball milling process can be liquefied and expelled at elevated temperatures above
873 K. The existence of liquid MgX2/LiX phase facilitates diffusion at the grain boundaries
of HHs, which upon heating and compacting, leads to the appearance of dense dislocation
arrays at grain boundaries, resulting in a semi-coherent interface structure for the HH
material.

This article is protected by copyright. All rights reserved.

Figure 3. TEM images showing dislocations along the grain boundaries. (a) Low-magnification TEM image of

NTFC-20min; (b) enlarged view of the boxed region in (a); (c) high resolution TEM image of a grain boundary;

(d) inverse FFT image of (c), which shows the dislocation arrays as indicated by the red symbols, inset is the

enlarged view of a dislocation; (e) high resolution TEM image of another grain boundary; (f) inverse FFT image

of (e), which shows some dislocations encircled in red, insets are the enlarged view of a dislocation (lower left)

and FFT image (upper) of the right grain; (g)-(l) TEM images featured with the dense dislocation arrays at the

grain boundaries of HZNSS-20min.

The temperature dependence of the total thermal conductivity tot for six samples of
Nb0.8Ti0.2 FeSb and Hf0.25Zr0.75Ni1.05Sn0.97Sb0.03, as shown in Fig. 4a, are calculated from tot =
DCpd, where D is the thermal diffusivity coefficient (Fig. S3a), Cp the specific heat capacity

This article is protected by copyright. All rights reserved.

(Fig. S3b), and d the density. Fig. 4b shows the temperature dependent electrical thermal

conductivity e for all samples, which was determined via the Wiedemann-Franz relation e =
LσT, where L, σ, and T is the Lorenz number (Fig. S4), temperature dependent electrical
conductivity, and the absolute temperature, respectively.40 The temperature dependent lattice
thermal conductivity L and bipolar thermal conductivity bip for the six Nb0.8Ti0.2 FeSb and
Hf0.25Zr0.75Ni1.05Sn0.97Sb0.03 samples were calculated by subtraction e from tot. The
contribution of bip arises slightly at high temperature for Nb0.8Ti0.2 FeSb. As shown in Fig. 4c
and 4d, the temperature dependence of L for NTFC-5min, NTFC-20min, HZNSS-5min, and
HZNSS-20min are significantly reduced，particularly by 40% and 44% at 300 K for the 5
minutes sintered samples in comparison to the literature values,21, 23 respectively. For the
HZNSS-5min sample, the L approaches ~1 W m-1 K-1 at 900 K, which is the minimum
lattice thermal for HH materials. Whereas, the L of 4 hours sintered samples increase
substantially and become comparable to the literatures,21, 23 which can be ascribed to the
growth of grain sizes and annihilation of dislocations, as shown in TEM images in Fig. S5.
We noticed that from 5min to 20min and 4hours, the bipolar effect in HZNSS samples
became slightly obvious, which is probably due to the slight increase of the concentrations of
the minor carrier originated from the intrinsic antisite defects between Zr and Sn sites, and in

turn the band narrowing effect in HZNSS materials.41 The reductions on L reveal that the
unique dislocation arrays at the grain boundaries might have profound impact on the phonon
scattering mechanism for these samples. In order to understand the details of the phonon
scattering mechanisms, the lattice thermal conductivity L was analyzed using the expression
based on Callaway model as following:29

∫ (1)

where , D, and  is the phonon velocity, the Debye temperature, and the phonon frequency,
respectively, x is defined as x = ħ/kBT, total (x) is the total effective relaxation time, which
can be determined from the individual relaxation time originated from different scattering

This article is protected by copyright. All rights reserved.

mechanisms. Based on the Matthiessen’s rule, the total phonon relaxation time (τtot) is a
reciprocal sum of the relaxation times of the following pertinent scattering mechanisms,
which traditionally comprise phonon-phonon Umklapp scattering (U), electron-phonon
scattering (EP) , point defect scattering (PD), and boundary scattering (B). 39

However, in this work, due to the dense dislocations at the grain boundaries in the
Nb0.8Ti0.2 FeSb and Hf0.25Zr0.75Ni1.05Sn0.97Sb0.03 bulk samples, additional considerations have

to be put forth regarding the phonon scattering from the dislocation cores (DC) and the
dislocation strain (DS). The details of the calculation on the total phonon relaxation time
(τtot), as well as the individual U, EP, PD, B, DC, and DS can be found in the supporting
information.

Based on the calculations, phonon relaxation times from different scattering mechanisms
are calculated and shown in Fig. 4e and 4f for NTFC-20min and HZNSS-20min samples,
respectively. The results show that dislocation strain is the dominant mechanism for low and
middle frequency phonon scattering, and meanwhile the point defect and electron-phonon
scatterings are the dominant mechanisms for high frequency phonons in both materials. The

comparison between experimental L for six Nb0.8 Ti0.2 FeSb and Hf0.25Zr0.75NiSn0.97Sb0.03
samples and the fitted L based on the parameters from NTFC-20min and HZNSS-20min are
shown in Fig. 4g and 4h, respectively. For NTFC-5min, NTFC-20min, HZNSS-5min and
HZNSS-20min, the dislocation strain scattering is believed to contribute significantly on the
reduction of lattice thermal conductivity below 800 K, and among them the experimental and
fitting data of the 20 minutes sintering samples match well with each other. For
Nb0.8Ti0.2 FeSb samples, it is noted that the experimental thermal conductivity start to arise
slightly above 800 K, which indicates that the minority carriers could have been excited and
bipolar thermal conduction emerges. For sample NTFC-4h and HZNSS-4h, significant grain
growth accompanied by annihilation of dislocation defects at the grain boundaries can be

This article is protected by copyright. All rights reserved.

observed, as shown in the TEM images in Fig. S5, which would account for the overall

higher L with respect to the other short time sintered samples.

(a) 8 NTFS-5min (b) 3

NTFS-20min
NTFS-4h

κe(Wm-1K-1)
HZNSS-5min
κtot(Wm-1K-1)

6 HZNSS-20min 2
HZNSS-4h

4 1

2 0
400 600 800 1000 400 600 800 1000
T(K) T(K)
(c) 5 (d) 4
NTFS-5min HZNSS-5min
NTFS-20min
κL+κbip(Wm-1K-1)

κL+κbip(Wm-1K-1)

4 HZNSS-20min
NTFS-4h HZNSS-4h
3

2
2

1 1
400 600 800 1000 400 600 800 1000
T(K) T(K)
(e) point defect Umklapp
(f) point defect Umklapp
101
Phonon relaxation time (ns)

electron-phonon boundary electron-phonon boundary

Phonon relaxation time (ns)

101 dislocation cores dislocation strain

dislocation cores dislocation strain
total Effective total Effective

10-1 10-1

10-3 10-3

Nb0.8Ti0.2FeSb 400 K Hf0.25Zr0.75NiSn0.97Sb0.03 400 K

10-5 10-5
0 10 20 30 40 50 ωD 60 0 10 20 30 40 ωD 50
Frequency (THz) Frequency (THz)
(g) 6 U+PD+B
(h) 5
U+PD+B
U+PD+B+EP U+PD+B+EP
5 U+PD+EP+DC+DS U+PD+EP+DC+DS
4
κL(Wm-1K-1)
κL(Wm-1K-1)

NTFS-5min HZNSS-5min
NTFS-20min HZNSS-20min
4 NTFS-4h HZNSS-4h
3

3
2

2
1
400 600 800 1000 400 600 800 1000
T(K) T(K)

This article is protected by copyright. All rights reserved.

Figure 4. Thermal conductivity dependence of temperature. (a) Total thermal conductivity tot of all samples; (b)

electron thermal conductivity e of all samples; (c) lattice thermal conductivity and bipolar thermal conductivity

(L+bip) of NTFC-5min, NTFC-20min, and NTFC-4h samples; (d) (L+bip) of HZNSS-5min, HZNSS-20min,

and HZNSS-4h samples; (e) calculated phonon relaxation time  versus frequency  for NTFC-20min sample and

for (f) HZNSS-20min sample; (g) comparison between experimental and the fitted L for three Nb0.8Ti0.2FeSb

samples, and for (h) three Hf0.25Zr0.75NiSn0.97Sb0.03 samples.

Figures 5a and 5b display the temperature dependence of electrical resistivity  and

Seebeck coefficient S for all the Nb0.8Ti0.2FeSb and Hf0.25Zr0.75NiSn0.97Sb0.03 samples,

respectively. For both compositions, the  and S increase with temperature and saturate at
around 700 K for Hf0.25Zr0.75NiSn0.97Sb0.03, which is consistent with the literature results for
the similar compositions and also indicates the typical electrical transport behavior for
degenerated HH semiconductors.21, 23 It is noted here that with respect to the reported data for

HfxZr1-xNiSn0.99Sb0.0123 and HfxZr1-xNiSn0.985Sb0.01541 compositions, the relatively higher 

and lower S of HZNSS-5min and HZNSS-20min lead to lower power factors in the whole
temperature range with a maximum value of 32×10-4 W m-1 K-2 at 673 K, while the power
factors were increased by sintering process with a maximum of 38×10-4 W m-1 K-2 for
HZNSS-4h, as shown in Fig. 5g. The temperature dependent electron carrier concentration,
obtained from Hall measurement as shown in Fig. 5c, reaches to 3.11020 cm-3 for
HZNSS-20min. The nearly temperature-independent carrier concentration might be slightly
higher in comparison to the optimized literature values,31 and it probably is the result of over
doping based on the calculation for nH dependence of power factor (PF) using the SPB model
(SI) as shown in Fig. 5h. Lowering the carrier concentration would favor higher Hall mobility
and Seebeck coefficient. For instance, reducing the Sb content from 0.03 to 0.01 leads to S
improvement from -130 to -160 µV/K, but no net gains on the PF. The temperature
dependence of Hall mobility for HZNSS-20min, as shown in Fig. 5d, is around H ~29 cm2

This article is protected by copyright. All rights reserved.

V-1 s-1 at near room temperature, which is comparable to the results for

HfxZr1-xNiSn0.985Sb0.015 compositions,41 whereas the carrier concentration is as high as 41020

cm-3 for the later. Thus, the nominal doping level of Sb~0.03 in HZNSS-20min is most
probably higher than the actual level in the sample, which should be close to Sb~0.015 based
on the relation between carrier concentration and the Sb content in literature.41 This argument
can also be supported by the fittings of μH as a function of nH for HZNSS-20min under the
assumption of homogeneous alloying, as shown in Fig. 5f. It is observed that the

experimental H ~ 29 cm2 V-1 s-1 is higher than the fitting value of 19 cm2 V-1 s-1 in the curve
under the same carrier concentration of 3.11020 cm-3, which obviously means the Sb content
is less than 0.03. This is because the difference between the experimental and fitting values
was supposed to be induced by the Sb content difference between actual and nominal values.
Homogeneous doping or alloying is particularly important for an alloying scattering
dominated carrier transport system, such as in ZrNiSn based materials since the relatively low
deformation potential and alloy scattering potential favor high carrier mobility. 42 The
relationship of μH ~ T-0.5 for HZNSS-20min shown in Fig. 5d indicates alloying scattering
mechanism for the system and this is further confirmed by the modeling in Fig. 5f. The
details of the modeling can be found in the supporting information.

Here the question still arises why the PF in HZNSS-5min and HZNSS-20min is lower than
those samples with similar carrier concentration in the literature? Despite of the scattering
induced by the minor inhomogeneity of Sb doping atoms in the sample, more contributions
may come from the crystal defects at the nanoscale level in the bulk sample, which would
scatter phonon and electron at the same time, given that the mean free paths for electron and
phonon are both revealed to be around 5 nm above 200 K for the ZrNiSn based materials. 41
The aforementioned dense dislocation strains and dislocation cores in the HZNSS-5min and
HZNSS-20min bulk samples might be the key reason. As can be seen from HZNSS-4h, with
the grain growth and annihilation of strain defects at the grain boundaries compared to two

This article is protected by copyright. All rights reserved.

short time sintered samples, which is detected by the TEM techniques (Fig. S5), the electrical
transport performance was enhanced as proved by the carrier concentration and increased
carrier mobility in HZNSS-4h in Fig. S6a and 6b, leading to comparable power factors to the
literature value.23

On the contrary to ZrNiSn based materials, the p-type Nb0.8Ti0.2 FeSb system is
characterized by acoustic scattering dominated carrier transport properties, which is
consistent with our analysis in the following context. In comparison with the literature data
for the same composition,21 the Nb0.8Ti0.2FeSb bulk samples in our synthesis exhibit some
interesting features in terms of its electrical transport properties. First of all, among three
samples the temperature dependent electrical resistivity ρ (Fig. 5a) decrease with the
increasing of sintering time from 5 minutes to 4 hours, and their differences become
significant at high temperatures, eventually NTFC-4h exhibits the lowest ρ and it reaches to
9.23 μΩ m at 925 K, which is higher than the literature’s value of 7.15 μΩ m.21 Second, as for
the S, there is not as much difference as that in the ρ for three samples except in high
temperature range, where the S of NTFC-4h reaches 187 μV/K at 925 K, nearly 14 % higher
than those of NTFC-5min and NTFC-20min. Moreover, in comparison to the literatures, the S
of three samples are higher at 300 K (80 vs. 70 μV/K), and gradually coalesce with the
literature values until 925 K (184 μV/K from Ref.18). This leads to 56% increase on the PF
at 300 K for NTFC-4h sample with respect to literature, 21 reaches 39×10-4 W m-2 K-1 in our
case, as shown in Fig. 4g. Meanwhile, the PF of our sample gradually increase with
temperature until reaches 47 ×10-4 W m-2 K-1 at 535 K, than decreases slowly to 37 ×10-4 W
m-2 K-1 at 925 K, in comparison to 45 ×10-4 W m-2 K-1 at 925 K from literature. 21 The
temperature dependence of carrier concentration and Hall mobility of NTFC-20min are
shown in Fig. 5c and d, in which the carrier concentration of 201020 cm-3 is lower than the
literature by 20%, 21 probably reflecting inhomogenous Ti doping in this sample. On the other
hand, the mobility start from 17 cm2 V-1 s-1 at 300 K, then degrade in a faster way to 5 cm2 V-1

This article is protected by copyright. All rights reserved.

s-1 at 800 K compared to literature. 21 The relatively lower Hall mobility in NTFC-20min at
high temperatures is believed to be responsible for the low PF in the same high temperature
range. However, the carrier transport performance of NTFC-20min can be improved, as
shown in Fig. S6a and 6b, higher carrier mobility was found in the NTFC-4h samples. To
better understand the charge carrier transport mechanism, we found the H exhibits a μH ~ T-1
relationship for NTFC-20min, as shown in Fig. 5d, which is the feature of acoustic scattering
mechanism. Further analysis in Fig. 5e indicates that the acoustic scattering play dominating
role in the charge transport for NTFC-20min, which is well consistent with the literature. 21
Here the details of modeling and calculation of the carrier transport mechanism can be found
in supporting information. In connection with the new phonon scattering centers induced by
the dislocation cores and strains at the boundaries of crystal grains in NTFC-20min bulk
sample, it shows no significant influence on the charge transport, at least in the low
temperature range. However, as shown in Fig. 3b, bipolar effect arises at high temperatures
for NTFC-5min and NTFC-20min, which would be responsible for their relatively lower S. In
addition, this is consistent with the slight inhomogeneity of Ti doping in both samples.
Furthermore, the theoretical PF in dependence of carrier concentration calculated based on
the SPB model is shown in Fig. 5h, which clearly shows that the improvement in the
experimental PF through increasing the sintering time, as observed in NTFC-4h.

This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.
Figure 5. Temperature dependent thermoelectric properties. (a) electrical resistivity (T) of all samples; (b)

Seebeck coefficient S(T) of all samples; (c) carrier concentration nH, and (d) Hall mobility H for NTFC-20min

and HZNSS-20min; (e) and (f) are the room temperature Hall mobility as the function of carrier concentration for

NTFC-20min and HZNSS-20min, respectively. The curves were calculated by using m* = 6.9 me for NTFC-20min

and 3.05 me for HZNSS-20min; (g) power factor PF for all samples of Nb0.8Ti0.2FeSb and

Hf0.25Zr0.75NiSn0.97Sb0.03; (h) modulated PF as the function of carrier concentration for NTFC-20min and

HZNSS-20min at 300 K with experimental data points for NTFC-20min, NTFC-4h, HZNSS-20min and

HZNSS-4h samples for comparison.

The temperature dependence of ZTs for the six Nb0.8Ti0.2 FeSb and
Hf0.25Zr0.75NiSn0.97Sb0.03 bulk samples are calculated based on the measured ρ, S, and κ
values, and plotted in Fig. 6a, 6b. The TE properties of Hf0.44Zr0.44Ti0.12CoSb0.85Sn0.15 are also
put in Fig. S6. Overall, among the Nb0.8Ti0.2 FeSb samples, the ZTs increase with the sintering
time, particularly at high temperatures, where the maximum ZT of NTFC-4h reaches 0.73 at
925 K. The ZTs between room temperature and 700 K for all Nb0.8Ti0.2FeSb samples are
about 30% higher than the literature value. 21 Meanwhile, above 750 K the literature’s ZT
gradually outperforms our samples.21 These results are fully consistent with the characters of
the interplays between temperature dependence of thermal conductivity and the power factors
for the Nb0.8Ti0.2FeSb samples. Namely the enhanced PF and reduced thermal conductivity at
lower temperatures increase the ZTs over the references,21 while the lower PF at higher
temperatures overturns the reduction in thermal conductivity, leading to lower ZTs,
particularly in NTFC-5min, compared to the literature values.21 Inset in Fig. 6b shows the
carrier concentration dependence of calculated ZT based on the two-band model (conduction
band and valence band) for NTFC-20min at 300, 600, 800 K. The measured ZTs of
NTFC-20min and NTFC-4h at three different temperatures locate closely to the maximum
positions in the curves, meaning that the carrier concentration and ZTs could have been

This article is protected by copyright. All rights reserved.

optimized. On the other hand, for three Hf0.25Zr0.75NiSn0.97Sb0.03 materials, their temperature
dependent ZTs increase continuously and show no significant differences until 820 K, at
which HZNSS-5min continues to be comparable to the literature values for the
Hf0.25Zr0.75NiSn0.99Sb0.01 composition until 925 K due to its lowest thermal conductivity
among three samples.23 Unlike the p-type Nb0.8Ti0.2FeSb materials, the interplays between
suppressed lattice thermal conductivity and reduced electrical transport properties in
HZNSS-5min and HZNSS-20min samples in comparison to the HZNSS-4h are more like
dominated by the lattice thermal conductivity part in the high temperature range, which
favors higher ZT peaks. Inset in Fig. 6a shows the carrier concentration dependence of the
calculated ZT based on the two-band model for HZNSS-20min at 300, 600, 800 K. The
measured ZTs of HZNSS-20min and HZNSS-4h at three corresponding temperatures were
compared with the modeling, which indicate that the carrier concentration and ZTs have been
optimized. The details of modeling and calculation of the ZT can be found in supporting
information.

This article is protected by copyright. All rights reserved.

Figure 6. Temperature dependent thermoelectrical properties. Figure of merit ZT for (a) Nb0.8Ti0.2FeSb and (b)

Hf0.25Zr0.75NiSn0.97Sb0.03 samples, the calculated respective ZTs as the function of carrier concentration are shown

in the insets. The solid lines were calculated based on the two-band model. The calculated cumulative temperature

(T) dependent TE properties for all Hf0.25Zr0.75NiSn0.97Sb0.03 and Nb0.8Ti0.2FeSb samples are plotted and

compared with the literatures values for the same compositions, 21, 23 the cold side temperature is fixed as Tc = 323

K for all analysis. (c) T dependence of (ZT)eng, (d) T dependence of , (e) T dependence of (PF)eng and (f) T

dependence of power density Pd.

This article is protected by copyright. All rights reserved.

 Furthermore, it was confirmed that the engineering (ZT)eng and (PF)eng of materials can
predicted more accurately the practical performance of TE materials at any given temperature
difference than the average (ZT)avg based on the constant property model.43, 44 In terms of TE
power generations, the output power and power density are equally important as that of ηmax
and (ZT)eng under the assumption of unlimited or free heat sources. Thus the (ZT)eng, (PF)eng,

Pd, and  are calculated for both Nb0.8Ti0.2 FeSb and Hf0.25Zr0.75NiSn0.97Sb0.03 materials and
compared with the closest composition in the literatures.21, 23

The (ZT)eng, , (PF)eng, and Pd were calculated based on the experimental results and
compared with the data calculated from the literatures’ values.21, 23 It is shown in Fig. 6c and
6d that the temperature difference dependence of (ZT)eng and  of three Nb0.8Ti0.2FeSb
samples are substantially higher than that of the reference data, with a maximum  of 7.8 %
for NTFC-4h at T~600 K. Regarding the temperature difference dependence of (PF)eng and
Pd of the Nb0.8Ti0.2FeSb samples as shown in Fig. 6e and 6f, the NTFC-4h outperforms
NTFC-5min, NTFC-20min and the reference data in the whole range, owing to its much
enhanced PF at low temperatures. Meanwhile, for all Hf0.25Zr0.75NiSn0.97Sb0.03 samples, the
temperature difference dependence of (ZT)eng and  as shown in Fig. 6c and 6d are
competitive to reference data with a maximum  of 10.5 % at T~600 K. However, the
temperature difference dependence of (PF)eng and Pd were outperformed by reference data as
shown in Fig. 6e and 6f, which also exhibits the trend of improvement with the increase of
sintering time for the samples.

This article is protected by copyright. All rights reserved.

 (a) 1.00 60 (b) 12 20

0.75 45 15

PF (10 Wm K )
9

-2

Pd (Wcm )
-1

-2
 (%)
10
ZT

0.50 30 6

-4
0.25 15 5
3
FeNb0.56V0.24Ti0.2Sb FeNb0.56V0.24Ti0.2Sb
0.00 0
0 0
400 600 800 1000 0 200 400 600
T (K) T (K)
Figure
7. Thermoelectric performance for FeNb0.56V0.24Ti0.2Sb: (a) temperature dependence of ZT represented by red

filled circle, and power factor represented by blue filled squire; (b) T dependence of efficiency  and power

density Pd, represented by the red filled hexagon and blue filled diamond, respectively.

In order to increase the (ZT)eng and  for the p type Nb0.8Ti0.2FeSb and make them
comparable to the Hf0.25Zr0.75NiSn0.97Sb0.03 materials for better devices, the vanadium doped
samples were prepared by the similar procedures, which was enlightened by outstanding
performance of vanadium doped materials in Zhu’s report.45 The temperature dependence of
thermoelectric performance for FeNb0.56V0.24Ti0.2Sb sample is shown in Fig.7. The details of
the temperature dependence of tot, , and S for this sample can be found in Fig. S7. Fig. 7a
shows temperature dependence of ZT and PF, represented by red filled circle and blue filled
square, respectively. It is noticed that although the PF were lower than that of Nb0.8Ti0.2FeSb
owing to the decreased hole mobility and electrical conductivity, the ZT curve in convex
upward shape reaches a peak of 0.92 at 925 K, favoring overall enhanced  as shown in Fig.
7b. As a result, the maximum T dependence of  reaches to 10.5 % at T~600 K
accompanied by Pd of 15 W cm-2, which is just slightly lowered than the Nb0.8Ti0.2 FeSb.

3. Conclusion.

This article is protected by copyright. All rights reserved.

 In conclusion, a series of representative HH thermoelectric materials were prepared by a
bottom-up nanostructure synthesis approach based on the displacement reaction between
metal chlorides and metal magnesium or lithium, followed by vacuum-assisted SPS process.
The samples are featured with dense dislocation arrays at the grain boundaries, leading to a
minimum κL of ~1 W m-1 K-1 at 900 K and comparable ZT (~1) and η (~11 %) for n-type
Hf0.25Zr0.75NiSn0.97Sb0.03. The phonon and electrical transports of p-type Nb0.8Ti0.2FeSb can
also be manipulated by mainly adjusting the dislocation defects at the grain boundaries,
leading to enhanced maximum power factor of 47×10-4 W m-1 K-2 and the predicted η of ~7.5
%. Further vanadium substitution as shown in FeNb0.56V0.24Ti0.2Sb significantly promotes the
η to ~ 11 %. This strategy can be extended to some other notable TE materials and a broad
range of advanced alloys and compounds for improved properties.

Methods

Synthesis. Six typical half-Heusler materials were synthesized through the combination of
mechano-chemical reaction and spark plasma sintering (SPS) process. All treatment and
processing of the material were carried out in an argon-filled glove box to avoid oxidation.
High purity metal chloride powders NbCl5 (99.9%), TiBr4 (98%), HfCl4 (99.9%), ZrCl4
(98%), FeCl2 (99.5%), VCl5 (99.5%), CoCl2 (97%), NiCl2 (98%), SbCl3 (99%), SnCl2 (98%),
and metal Mg/Li (99.9%) granules from Alfa Aesar were used for the synthesis. Raw
materials with a total weight of 5 g and appropriate ratios corresponding to the stoichiometric
ratio of HH materials were transferred into the stainless ball milling jar. High energy ball
milling process was then performed for 2 – 5 hrs by using a SPEX Sample Prep 8000
Mixer/Mill. The finally produced fine powder that is composed by the mixture of transitional
metal nanoparticles and MgCl2/LiCl were carefully loaded into the graphite die in the glove
box, and then treated by SPS process at 850-950 oC in the vacuumed chamber under 50 MPa

This article is protected by copyright. All rights reserved.

for 5min, 20 min and 4h, respectively. The final bulk pellet has the dimension of 12.7 mm

(diameter)  2 mm (thickness). The measured density of all the pellets was up to 94~98% of
the theoretical density. The obtained samples were further characterized by XRD, SEM and
TEM techniques.

Powder X-ray Diffraction (XRD). Powder XRD patterns were collected on a diffractometer
(PANalytical X’Pert Pro) in Bragg-Brentano reflection geometry with Cu K radiation
(=1.5418 Å) and operated at 40 kV and 40 mA. The patterns were carried out in the 2
range of 10 o -90 o with a step size of 0.02o at room temperature in air.

Microstructure analysis. The morphologies and element distribution of bulk samples were
characterized by a scanning electron microscopy (SEM) and energy dispersive X-ray
spectroscopy (EDX) on the S-5200 field emission SEM (Hitachi). The chemical compositions
were further checked by Inductively Coupled Plasma Mass Spectrometry (ICP-MS, X Series
2, Thermo Fisher Scientific). The microstructure and crystal defects at the atomic and lattice
levels in the bulk samples were investigated by a transmission electron microscope (TEM,
F20, FEI).

Thermoelectric transport properties. The temperature dependence of electrical resistivity

and Seebeck coefficient were measured by using a commercial system (LSR-3, Linseis). The

temperature dependence of thermal conductivity was calculated from the equation,  = DCpd,
where D, Cp, and d are the thermal diffusivity, specific heat capacity, and density,
respectively. The thermal diffusivity was measured by the laser flash method with a
commercial system (LFA-1000, Linseis). The specific heat capacity was measured by
differential scanning calorimeter (TA Q200). The density was determined by the Archimedes
method.

This article is protected by copyright. All rights reserved.

Reference

1. S. Challipalli, F. H. Froes, Mechanical Alloying of Titanium-Base Alloys. Adv Mater. 1993, 5, 96-106.

2. P. V. Liddicoat, X. Z. Liao, Y. Zhao, Y. Zhu, M. Y. Murashkin et al. Nanostructural hierarchy increases the
strength of aluminium alloys. Nat Commun, 2010, 1, 63.

3. J. R. Sootsman, D. Y. Chung, M. G. Kanatzidis, New and Old Concepts in Thermoelectric Materials. Angew
Chem-Int Edit, 2009, 48, 8616-8639.

4. I. S. Kim, K. H. Lee, H. A. Mun, H. S. Kim, S. W. Hwang et al. Dense dislocation arrays embedded in grain
boundaries for high-performance bulk thermoelectrics. Science, 2015, 348, 109-114.

5. B. Poudel, Q. Hao, Y. Ma, Y. C. Lan, A. Minnich et al. High-thermoelectric performance of nanostructured

bismuth antimony telluride bulk alloys. Science, 2008, 320, 634-638.

6. K. Biswas, J. Q. He, I. D. Blum, C. I. Wu, T. P. Hogan et al. High-performance bulk thermoelectrics with
all-scale hierarchical architectures. Nature, 2012, 489, 414-418.

7. K. F. Hsu, S. Loo, F. Guo, W. Chen, J. S. Dyck et al. Cubic AgPbmSbTe2+m: Bulk Thermoelectric Materials with
High Figure of Merit. Science, 2004, 303, 818-821.

8. T. Caillat, J.-P. F., A. Borshchevsky. Preparation and thermoelectric properties of semiconducting Zn4Sb3. J
Phys Chem Solids, 1997, 58, 1119-1125.

9. S. R. Brown, S. M. Kauzlarich, F. Gascoin, G. J. Snyder, Yb14MnSb11: New High Efficiency Thermoelectric

Material for Power Generation. Chem Mater, 2006, 18, 1873–1877.

10. G. Joshi, H. Lee, Y. C. Lan, X. Wang, G. H. Zhu et al. Enhanced thermoelectric Figure-of-Merit in
nanostructured p-type silicon germanium bulk alloys. Nano Lett, 2008, 8, 4670-4674.

11. X. W. Wang, H. Lee, Y. C. Lan, G. H. Zhu, G. Joshi et al. Enhanced thermoelectric Figure-of-Merit in
nanostructured n-type silicon germanium bulk alloys. Appl. Phys. Lett, 2008, 93, 193121.

12. H. L. Liu, X. Shi, F. F. Xu, L. L. Zhang, W. Q. Zhang et al. Copper ion liquid-like thermoelectrics. Nat Mater,
2012, 11, 422-425.

13. W. Y. Zhao, Z. Y. Liu, P. Wei, Q. J. Zhang, W. T. Zhu et al. Magnetoelectric interaction and transport
behaviours in magnetic nanocomposite thermoelectric materials. Nature Nanotech, 2017, 12, 55-60.

This article is protected by copyright. All rights reserved.

14. H. Z. Zhao, J. E. Sui, Z. J. Tang, Y. C. Lan, Q. Jie et al. High thermoelectric performance of MgAgSb-based
materials. Nano Energy, 2014, 7, 97-103.

15. L. D. Zhao, S. H. Lo, Y. S. Zhang, H. Sun, G. J. Tan et al. Ultralow thermal conductivity and high
thermoelectric figure of merit in SnSe crystals. Nature, 2014, 508, 373-378.

16. L. D. Zhao, J. Q. He, D. Berardan, Y. H. Lin, J. F. Li et al. BiCuSeO oxyselenides: new promising
thermoelectric materials. Energy Environ Sci, 2014, 7, 2900-2924.

17. W. Liu, X. J. Tan, K. Yin, H. J. Liu, X. F. Tang et al. Convergence of Conduction Bands as a Means of
Enhancing Thermoelectric Performance of n-Type Mg2Si1-xSnx Solid Solutions. Phys Rev Lett, 2012, 108, 5.

18. W. S. Liu, H. S. Kim, S. Chen. Q. Jie, B. Lv et al. n-Type thermoelectric material Mg2Sn0.75Ge0.25 for high
power generation. PNAS, 2015, 112, 3269-3274.

19. G. D. Mahan, Figure of merit for thermoelectrics. Journal of Applied Physics, 1989, 65, 1578.

20. S. Öğüt, K. M. Rabe, Band gap and stability in the ternary intermetallic compounds NiSnM(M=Ti,Zr,Hf): A
first-principles study. Phys Rev B, 1995, 51, 10443-10453.

21. C. Fu, T. Zhu, Y. Liu, H. Xie, X. Zhao, Band engineering of high performance p-type FeNbSb based
half-Heusler thermoelectric materials for figure of merit zT>1. Energy Environ Sci, 2015, 8, 216-220.

22. Y. Z. Pei, H. Wang, G. J. Snyder, Band Engineering of Thermoelectric Materials. Adv Mater, 2012, 24,
6125-6135.

23. S. Chen, K. C. Lukas, W. S. Liu, C. P. Opeil, G. Chen et al. Effect of Hf Concentration on Thermoelectric
Properties of Nanostructured N-Type Half-Heusler Materials HfxZr1-xNiSn0.99Sb0.01. Adv Energy Mater, 2013,
3, 1210-1214.

24. R. He, D. Kraemer, J. Mao, L. P. Zeng, Q. Jie et al. Achieving high power factor and output power density in
p-type half-Heusler Nb1-xTixFeSb. PNAS, 2016, 113, 13576-13581.

25. L. D. Zhao, G. Tan, S. Hao, J. He, Y. Pei et al. Ultrahigh power factor and thermoelectric performance in
hole-doped single-crystal SnSe. Science, 2015, 351, 141-144.

26. G. Rogl, A. Grytsiv, M. Gürth, A. Tavassoli, C. Ebner et al. Mechanical properties of half-Heusler alloys. Acta
Materialia, 2016, 107, 178-195.

27. R. He, S. Gahlawat, C. F. Guo, S. Chen, T. Dahal et al. Studies on mechanial properties of thermoelectric
materials by nanoindentation. Physica Status Solidi (a), 2015, 212, 2191-2195.

This article is protected by copyright. All rights reserved.

28. C. Fu, H. Xie, T. J. Zhu, J. Xie, X. B. Zhao, Enhanced phonon scattering by mass and strain field fluctuations
in Nb substituted FeVSb half-Heusler thermoelectric materials. Journal of Applied Physics, 2012, 112,
124915.

29. J. Callaway, Model for Lattice Thermal Conductivity at Low Temperatures. Phys Rev, 1959, 113,
1046-1051.

30. C. Fu, H. Wu, Y. Liu, J. He, X. Zhao et al. Enhancing the Figure of Merit of Heavy-Band Thermoelectric
Materials Through Hierarchical Phonon Scattering. Advanced Science, 2016, 1600035.

31. C. Yu, T.-J. Zhu, R.-Z. Shi, Y. Zhang, X.-B. Zhao et al. High-performance half-Heusler thermoelectric
materials Hf1−xZrxNiSn1−ySby prepared by levitation melting and spark plasma sintering. Acta Materialia,
2009, 57, 2757-2764.

32. D. G. Cahill, S. K. Watson, R. O. Pohl, Lower limit to the thermal-conductivity of disorderd crystals. Phys
Rev B, 1992, 46, 6131-6140.

33. V. I. Fistul, Heavily Doped Semiconductors. Plenum Press , New York(1995)

34. Z. W. Chen, B. H. Ge, W. Li, S. Q. Lin, J. W. Shen et al. Vacancy-induced dislocations within grains for
high-performance PbSe thermoelelctrics. Nature Comm, 2017, 8, doi:10.1038/ncomms13828.

35. F. Casper, T. Graf, S. Chadov, B. Balke, C. Felser, Half-Heusler compounds: novel materials for energy and
spintronic applications. Semicond Sci Tech, 2012, 27, 063001.

36. F. J. Humphreys, M. Hatherly, Recrystallization and Related Annealing Phenomena. Elsevier, Oxford(2004)

37. M. Nader, F. Aldinger, M. J. Hoffmann, Influence of the α/β-SiC phase transformation on microstructural
development and mechanical properties of liquid phase sintered silicon carbide. J Mater Sci, 1999, 34,
1197-1204.

38. P. J. Jorgensen, Liquid-phase sintering of SmCo5. Journal of Applied Physics, 1973, 44, 2876.

39. H. S. Kim, S. D. Kang, Y. L. Tang, R. Hanus, G. J. Snyder, Dislocation strain as the mechanism of phonon
scattering at grain boundaries. Materials Horizons, 2016, 3, 234-240.

40. M. Jonson, G. D. Mahan, Mott Formula for The Thermopower and The Widemann-Franz Law. Phys Rev B,
1980, 21, 4223-4229.

41. T. Zhu, C. Fu, H. Xie, Y. Liu, X. Zhao, High efficiency Half-Heusler thermoelectric materials for energy
harvesting. Adv Energy Mater, 2015, 5, 1500588.

This article is protected by copyright. All rights reserved.

42. H. H. Xie, H. Wang, Y. Z. Pei, C. G. Fu, X. H. Liu et al. Beneficial Contribution of Alloy Disorder to Electron
and Phonon Transport in Half-Heusler Thermoelectric Materials. Adv Funct Mater, 2013, 23, 5123-5130.

43. T. M. Tritt, Thermoelectric Phenomena, Materials, and Applications. Annu Rev Mater Res, 2011, 41,
433-448.

44. H. S. Kim, W. Liu, G. Chen, C. W. Chu, Z. F. Ren, Relationship between thermoelectric figure of merit and
energy conversion efficiency. Proceedings of the National Academy of Sciences of the United States of
America, 2015, 112, 8205-8210.

45. C. G. Fu, Y. T. Liu, X. B. Zhao, T. J. Zhu, Are solid solution better in FeNbSb-based thermoelectrics? Adv.
Electron. Mater, 2016, 1600394.

Acknowledgements

We acknowledge the funding support from the National Natural Science Foundation of China
(NSFC) under the grant number of 51572287, U1601213 and 51472052, and the financial
support from Institute of Physics at Chinese Academy of Sciences. The work performed at
the University of Houston is funded by “Solid State Solar Thermal Energy Conversion Center
(S3TEC)”, an Energy Frontier Research Center funded by the U.S. Department of Energy,
Office of Science, Office of Basic Energy Science under award number DE-SC0001299. We
would thank Professor Tiejun Zhu and Mr. Yintu Liu at Zhejiang University for their
assistance in the electrical transport property measurement for the p-type NdFeSb based
materials. We also thank Professors Lidong Chen and Xun Shi at Shanghai Institute of
Ceramics, Chinese Academy of Sciences for their insightful suggestions.

Author contributions

This paper was written by H.Z.Z. and S.M.L, and edited by G.J.S., Z.F.R., and X.L.C.
Sample synthesis, structural characterization and thermoelectric transport property

This article is protected by copyright. All rights reserved.

measurements were performed by B.L.C., S.M.L., Y.Z.P., and J.W.S. The computing
programs used for modeling of the inclusive phonon scattering mechanisms, and the
modeling and fittings of the electrical transport properties based on the two-band (conduction
and valence bands) electronic structure models were collaboratively written and performed
by N.L., S.M.L. with assistance from S.L.

Additional information

Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online. Correspondence and requests for materials
should be addressed to H. Zhao.

Competing financial interests

The authors declare no competing financial interests.

TOC for paper aenm.201700446

A series of representative HH thermoelectric materials have been prepared by a bottom-up

nanostructure synthesis approach based on the displacement reaction between metal chlorides
and lithium, followed by vacuum-assisted SPS process. These samples are featured with
dense dislocation arrays at the grain boundaries. The phonon and electrical transports can be
manipulated by adjusting the dislocations, leading to enhanced TE performances.

This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.