Materials Research Express

ACCEPTED MANUSCRIPT

Electronic, and thermoelectric properties of half-heusler compounds

MCoSb (M=Ti, Zr, Hf): a first principles study
To cite this article before publication: Himanshu Joshi et al 2019 Mater. Res. Express in press https://doi.org/10.1088/2053-1591/ab0c68

Manuscript version: Accepted Manuscript

Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”

This Accepted Manuscript is © 2019 IOP Publishing Ltd.

During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.
As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript is available for reuse
under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.

After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence https://creativecommons.org/licences/by-nc-nd/3.0

Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions will likely be
required. All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.

View the article online for updates and enhancements.

This content was downloaded from IP address 128.252.67.66 on 06/03/2019 at 15:04

Page 1 of 14 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2

IOP Publishing AUTHOR SUBMITTED MANUSCRIPT - MRX-112122 Materials Research Express

1
Journal XX (XXXX) XXXXXX https://doi.org/XXXX/XXXX
2
3
4
5
6
7
8
Electronic and Thermoelectric properties of Half-

pt
9
10 Heusler Compounds MCoSb (M=Ti, Zr, Hf): A first
11
12
13
principle study.

cri
14
15
16 H. Joshi1, D.P. Rai2, 3*, Amel Laref 4 and R. K. Thapa1,5
17
18
19 1
Condensed Matter Theory Research Group, Department of Physics, Mizoram University, Aizawl, Mizoram, India-796004.
20 2

us
Division of Computational Physics, Institute for Computational Science, Ton Duc Thang University, Ho Chi Minh City,
21
Vietnam
22 3
Faculty of Electrical and Electronics Engineering, Ton Duc Thang University, Ho Chi Minh City, Vietnam
23 4
24 Department of Physics, College of Science, King Saud University, Riyadh, Saudi Arabia
5
25 Condensed Matter Theory Research Center, Butwal, Rupendehi, Nepal.
26
27
28
29
Vietnam. E-mail address: [email protected] (D. P. Rai)
an
* Corresponding author at: Division of Computational Physics, Institute for Computational Science, Ton Duc Thang University, Ho Chi Minh City,

Abstract
30
31 Full Potential Linearized Augmented Plane Wave method (FP-LAPW) method based on the Density Functional Theory
32
dM
(DFT) in combination with Boltzmann semi-classical transport equation (BTE) have been used to investigate the electronic,
33 and thermoelectric properties of MCoSb type half-Heusler compounds. A most effective exchange potential developed by
34 Tran and Blaha by modifying Becke-Johnson exchange potential called TB_mBJ potential has also been used to open the
35 degenerated band gaps. The calculated results were compared with the available theoretical and experimental data. On careful
36
observation of the band structure one can predict that MCoSb is a narrow indirect band gap semiconductor with electron
37
transition along Γ-X symmetry point in the first Brillouin zone. The vibrational and thermodynamic properties were also
38
studied form the quasi-harmonic approximations to confirm their mechanical and thermodynamical stability. Furthermore, the
39
lattice thermal conductivity (kl) has been calculated from the Slack’s equation. The low value of kl is highly promising for
40
applications in the thermoelectric devices. The electronic part of thermoelectric parameters are calculated from Boltzmann
41
42 semi-classical transport equation (BTE). However, they are associated with the relaxation time (τ) and decoupling τ is
43 challenging. Therefore, we have calculated τ as a function of temperature by using the Bardeen and Shockley’s deformation
pte

44 potential theory. The calculated ZT value is highest in case of TiCoSb at 800 K. The order of ZT values in MCoSb at a
45 temperature range from 600-900 K is TiCoSb > ZrCoSb > HfCoSb. The carrier concentration has been optimized such as to
46 give the maximum thermoelectric performance.
47
48 Keywords: Heusler, Density Functional Theory, DOS, Band structure, Phonons, Seebeck Coefficient, Thermal
49 conductivity, Relaxation time, and Thermoelectric Figure of Merit.
50
51
ce

52
53 peculiar half metallic property in NiMnSb from an ab initio
54 1. Introduction approach [1]. Besides half metallicity, the other interesting
55 properties of these compounds include high Curie
In recent years, Half-Heusler (HH) compounds has
56 temperature, low thermal conductivity, high spin
become the center of research attention due to their thermal
Ac

57 polarization, and are considered as the most stable zinc-

58 stability, mechanical strength, non-toxicity and cost-
blende structure [2-6]. In general, the chemical formula of
59 effectiveness. A mass research interest has been focused on
half-Heusler compound is represented as XYZ, where X and
60 the Heusler compounds only after the discovery of its
Y are the transition metals and Z is the main group element

xxxx-xxxx/xx/xxxxxx 1 © xxxx IOP Publishing Ltd

 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2 Page 2 of 14

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 in the periodic table. It crystallizes in MgAgAs type structure detail electronic structure calculation of HfCoSb and ZrCoSb
4 with space group F 43m . The HH alloys comprises of X n+ have been performed by Lee et al. in 2011 from first
5 principle approach [26]. Shiomi et al. reported the phonon
ion, stuffing the zinc-blend YZn- sublattice, where the
6 dispersion and phonon relaxation time for ZrCoSb to
number of valence electron count of YZn- is 18 (d10+s2+p6).
7 compute the phononic contribution to the lattice thermal
These types of HH compounds are of particular interest
8 conductivity [27-29]. Although the theoretical studies on

pt
because they are non-magnetic and exhibits semiconducting
9 these type of compounds are already in progress but the
10 behavior with narrow band gaps, which are favorable for
thermoelectric applications [7-10]. The Slater-Pauling rule detail report on the optical, thermodynamic, elastic and
11 vibrational properties are still not concrete. Therefore, a
12 has its own importance in determining the magnetic
properties of HH compounds. The magnetic moment of HH detail studies from the scratch is much needed to understand
13
the physical properties in an atomic scale from the first

cri
14 compounds can be predicted using the Slater-Pauling
equation which is expressed as M = (Z-18)  B , where M and principles approach. Hence, in this present work, we attempt
15
a rigorous theoretical study by calculating the structural,
16 Z are the magnetic moment and the number of valence
electronic, vibrational, thermodynamic and thermoelectric
17 electron in the unit cell [7]. The total number of valence
properties of these compounds. However, the thermodynamic
18 electrons in MCoSb (M=Ti, Zr, Hf) is 18, so we can predict
19 properties are reported for the first time. Understanding the
it to be a non-magnetic semiconductor, having magnetic
20 vibrational properties can provide valuable information
moment 0  B . The total valence electron count can be

us
21 regarding the structural stability and phase transition of the
changed by varying the X, Y or Z compositions in the XYZ material. Thermoelectric properties are also covered by
22
23 formula and hence the physical properties of MCoSb can be calculating the lattice thermal conductivity and relaxation
24 tuned further. Thermoelectricity has emerged as a core time. The electronic part of thermoelectric parameters are
25 research problem in the field of material sciences because it calculated from Boltzmann semi- classical transport equation
26 is thought to be an alternative energy resource which can (BTE) within a rigid band approximation (RBA).
27
28
29
reduce the consumption of fossil fuels and provide green
energy [11, 12]. Continuous efforts are made to produce an
an
eco-friendly energy material which act smartly in response to
2. Computational Methodology
The electronic structure, optical and elastic properties were
30 external fields like pressure, light, temperature, wind, wave,
calculated using Full Potential Linearized Augmented Plane
31 etc. The utilization of waste heat from factories and
Wave Method (FP-LAPW), which is based upon Density
32 automobiles by the solid state materials and converting into
dM
Functional Theory (DFT) [30, 31] and is implemented in the
33 usable form of electricity is a thermoelectric innovation.
code Wien2K [32]. The electron exchange and correlation
34 Over the last few decades, several materials were studied
35 both theoretically and experimentally, in search of high was treated by Generalized Gradient Approximation (GGA)
36 efficient thermoelectric materials. Most effective ones found and Tran and Blaha’s modified Becke-Johnson potential
37 being PbTe (ZT~1.5), Bi2Te3 (ZT~1), Sb2Te3 (ZT~1), (TB-mBJ). Usually, the Generalized Gradient Approximation
38 Ba8Ga16Ge30 (ZT~1.7), Mg2Si (ZT ~1.4), Oxide materials [33] underestimates the band gap of semiconductors [34, 35].
39 (ZT ~2.7) [13-15] etc. However, their application is limited It is because GGA does not precisely account for the
40 due to poor physical and chemical stability at high quasiparticle self-energy [36], which leads to this
41 temperatures [16,17]. The other reason being their toxic underestimation. Therefore, for better computation of the
42 constituents [18-20]. The choice of compounds in our present band gap, the electronic band structure is also calculated
43 using mBJ potential. In order for scf calculation to converse
pte

work is free from toxic components and offers high

44 temperature working range. well, energy cut off = -8.0 Ry was used to separate out the
45 core states and the valence states. A fine 19 × 19 × 19 k-
46 TiCoSb was first synthesized by Gladysevskij and points mesh, with 7000 k-points for HfCoSb and 10 × 10 ×
47 coworkers in1963 and (Zr, Hf)CoSb by Marazza et al. in 10 k point mesh, with 1000 k-points each for ZrCoSb and
48 1974. Xia et al.(2000) experimentally determined the TiCoSb was chosen after optimization with respect to
49 thermoelectric properties of MCoSb [21], at low energy. To ensure better convergence of the self-consistency
50 temperatures ranging from 100K to300K. Stadnyk et al. cycle, we have optimized some of the important parameters
51
ce

doped TiCoSb with Ni impurities at the Co-sites and reported like RKmax (which is the product of the least atomic radii of
52 an enhanced electrical conductivity [22]. Sekimoto et al. the muffin-tin spheres and the maximum value of wave
53 experimentally determined the ZT value of about 0.02 in vector, in the interstitial region, for expansion of wave
54 TiCoSb at higher temperature range ~1000K, [23]. The same functions of electron), Gmax (The cut off vector in the
55 group reported an enhanced ZT value of 0.03 for TiCoSb interstitial region, for charge density and potential
56 with thermal conductivity (κl) at higher temperature[24]. The expansion), k-point (which is the number of selected points
Ac

57 aforementioned results favor that MCoSb is a potential in the first Brillouin zone) and RMT (The smallest atomic
58 candidate for thermoelectric application. Yang et al.(2008) radii of the muffin-tin spheres in the unit cell). The optimized
59 have also reported a high power factor in TiCoSb and values are shown in table 1 and is used throughout the
60 ZrCoSb from their theoretical investigation [25]. The first calculation. The energy convergence was set to 10-5 Ry and

2
Page 3 of 14 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 charge to 0.001 e-. The semi-core states were treated semi- a
4 relativistically i.e. it was treated ignoring the spin orbit s , ( a  a  a0 ) and a0 being the equilibrium lattice
a0
5 coupling. The phonon vibrational properties has been
6 constant. E is calculated as the energy change in
calculated using Phonopy code [37], employing the force
7 constant method, based upon density-functional perturbation
conduction band minimum (CBM) (for n-type material)
8 theory within the framework of quasi-harmonic
or the energy change in the valence band maximum

pt
9 approximation of phonon frequencies [38]. In order to
(VBM) (for p-type material) with the application of
10 calculate the phonon DOS and their dispersion relation, a
strain. The effective mass, m* was calculated within the
11 2×2×2 supercell was created and self-consistency
framework of parabolic band approximation [44], by
12 calculations were performed using Wien2K code under GGA
parabolic fitting of the wave vector (K) vs energy (E) of
13 the CBM or the VBM, depending upon the nature of the
approximation.

cri
14 charge carriers and is determined by [44]-
15 Table 1: Optimized K-points, RKmax; Gmax and RMT (a.u.) 2

16 values used for DFT calculation. m * (2)

dE 2 / dK 2
17
Compound k-points RKmax Gmax RMT The relaxation time is then calculated with respect to
18
19 temperature (K). Electrical conductivity (σ/τ) and the
TiCoSb 1000 8 13 Ti=1.812
20 electronic thermal conductivity (κe/τ), calculated from
Co=2.016
BoltzTraP is made τ independent as σ = σ/τ  τ and κ =

us
21 Sb=2.288
22 κe /τ  τ.
ZrCoSb 1000 9 13 Zr=2.346
23 Co=2.073 The important parameter involved in the calculation
24 Sb=2.573 of efficiency is Figure of Merit (ZT) and is also known
25 as the maximum efficiency condition given by
26 HfCoSb 7000 7 12 Hf=2.576
27
28
29 The temperature dependence of the thermodynamic
Co=2.035
Sb=2.313
an ZT 
S 2 T
e   p
(3)

30 parameters of MCoSb was calculated implementing the The phononic contribution to thermal conductivity (κp)
31 Gibbs package [39, 40], in the framework of quasi-harmonic is obtained from Slack’s equation [45] as
32 Debye model. The master input file, used in Wien2K code M 3D
dM
33 for calculation of electronic structure, acts as the key input to kp  A (4)
34  2 n 2/3T
this code. The approximations and equations involved in the
35 calculation are discussed in the results section. The where, M is the average Molecular weight, n is the
36 calculation of thermoelectric properties further involves a number of atoms per unit cell, T is the absolute
37 post DFT treatment involving a program cord BoltzTraP temperature,  is the Grüneisen parameter and  D the
38 [41]. The electronic structure obtained from DFT Debye’s temperature. A is a coefficient dependent upon
39 calculations act as the key input for the calculation of  in units of W mol/kg/m2/K2. The thermoelectric figure
40 transport properties. The electron transport properties like
41 of merit is finally computed by combining the results
electrical conductivity, electronic thermal conductivity and obtained after solving Slack’s and Boltzmann transport
42 Seebeck coefficient has a strong dependency of electronic
43 equations.
pte

band structures. However, the transport properties calculated

44
from BoltzTraP are coupled with the relaxation time (τ), 3. Results & Discussions
45
which is an unknown parameter. Therefore, we have
46 3.1 Structural Properties
calculated the relaxation time to decouple and to obtain the
47
48 solid electron transport properties independent of (τ). The The half-Heusler MCoSb crystallizes in MgAgAs type
49 relaxation time is calculated by using the Bardeen and structure, possessing cubic phase with space group F 43m .
50 Shockley’s deformation potential theory [42, 43] as given in The unit cell of MCoSb consists of four formula units with
51 Eq. (1)
ce

M, Co and Sb atoms located at the Wyckoff positions 4c:

52 2 2 4
C3 D (0.25, 0.25, 0.25); 4a: (0, 0, 0) and 4d: (0.75, 0.75, 0.75),
 (1)
53 3E (m * k BT )3/ 2
2
while the 4b (0.5, 0.5, 0.5) position is vacant [29, 46, 47]. Its
54 unit cell structure is shown in Fig.1 (a). Previous studies
Here, C 3D is a combination of elastic constants, which
55 reported that MCoSb compounds crystallizes in AlLiSi type
56 was calculated within the DFT framework. E is the structure [48]. Also, MgAgAs type structure can have
Ac

57 deformation potential coefficient along the direction β, Wyckoff positions 4a: (0, 0, 0); 4c: (0.25, 0.25, 0.25) and 4b:
58
E (0.5, 0.5, 0.5), with the 4d (0.75, 0.75, 0.75) position empty
59 defined as , where s is the strain given by [49,50].Thus, we have made a simple investigation to find
60 s
out the most favourable structure by calculating their

3
 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2 Page 4 of 14

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 minimum energy. The optimized lattice constant is obtained
4 by fitting the Energy vs Volume curve with respect to E form 
1
x yz
 EtotMCoSb  xEsolid
M
 yEsolid
Co
 zEsolid
Sb
 (7)
5 Murnaghan’s equation of state [51], using Eq. 5
6 with M = Ti, Zr and Hf. The calculation of formation
 V  B0  
*

7 BV  V energies are presented in Table 3. The calculated formation

Etot (V )  * 0*  0   B0* 1  0   1  E0
8 B0 ( B0 1)  V   V   energies for TiCoSb, ZrCoSb and HfCoSb are –4.04
 eV/atom, –4.9022 eV/atom and –4.4392 eV/atom,

pt
9
(5) respectively. Negative values of formation energy indicates
10
where, V and V0 are the unit volumes at normal condition and that the structure is energetically favourable. Such structures
11
ground state, B0 is the bulk modulus and B0* is the derivative are most likely to be formed from experiments [60].
12
of B0 at V=V0. The energy versus volume curves for MCoSb Table 2: Lattice constants (a) in Å, Bulk modules (B) in GPa
13
and pressure derivative of bulk modules (B’) obtained using

cri
14 is shown in Fig. 1 (b-d). Two atomic positions are considered
for calculation. Position 1: M = (0, 0, 0); Co = (0.25, 0.25, Murnaghan’s equation of state.
15
16 0.25); Sb = (0.5, 0.5, 0.5). Position 2: M = (0.25, 0.25, 0.25); Present work Previous report
17 Co = (0, 0, 0); Sb = (0.75, 0.75, 0.75). The position with
Compd. a B B’ a B B’ Refs.
18 lowest ground state energy is the most stable position and
19 correspond to a stable structure. From Fig.1(b-d) it can be TiCoSb 5.88 146.91 4.59 5.90 xx xx [47]
20 seen that the atomic positions corresponding to position 2 is 5.88 xx xx [24]

us
5.88 xx xx [22]
21 the most stable one, so our calculations are based on this
5.89 xx xx [29]
22 structure of MCoSb. 5.88 xx xx [23]
23 5.97 xx xx [55]
24
25 ZrCoSb 6.09 139.86 4.85 6.06 xx xx [24]
26 6.06 xx xx [56]
27
28
29
an 6.06
6.06
6.09
xx
xx
xx
xx
xx
xx
[23]
[48]
[57]
6.07 xx xx [55]
30 6.10 xx xx [29]
31
32 HfCoSb 6.05 145.12 4.93 6.05 xx xx [57]
dM
6.04 xx xx [45]
33
6.04 xx xx [24]
34 6.03 xx xx [23]
35 6.06 137.7 4.33 [58]
36
37 Table 3: Calculated individual energies of Ti Zr, Hf, Co, Sb,
38 and formation energy (Efor) in eV x 103.
39 Fig. 1: (a) Unit cell structure of MCoSb, Volume Optimization
40 curve for (b) TiCoSb, (c) ZrCoSb and for (d) HfCoSb.
Compd. ETi EZr E Hf ECo ESb E for
41
TiCoSb -1.71 xxxx xxxx -2.79 -12.97 -0.0040
42 The calculated lattice constants are 5.8839 Å, 6.0912 Å and
ZrCoSb xxxx -7.20 xxxx -2.79 -12.96 -0.0049
43
pte

6.0574 Å respectively for TiCoSb, ZrCoSb and HfCoSb. HfCoSb xxxx xxxx -30.2 -2.79 -12.97 -0.0044
44 These values are overestimated as compared to available
45 experimental data [22, 23, 47, 52]. However, our results are
3.2 Electronic Properties
46 in good agreement with the previous theoretical results
47 obtained from GGA functional [53, 54]. The calculated The electronic structure (density of states and band
48 lattice parameters, bulk modules and the pressure derivative structures) of MCoSb were calculated for the relaxed
49 of bulk modules are listed in Table 2. For further structures using GGA and mBJ approximations [see Figs. (2-
50 confirmation of structural stability, we have calculated the 6)]. In Fig. 2 (a-c), we found that both GGA and mJB gives
51
ce

formation energy for MCoSb (M=Ti, Zr and Hf) from the the similar electronic structures in MCoSb, no significance
52 following Eq. (6) [58-61] changes were observed. The similarity in the electronic
53
 Etot  xEsolid 
1 structure arises due to the presence of localized electron
54 E form  A
 yEsolid
B
(6) density. However, the mBJ is more effective for the kinetic
55 x y
density with orbital independent electrons. The DOS peaks
56 where, Etot is the total energy of the unit cell, x and y are the within mBJ are slightly reduced and shifted towards higher
Ac

57 total number of the composite atoms A and B respectively. energy ranges due to Ti-d electrons. The shifts are more
58 A
Esolid B
and Esolid are the energies of atom A and atom B in solid distinguished towards the conduction region. Whereas, GGA
59
state. In case of half-Heusler MCoSb, equation (6) can be includes orbital energy obtained from self-consistent
60 modified as given by Eq. 7-

4
Page 5 of 14 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 calculations, such that DOS results depends strongly on the Similar DOS characteristics are presented in Fig. 4 (a-c) for
4 energy exchange correlation functional. Thus, due to this ZrCoSb and HfCoSb [Fig.5 (a-c)]. Thus, it can be claimed
5 reason the DOS peaks within mBJ are slightly lower than that the high DOS peaks in the conduction region of HH
6 that of GGA. Sharp peaks are observed from both the MCoSb is due to the d-states of M atom (M=Ti, Zr and Hf)
7 approximations, near the Fermi level (EF), for all MCoSb and and that in the valence region is mainly due to the d-states of
8 can be clearly seen that zero density of states exists at EF, Co atom.

pt
9 indicating no presence of pseudogap in the compounds.
10
11
12
13

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an Fig. 3: Partial Density of States plot for (a) Ti (b) Co and (c) Sb
Fig. 2: Total Density of States plot (red mBJ and green GGA) for atom of TiCoSb compound.
30 (a) TiCoSb (b) ZrCoSb and (c) HfCoSb.
31
32 Fig. (3-5) (a-c), displays the partial DOS of MCoSb. It has
dM
33 been noticed that Co atom has the maximum contribution,
34 whereas, Sb atom has the lowest contribution to the total
35 DOS. This indicates that more number of Co atomic states is
36 available for occupation. In order to explain the partial DOS,
37 we have divided the valence and the conduction region into
38 the following category: lower valence region (between -8eV
39 to -6 eV), middle valence region (between -6 eV to -4 eV),
40 upper valence region (between -4 eV to 0 eV), lower
41 conduction region (between 1 eV to 4 eV) and upper
42 conduction region (between 4 eV to 8 eV). A closer look at
43
pte

Fig.3 (a-c) shows that the maximum contribution to the total

44 DOS of TiCoSb comes from the Ti-3d and Co-3d states. The
45
lower valence region has no contribution to the total DOS of
46
TiCoSb. The middle valence region has mixed DOS arising
47
from Ti-3d, Co-3d and Sb-5p states. All these states has
48
almost equal contribution (<0.3 states/eV) in this region. The
49
50 upper valence region is dominated by the 3d states of atom
51 Co. High peaks are observed at around 1 eV and 1.5 eV.
ce

52 Considerable contribution from 3d states of Ti atom is Fig. 4: Partial Density of States plot for (a) Zr (b) Co and (c) Sb
53 observed (highest contribution being <1.5 states/eV) and a atom of ZrCoSb compound.
54 very small (~0.54 states/eV) contribution from Sb-5p states.
55 The lower conduction region shows strong dominance of Ti-
56 3d (d-eg, d-t2g) states. Sharp DOS peaks (~5 states/eV) are
Ac

57 observed in this region. The upper conduction band region

58 has negligible contribution from Co and Sb atoms. Therefore
59 it can be reported that the DOS in the conduction region is
60 mainly due to 3d states of Ti atom in half Heusler TiCoSb.

5
 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2 Page 6 of 14

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 In the upper valence band region (0 eV to -4 eV),
4 threefold degenerate bands are observed at the Γ point,
5 whereas a twofold degenerate bands exists at L point. The
6 band just above -2 eV at the Γ point does not participate in
7 carrier transport directly, but its position affects the transport
8 properties associated with the valence band. These bands are

pt
9 mainly formed due to the d orbitals (d-t2g) of Co atoms.
10 Other bands seen at the top of the valence band at the same
11 symmetry point comes from the hybridization of d-eg and d-
12 t2g orbitals of M and Co atoms (M = Ti, Zr, Hf). Since the
13
VBM differs in case of ZrCoSb, may change the nature of

cri
14
charge carriers. ZrCoSb will be more suitable to dope with p-
15
type impurity instead of the n-type. This is because, the
16
VBM of ZrCoSb lies in the L-symmetry point instead of Γ as
17
in case of TiCoSb and HfCoSb, which will be more effective
18
19 towards n-type dopant.
20

us
21
22
23
24 Fig. 5: Partial Density of States plot for (a) Hf (b) Co and (c) Sb
25 atom of HfCoSb compound.
26
27
28
29
The band structure of MCoSb is plotted along high
symmetry directions W-L-Γ-X-W-K in the irreducible part of
the first Brillouin zone [see Fig.6 (a-c)]. The Fermi level is
an
30 set at zero energy. All these compounds exhibit a small
31 indirect band gap. The red lines in the figures refers to mBJ
32 approximation, while the green lines are from GGA Fig. 6: Band structure plots (red mBJ and green GGA) for (a)
dM
33 approximation. The valence band maxima (VBM) and TiCoSb, (b) ZrCoSb and (c) HfCoSb along W-L-Γ-X-W-K.
34 conduction band minima (CBM) lies at Γ and X point, Table 4: Calculated band gap (∆EG) in eV for TiCoSb,
35 respectively for both TiCoSb and HfCoSb. Whereas, in case ZrCoSb and HfCoSb
36 of ZrCoSb the VBM shifted to L symmetry point and CBM GGA mBJ
37 is fixed at the symmetry point X. Similar trends in the band
38 gap had been reported in the previous literature [29,57]. The Compd. ∆EG previous works ∆EG Ref.
39 calculated values of the band gap from GGA and mBJ TiCoSb 1.04 1.05 1.057 [29]
40 approximation is presented in Table 4. The highest band gap 1.06 [44]
41 value is obtained for HfCoSb (1.137 eV) and the lowest is 0.19 (400K) [43]
42 observed for TiCoSb (1.04 eV). These values are obtained 0.95 (300K) [21]
43
pte

0.40 (900K) [23]

from GGA functional and are in close agreement with the
44 0.16 (400K) [23]
available theoretical results (see Table 4). The band gap
45 1.05 [25]
values drastically varies as compared to the experimental
46
one. The discrepancy may be attributed to the temperature ZrCoSb 1.073 1.06 1.13 [54]
47
48 influence, obtained at 300-900K. The band gap varies as ΔEG 1.061 [29]
0.14 (400K) [23]
49 (TiCoSb) < ΔEG (ZrCoSb) < ΔEG (HfCoSb). From the band
1.06 [25]
50 structure plot (Fig. 6), it is observed that the calculation of
51 electronic structure with mBJ potential is almost ineffective
ce

HfCoSb 1.137 1.13 1.178 [54]

52 as there is no significance improvement in the band gap. 1.03 [28]
Several authors [1, 62] have reported the inefficiency of mBJ 1.135 [29]
53 0.07 (400K) [23]
54 potential on narrow band gap HH compounds. We predict
55 that mBJ potential is not an efficient theoretical technique for
Here, also we have divided the conduction band into two
56 calculation of band structure of MCoSb, which falls under
regions, the lower conduction band (between 0-4 eV) and the
Ac

57 IVB-VIIIB-VA type semiconductor. Even highly

upper conduction band (between 4-8 eV). For ZrCoSb and
58 sophisticated hybrid functional proposed by Heyd-Scuseria-
HfCoSb a high dispersion of band energies are observed at Γ
59 Enzerhof (HSE06) [62] may have negligible effect on the
point at ~2 eV. The dispersion continues up to X symmetry
60 band gap.

6
Page 7 of 14 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 point. This increases the effective mass of the electron, near the phonon density of states. The maximum amplitude of
4 the conduction band minima, along the X-Γ direction. This acoustic frequency mode are found to be ~5 THz for TiCoSb
5 band is a result of hybridization between the d-t2g states of M and ZrCoSb, whereas for HfCoSb, it is found to be ~4 THz.
6 atom and the d-eg states of Co atom. The sharp peaks at the Γ Unfortunately, no experimental data are available for
7 point in the lower conduction band region (near about 4 eV) comparison of the dispersion results, however our results are
8 are due to the s orbitals of Sb atom. The bands originating at in good agreement when compared with the theoretical

pt
9 the upper conduction band region is mainly due to the d results reported in Ref.[28, 29]. The high ranges of
10 states of M atom and p states of Sb atom. Thus the origin of frequencies in the acoustic mode resulting in large group
11 band gap in HH MCoSb can be explained as, due to the velocities along with strong dispersion in the acoustic mode,
12 hybridization of d-t2g orbitals of M atom and p orbitals of Sb results in maximum contribution to heat transfer by the
13
atom with d-eg orbitals of Co atom. In the valence band acoustic branch. The heat transfer by optical branches is

cri
14
region, large peaks of DOS are observed near the band edges. negligible. Thus, lattice thermal conductivity is mainly due
15
This indicates good mobility of charge carriers in MCoSb to the acoustic branches. The mixing of acoustic and optical
16
compounds and is beneficial for high Seebeck coefficient (S) mode indicates phonon-phonon scattering which as a result
17
values. It will be discussed in more details in the reduces the lattice thermal conductivity of a compound.
18
19 thermoelectric properties section. Therefore, TiCoSb should have higher lattice thermal
20 conductivity than ZrCoSb and HfCoSb. The lattice thermal

us
21 3.3 Vibrational Properties conductivity calculations are done in thermoelectric
22 properties section. The peaks in the phonon DOS which
The phonon dispersion relations and the phonon DOS are
23 corresponds to the flat regions in the phonon dispersion
obtained using harmonic (second order) force constants
24 within the framework of density functional perturbation
curve indicates localized states. The difference in the
25 theory (DFPT), with an assumption that the atoms displaced
contribution to the phonon DOS by different atoms occurs
26 due to the difference of mass of the atoms. Heavier atom
27
28
29
crystal. Analyzing a force that is associated with such a
displacement provides phonon frequencies. The second order
an
from their equilibrium position raises force on all atoms in a
contributes to the low modes of frequencies while light atom
contributes to the higher ones. From the DOS plot it can be
seen that the main contribution to the acoustic mode of
force constant is obtained by [37]
30 vibration comes from the Sb atom of HH compound (TiCoSb
F (lk )
31  i j (lk , l * k * )   i * * (8) and ZrCoSb).
32 u j (l k )
dM
33 where, i and j are the Cartesian index and l and k are the label
34 of unit cell and atoms in each unit cell. The phonon band
35 structure and phonon density of states are shown in Fig. 7.
36 The phonon DOS is defined by the expression-
37 1
38 g ( )    (  ij ) (9)
N ij
39
40 N is the number of unit cells in a crystal. The plot of phonon
41 dispersion curves along the Γ-W-Γ-U-L-X symmetry
42 directions along with the phonon total and partial density of
43
pte

states are shown in Fig. 7. The phonon dispersion band

44 occurs along the Γ-symmetry point for all of MCoSb (M=Ti,
45 Zr and Hf). Since, the primitive cells of MCoSb contains 3
46 atoms each, hence nine modes of vibration exists, out of
47 which three are acoustic modes and six are optical modes of
48 vibration. The acoustic modes of vibration consists of one
49 longitudinal and two transverse branches, while the optical
50 mode consists of two longitudinal and four transverse
51
ce

branches. The bands that lies in the frequency range of 0-4

52 THz are the acoustic modes of vibration, the bands lying
53 above 4 THz are optical modes. It can be seen from the
54 figure that the imaginary (negative) frequencies are absent in
55 the dispersion curve, this strongly supports the dynamical
56 stability of HH compound MCoSb. The compound TiCoSb
Ac

57 shows a small frequency gap between the optical and Fig. 7: The phonon dispersion curves and the phonon density of
58
acoustic modes. ZrCoSb and HfCoSb shows a mixing of the states (DOS) for (a) TiCoSb (b) ZrCoSb and (c) HfCoSb.
59
acoustic branches with the optical branches as evident from
60

7
 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2 Page 8 of 14

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 The lower optical mode has maximum contribution from Thermal properties such as the constant volume heat
4 Ti/Zr atoms whereas the upper optical mode comes from Co capacity (CV), constant pressure heat capacity (CP), Entropy
5 atom. In HfCoSb, the acoustic mode has its main (S) and the coefficient of thermal expansion (α) are
6 contribution from Hf atom whereas Co and Sb atom calculated with respect to temperature, using the relations-
7 contributes to the optical mode.   
8   D  3 D 
Cv  3nk  4 D     D / TT 

pt
9 (16)
3.4 Thermodynamic Properties  1
10  T  e
 
11 The thermodynamic properties can be calculated fully ab
The Debye integral is defined by,
12 initio with precise accuracy using quasiharmonic Debye
  3 D /T x3
3 0
13 approximations. One of the non-negligible deficiencies of the D D   dx (17)
 T  ( D / T ) e 1

cri
x
14 harmonic approximation is the independency of the
15 temperature and geometry of the system. This means that in    D /T 
16 the frame work to this assumption, thermal expansion of a S  nkB  4 D  D   3ln(e  1)  (18)
 T  
17 system cannot be explained, as the distance between the
18  CV
atoms in a crystal is independent of temperature. The  (19)
19 quasiharmonic approximation in turn assumes the volume BT V
20 dependence of phonon frequencies, to achieve the volume BT is the isothermal bulk modulus and can be computed as

us
21 and temperature dependence such that the thermal expansion   2 G *(V ; P, T ) 
22 of a crystal can be predicted. In the framework of BT (T , P)  V   (20)
 V 2  P ,T
23 quasiharmonic approximation, from Debye’s model, the
24 ɤ is the Grüneisen parameter,
Gibbs function can be defined as [39]
25 d ln  D (V )
G *(V ; P, T )  E (V )  PV  AVib ( D (V ); T ) (10)   (21)
26 d ln V
27
28
29
E(V) is the total energy per unit cell, PV denotes constant
pressure,  D (V ) is the Debye temperature and AVib is the
vibrational parameter expressed by [63]
an and, CP  CV (1    T ) (22)

The temperature dependence of thermal properties are

30  9   
A Vib ( D , T )  nk BT   3ln(1  e  / T )  D  D   (11) studied in a wide range of temperature (0-1000 K) and the
31  8T  T  plots are shown in Fig. 8. The volume of the unit cell is
32
 
dM
33 found to increase with the increase of temperature, Fig.8 (a),
where, n is the number of atoms per unit cell and D  D  is
34 T  for all of MCoSb (M=Ti, Zr and Hf). This is usually
35 the Debye integral. The Debye temperature  D for an expected as lattice constant increases with the increase in
36 temperature. The entropy of the system increases with the
isotropic solid can be expressed by [64]
37 increase of temperature as evident from Fig.8 (d), thus the
 6 V n  B
1/3
38 D  1/ 2
f ( ) (12) Gibbs free energy of the system decreases when temperature
39 kB M increases. As the temperature tends toward absolute zero
40 M is the molecular weight per formula unit and B is the bulk (0K), the entropy tends toward a constant value and the
41 modulus which is assumed to be equal to the static bulk system MCoSb has almost same entropy values for
42 modulus as expressed by the relation temperatures below 25 K. The entropy is zero at the absolute
43 d 2 E (V )
pte

temperature. Thus, the change in entropy with temperature is

B  Bs t a t i c  V (13)
44 dV 2 in accordance with the third law of thermodynamics for HH
45 f ( ) is a function dependent on the Poisson’s ratio (σ) compounds MCoSb. Fig. 8 (g) gives the variation of thermal
46 defined as expansion coefficient (α) with temperature. α increases
47 1/3 almost linearly with the rise of temperature till 300 K, after
  3/ 2 1 
48  (1   )  
3/ 2
   2(1   )   300 K, the α value tends towards a constant value. The
49 f ( )  3 2 
   3(1  2 ) 
    (14)
 3(1   )     values of α calculated at room temperature (300 K) and
50   constant pressure (0 BPa) are 3.28x10-5 / K, 3.361x10-5 / K
51
ce

Thus, the Gibbs free energy is obtained with respect to and 3.417x10-5 / K respectively for TiCoSb, ZrCoSb and
52 volume at constant temperature and pressure as-
HfCoSb. Debye’s temperature (  D ) continuously decreases
53  G * (V ; P, T ) 
54  V  0 (15) with the increase of temperature, with a small variation. The
  P ,T
55 obtained  D values at 0 K and at 0 GPa pressure are very
From equation (15), the equilibrium volume curve V (T, P)
56 can be obtained, i.e. the thermal equation of state. The
close to the ones obtained from elastic result. At room
Ac

57 different thermal properties are now calculated with respect temperature, the values obtained for  D are 431.57 K, 392.13
58 to equation 15. K and 346.58 K. The decrease of  D with temperature
59
60 indicates the decrease of anharmonicity with the rise in

8
Page 9 of 14 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 temperature. The Grüneisen parameter (ɤ) decreases by very
4 small amount with rise in temperature. From temperature
5 range of 0 to 200 K, ɤ is almost constant with temperature at
6 a given pressure. The room temperature values of ɤ at 0
7 pressure are listed in Table 5. For compound MCoSb, the
8 temperature does not show much significant effect on the

pt
9 Grüneisen parameter. The specific heat at constant volume
10 CV increases linearly upto 200 K with the increase of
11 temperature, after 200K upto 800 K, CV increases slowly,
12 Fig. 8: Temperature dependence of (a) Volume (a.u)3, (b) Gibbs
with the variation becoming smaller and smaller as
13 free energy (kJ/mol), (c) Heat capacity at constant volume (J/mol
temperature rises, after 800K, CV attains a constant value. At

cri
14 K), (d) Entropy (J/mol K), (e) Debye Temperature (K), (f)
lower temperatures upto 200K, the specific heat at constant Grüneisen parameter, (g) Volumetric thermal expansion coefficient
15
volume CV obeys the Debye’s T3 law, at constant pressure. (K-1) and (h) Heat capacity at constant pressure (J/mol K) for
16
At higher temperature range (T>800K), the specific heat CV MCoSb (M= Ti, Zr and Hf).
17
follow the Dulong and Petie law, i.e it becomes independent
18 Table 5: Volume (a.u)3, Thermal expansion coefficient α
19 of temperature. The temperature variations for CV of MCoSb,
see Fig. 8 (c), shows the validation of the two important laws x10-5 (K-1), heat capacity at constant volume CV (J/mol K),
20

us
of thermodynamics, thus giving a measure of reliability of heat capacity at constant pressure CP (J/mol K), Entropy S
21
the calculated thermodynamic results. The specific heat (J/mol K) and the Grüneisen parameter (ɤ) for MCoSb at
22
capacity at constant pressure (CP) also shows similar room temperature (300 K).
23
24 characteristics as CV, except that at high temperature range,
Compd. Vol α CV CP S ɤ
25 CP is less influenced by temperature but not completely
26 independent, Fig.8 (h). The trends shown by CV and CP are TiCoSb 347.76 3.28 67.63 69.05 76.33 2.13
27
28
29
shows the calculated values of the thermodynamic
parameters at room temperature. Unfortunately, no
an
very similar to that reported by Gaili Sun et. al [29]. Table 5 ZrCoSb
HfCoSb
385.78
379.30
3.36
3.41
68.81
70.06
70.38
70.38
82.84
91.41
2.26
2.30
30 theoretical or experimental values on the thermodynamic
31 properties of MCoSb (M= Ti, Zr and Hf) has been reported 3.5 Thermoelectric Properties
32 earlier for comparison, to the best of our knowledge.
dM
33 In this section, the thermoelectric properties of HH
34 compound MCoSb are presented. We have calculated the
35 Seebeck coefficient (S), the electrical conductivity (σ),
36 thermal conductivity (κ) and the thermoelectric figure of
37 merit (ZT). These parameters are sufficient to characterize
38 the efficiency of a material for thermoelectric energy
39 applications and are calculated from the semi classical
40 transport equation of Boltzmann using rigid band
41 approximation [65]. According to this approximation, the
42 band structure of a material remains unchanged with the
43
pte

application of temperature, changing only the chemical

44 potential of the material. This approximation holds good for
45 low doping levels and are found to yield results close to the
46 experimental results [41]. The band structure is directly
47 related to the Seebeck coefficient of the material and the
48 Seebeck coefficient to the other thermoelectric parameters.
49 Except the Seebeck coefficient, the electrical conductivity
50 and the electronic thermal conductivity (κe) obtained from
51
ce

BoltzTraP are found to be dependent on the relaxation time

52 (τ). Therefore, the dependence of the relaxation time (τ) on
53 temperature does not affect the dependence of Seebeck
54 coefficient on temperature.
55
56
Ac

57
58
59
60

9
 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2 Page 10 of 14

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 ekB       f 0 (T ,  ,  ) 
4 S
  i, k  
 k BT     d  (27)

5
6 The calculated electrical conductivity and the electronic
7 thermal conductivity from BoltzTraP are expressed in the
8 ratio σ/τ and κe /τ. For better computation purpose, a dense k-

pt
9 mesh of 50,000 k points were used. The results obtained
10 from the first principle calculations using WIEN2k code acts
11 as the key input for the calculation of thermoelectric
Fig. 9: Variation of relaxation time as a function of temperature in
12 properties. The following figure 10, shows the plot of
TiCoSb, ZrCoSb and HfCoSb half Heusler compounds.
13 different thermoelectric parameters as a function of chemical

cri
14 However, for accurate determination of the transport potential (  ) at different temperatures. They are plotted in
15 properties it is important to account for the relaxation time. It the chemical potential range of 0.58569 Ry - 0.96469 Ry,
16 is calculated using the Bardeen and Shockley’s deformation 0.475 Ry – 0.825 Ry and 0.525 Ry -0.875 Ry respectively
17
potential theory using Eq.1, as discussed in the methodology for TiCoSb, ZrCoSb and HfCoSb at different temperatures.
18
section. Our calculation shows that the relaxation time of In Fig.10 (a), the negative trend in the Seebeck coefficient
19
MCoSb (M=Ti, Zr and Hf) decreases with the increase in represents n-type carriers, while the positive trend represents
20

us
21
temperature and for higher temperature ranges, it tends p-type. For each temperature range from 100 K – 1000 K,
towards a constant value. The plot is shown in the following distinct positive trend of the Seebeck coefficient indicates the
22
23 Fig.9. Table 6 shows the calculated values of C 3D and m* involvement of only p-type of charge carriers for the range of
24 for TiCoSb, ZrCoSb and HfCoSb. chemical potential below Fermi level (EF). When the
25 chemical potential (μ) becomes equal to EF, the contribution
26 Table 6: Calculated values of C 3D and m* used in the of both the p and n type charge carriers becomes maximum.
27
28
29
calculation of τ.
an As the μ value increases from EF, only n-type charge carriers
contributes to S. This is beneficial for a thermoelectric
material as it ensures high S value. Mixed n and p-
TiCoSb ZrCoSb HfCoSb
30 contribution cancels the induced Seebeck voltage thus
C (GPa)
3D
254.8 263.0 257.2
31 decreasing the S value. The DOS peaks near the band edges
32 m*(me) 1.48 2.01 1.62 and the band gap affects the Seebeck coefficient of the
dM
33 material and thus the thermoelectric properties. The flat
34 valence and conduction band near the Fermi level in the band
The solution of the semi-classical Boltzmann equations
35 structure prevents the Seebeck coefficient from attaining
are obtained by performing Fourier interpolation of the band
36 high values. The band degeneracy and the bands along the L-
37 energies which yields the conductivity tensor and the other
Γ direction in the valence band are responsible for positive
38 transport coefficients. The electrical conductivity tensor is
values of S, while that along the Γ-X direction in the
39 expressed as [65, 66]
conduction band contributes to negative S values. The band
40
 f 0 (T ,  ,  )  degeneracy is beneficial for high Seebeck coefficient. The S
41   ,   e2         k (23)
   value decreases with respect to chemical potential as the
42 i, j
temperature increases. It is seen that both σ/τ and κe/τ are
43 Where, α and β tensor indices, να and νβ are the group
pte

higher for those range of chemical potential which is greater

44 velocities, τκ is the relaxation time, e is the electronic charge
than EF. Thus, doping MCoSb with n-type impurities will
45 and f0 is the Fermi- Dirac distribution function. The electrons
46 contribute near the chemical potential (  ) in a narrow range contribute more towards the electrical and electronic thermal
conductivity. Unlikely to S, here both the electrical and
47 of   kBT      kBT . The transport distribution can then
thermal conductivity tends towards zero as the values of
48 be expressed as the kernel of the transport coefficients given chemical potential approaches near EF. κe/τ increases with the
49 by-
increase of chemical potential and temperature whereas σ/τ
50 i , k   i , k    k (24) decreases with the increase of temperature, except for those
51
ce

52 Thus, the electrical conductivity (σ), the electronic thermal ranges of μ near to EF. The decrease in electrical conductivity
53 conductivity (κe) and the Seebeck coefficient (S) can be is due to the increase in charge carriers concentration and the
54 obtained as a function of chemical potential and temperature, mobility near EF. With the increase in temperature, scattering
55  f (T ,  ,  )  becomes strong leading to high thermal conductivity and
  e2  i , k   0  d (25)
56    hence reducing the electrical conductivity. Therefore, in
Ac

57 2 general, Fig.10 reveals that the material MCoSb will be more

      f 0 (T ,  ,  ) 
58  e  k T  i , k 
2
   d (26) efficient when doped with n-type impurity and will have

B
59  k BT    maximum efficiency in the chemical potential range above
60 EF, where the electronic thermal conductivity is minimum.

10
Page 11 of 14 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 We have calculated the lattice part of thermal conductivity Fig.10: Plot of thermoelectric parameters with respect to chemical
4 (κl) by solving the Slack’s model expressed by Eq.4. The potential (μ) for TiCoSb, ZrCoSb and HfCoSb at different
5 calculated results of (κl) are plotted against the temperature temperatures. All (a) in the figure represents Seebeck coefficient,
6 in Fig. 11. The parameter A present in Slack’s equation (Eq. (b) The relaxation time dependent electrical conductivity and (c) the
7 4) is a constant dependent upon ɤ given by [67]-
relaxation time dependent electronic thermal conductivity.
8
5.720 107  0.849

pt
9 A( )  (28)
10 2 1  (0.514 /  )  (0.288 /  2 ) 
11 The lattice thermal conductivity of MCoSb compound
12 increases in the order TiCoSb>ZrCoSb>HfCoSb. The
13 calculated values of thermal conductivity are plotted in

cri
14 Fig.11 (a) with respect to temperature. Andrea et. al. [68] had
15 calculate κp for ZrCoSb by using the many body perturbation
16 theory and the Boltzmann transport equation, in the Fig. 11: (a) lattice thermal conductivity (κp) of MCoSb (M=Ti, Zr
17 temperature range of 300K-900K. We have compared our and Hf) (b) Comparison of κp for ZrCoSb with the available
18 theoretical result.
results with that obtained by Andrea et. al. and are presented
19
in Fig.11 (b). In the temperature range of 300K-700K our
20 Further to get the ZT results, two methods can be chosen.

us
calculated values are bit higher than the compared theoretical
21 One can calculate ZT with respect to temperature at constant
results.
22 chemical potential or can calculate ZT with respect to
23 constant carrier concentration. In this present work, we will
24 use the later method, our previous works are reported using
25 the former method [69]. The following Fig.12 shows the plot
26
27
28
29
an of Seebeck coefficient at different carrier concentration of n
and p-type charge carriers, with temperature. Unlike with
chemical potential, when Seebeck coefficient is varied with
carrier concentration, it is found to increase with temperature
30
except for the carrier concentration of 10 19 cm-3, where the
31
carrier concentration decreases from 900K, (for both n and p-
32
dM
33 type doping). At higher values of temperature, Seebeck
34 coefficient is expected to decrease because of increasing
35 presence of thermally excited charge carriers.
36 The Seebeck coefficient has been plotted in the carrier
37 concentration range of  1017 cm-3 to  1022 cm-3. Even
38 though the charge carrier concentration of  1017 and  1018
39 shows high S values, it can be seen from figure 12 that at
40 high temperatures, S decreases abruptly with increasing
41 temperature. Also, above 900 K, mixed n and p-type
42 contribution is observed in this concentration range.
43
pte

Therefore, we have chosen 1019 cm-3 as the optimised

44
concentration value for which S is maximum at higher
45
temperatures. The positive trend of the Seebeck coefficient is
46
higher than the negative trend, therefore, p-type charge
47
48 carriers dominates the charge transport in HH compound
49 MCoSb. This indicates the fact that the maximum
50 thermoelectric ZT value is achievable for p-type case at a
51 carrier concentration of 1019 cm-3 and will play an important
ce

52 role to achieve high ZT values at high temperatures. Yang et.

53 al. [55] have also reported the high power factor of these
54 materials for p-type than the n-type charge carriers.
55 Therefore, our further results are reported on this
56 concentration. The highest value of the Seebeck coefficient
Ac

57 observed are respectively 675 (μV/K), 654 (μV/K) and 672

58 (μV/K) at 800 K for Ti, Zr and Hf of MCoSb. Since, the
59 band gap values of TiCoSb, ZrCoSb and HfCoSb are close to
60 each other, hence their S values are also close to each other.

11
 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2 Page 12 of 14

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 ZrCoSb has the lowest S value in comparison to the other
4 MCoSb compounds. This is because of the reduced
5 degeneracy leading to enhanced effective mass of the carriers
6 which results due to the band gap originating between L-X
7 symmetry directions. S value decreases with the increase in
8 concentration from 1019 cm-3. Similar decrease in S with the

pt
9 increasing value of concentration was reported by Lee et. al.
10 [57]. When the concentration level was increased above
11 ±1022 cm-3, the charge flow ceased due to the over population
12 of the Fermi level. The value of S is limited by the slopes of
13
the DOS near the Fermi energy which indicates high

cri
14
effective mass of holes in the upper valence band region. The
15
high Seebeck coefficient at 800 K indicates good
16
thermoelectric performance at higher temperatures.
17
18
19 Fig. 13: Thermoelectric parameters (a) Electrical conductivity, (b)
20 Electronic thermal conductivity, (c) Total thermal conductivity of

us
21 TiCoSb, (d) ZrCoSb, (e) HfCoSb and (f) the thermoelectric figure
22 of merit ZT plotted with respect to temperature. The calculated
23 lattice thermal conductivity is compared with the experimental
results of Sekimoto et. al. [23, 70] in a temperature range of 400 K-
24
1000 K.
25
26
27
28
29
values of carrier concentration for (a) TiCoSb (b) ZrCoSb and (c)
HfCoSb compounds.
an
Fig. 12: Seebeck coefficient with respect to temperature at different
The electronic thermal conductivity (κe), is very low when
compared with the lattice thermal conductivity and is found
to decrease with temperature in the range 300 K to 900 K,
see Fig.13 (b). This indicates phase change from
30
semiconductor to metal in this temperature range, as the
31 The variation of electrical conductivity (σ), total thermal
thermal conductivity decreases with the increasing
32 conductivity (κ) and the thermoelectric figure of merit (ZT) is
dM
temperature only in metals. This change in phase is indeed
33 shown in Fig.13. The obtained thermal conductivity results
the key for high electrical conductivity in these compounds.
34 are compared with the available experimental results and it
35 can be seen that our calculated values of κ are in close However at temperatures above 2000 K, it is expected that κe
36 agreement to the available experimental data. The total will be significantly high and will contribute solely to the
37 thermal conductivity decreases in the higher temperature total thermal conductivity as κl will be negligibly small in
38 regions, representing the thermoelectric applicability of these those temperature ranges. Unfortunately no experimental
39 materials in higher temperature domains. The thermal data are available to compare with the obtained κe results and
40 conductivity value calculated at room temperature are the available theoretical results on κe lack the computation of
41 presented in table 7. Our calculated value of electrical relaxation time. The ZT values for these materials are much
42 conductivity and the figure of merit ZT is higher than the lower than the benchmark value (~1), however, owing to the
43 high Seebeck coefficient values and the low thermal
pte

available experimental results, this may be because of the

44 optimized carrier concentration in our results, which showed conductivity, MCoSb would be a promising thermoelectric
45 optimum thermoelectric performance for p-type MCoSb, material on further optimization of the thermoelectric
46 while the experiments report for n-type characteristics of the
parameters. It was also found that decreasing the carrier
47 concentration increased the S value and decreased the
compound. The highest ZT values were observed at 800K in
48 thermal conductivity, but it also decreased the electrical
TiCoSb (0.29) and ZrCoSb (0.24). HfCoSb had the highest
49 conductivity. Therefore, the numerator in the ZT equation
ZT at 900K (0.23). As mentioned, our ZT values are about
50 was not able to attain high values. However, if MCoSb is
ten times higher than those reported by Sekimoto et. al.
51
ce

doped with impurities which are a good conductor of

(2005) [23]. However, authors have reported higher values of
52 electrical conductivity, such that σ would not become too
53 ZT~0.5, when doped with suitable impurities [70].
less, it would ensure high numerators in the ZT equation and
54 ultimately increase the ZT values.
55
56
Ac

57
58
59
60

12
Page 13 of 14 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 Table 7: Values of different thermoelectric parameters S Acknowledgements
4 (μV/K), σ x 104 (Wm)-1, ke (W/mK), kl (W/mK), k (W/mK)
5 ZT calculated at room temperature. HJ and RKT acknowledge research project grant from
6 SERB (Govt. of India) EMR /2015/001407.
7 Comp. S σ κe κp κ ZT One of the authors A. Laref acknowledge the ”Research
8 TiCoSb 523.80 6.636 5.9 28.25 34.15 0.16 Center of Female Scientific and Medical Colleges, King

pt
9 Saud University” for financial support.
10 ZrCoSb 518.29 3.806 3.3 23.06 26.33 0.12
11 HfCoSb 559.28 3.089 3.06 20.42 23.48 0.12 References
12
13 3. Conclusion [1] R. A. de Groot, F. M. Mueller, P. G. van Engen, and K. H. J.

cri
14 Buschow, Phys. Rev. Lett. 50, 2024 (1983).
15 [2] P. J. Webster and K. R. Ziebeck, Alloys and Compounds of
Our main aim was to study the different properties of HH
d-elements with Main Group Elements. Part 2 edited by H. R. J
16 compound MCoSb not yet reveled theoretically. In order to Wijn, Landolt-Bornstein, New series Group III, Vol. 32/c, Springer
17 achieve this, we used the FP-LAPW method based upon the Berlin 64-414, (2001).
18 density functional theory along with quasiharmonic [3] F. G. Aliev, Physics B, 171, 199 (1991).
19 approximation and Boltzmann transport theory. The [4] Fanjie Kong, Yanfei Hu, Haijun Hou, Yanhua Liu, Baolin
20 structural and the electronic parameters such as the lattice Wang and Lili Wang, JPCS, 196, 511–517 (2012).

us
21 constant and the energy band gap are in agreement with the [5] Shakeel Ahmad Khandya, Ishtihadah Islamb, Dinesh C.
22 available literature. We found that HH compound MCoSb is Gupta and Amel Laref, Journal of Solid State Chemistry, 270, (173-
23 179), 2019.
more stable and has greater probability to be synthesized
24 [6] Shakeel Ahmad Khandya and Dinesh C. Gupta,
experimentally (corresponding to negative formation energy)
25 Semiconductor Science and Technology, 32 (12), 125019 (2017).
when we considered the position M; x=y=z=0.25, Co;
26 [7] I. Galanakis, P. Mavropoulos, P. H. Dederichs, J. Phys.D;
27
28
29
x=y=z=0 and Sb; x=y=z=0.75, rather than considering the
general position of HH compounds which is M; x=y=z=0,
Co; x=y=z=0.25 and Sb; x=y=z=0.5. The modified Becke-
an Appl. Phys, 39 (5) (2006).
[8] H. Shi, D. Parker, M. H. Du and D. J. Singh, Phys. Rev. Appl.
3, 014004 (2015).
Johnson potential is almost ineffective to compute the band [9] H. Osterhage, J. Gooth, B. Hamdou, P. Gwozdz, R. Zierold
30
gaps of the compounds and thus the customary GGA and K. Nielsch, Appl. Phys. Lett. 105, 123117 (2014).
31 [10] S. A. Khandy, I. Islam, D. C. Gupta, A. Laref, Int J
32 functional yields appreciable band structure related results.
dM
Energy Res. 1–8 (2018), DOI: 10.1002/er.4182.
33 The phonon calculations strongly supports the indication that
[11] Kulwinder Kaur, D. P. Rai, R. K. Thapa and Sunita
34 MCoSb is mechanically and thermodynamically stable as no Srivastava, J Appl Phys, 122, 045110 (2017).
35 negative frequencies were observed in the phonon dispersion [12] G. J. Snyder and E. S. Toberer, Nat. Mater, 7, 105 (2008).
36 curve and the stability increases from TiCoSb  HfCoSb  [13] D.P.Rai, Amel Laref, A.Shankar, Sandeepd, Anup
37 ZrCoSb. Thermodynamic properties analysis showed the Pradhan Sakhya, R.Khenata and R.K.Thapa, J. Phys. Chem. Solids,
38 decrease of Debye’s temperature with the rise in temperature. 120, 104-108 (2018).
39 This indicates that at higher temperatures the compound [14] L. D. Zhao, G. J. Tan, S. Q. Hao, J. Q. He, Y. L. Pei and
40 MCoSb becomes less anharmonic and thus tends towards H. Chi, Science, 351, 141 (2016).
41 harmonic oscillations. Finally, the high Seebeck coefficient [15] S. I. Kim, K. H. Lee, H. A. Muu, H. S. Kim, S. W. Hwang
42 and J. W. Roh, Science, 348, 109 (2015).
and the low thermal conductivity showed the carrier
[16] A. Weidenkaff, R. Robert, M. H. Aguirre, L. Bocher, T.
43
pte

concentration is optimized at 1019 cm-3 and p-type MCoSb Lippert and S. Canulescu, Renewable Energy, 33, 342 (2008).
44 will be more efficient than the n-type MCoSb HH [17] P. Tomes, C. Suter, M. Trottmann, A. Stenfeld and A. J.
45 compounds. The lattice thermal conductivity are in close Weidenkass, Mater. Res., 26, 1975 (2011).
46 agreement with the experimental results. Our calculation is [18] K. Kaur, S. Dhiman and R. Kumar, Phys. Lett. A, 381, 399
47 done computing the variation of relaxation time with (2017).
48 temperature, which many theoretical studied lack. The [19] S. R. Clup, J. W. Simonson, S. J. Poon, V. Ponnambalam, J.
49 calculated ZT value, even though not high, is still higher than Edwards and T. M. Tritt, Appl. Phys. Lett., 93, 022105 (2008).
50 the available experimental results and we believe that [20] D. P. Rai, A. Shanker, Sandeep, M. P. Ghimire, R.
51
ce

Khetana and R. K. Thapa, RSC Adv, 5, 95353 (2015).

optimizing the transport parameters like Seebeck coefficient,
52 [21] Y. Xia, S. Bhattacharya, V. Ponnambalam, A. L. Pope, S.
electrical and thermal conductivity will further reveal high
53 J. Poon, T. M. Tritt, Journal Of Applied Physics, 88, 4 (2000)
ZT values. This will add a strong support on our prediction
54 [22] Yu. Stadnyk, Yu. Gorelenko, A. Tkachuk, A. Goryn, V.
that MCoSb has possible thermoelectric applications. Davydov and O. Bodak, J. Alloys and Compd. 329, 37–41 (2001)
55
[23] T. Sekimoto, K. Kurosaki, H. Muta, and S. Yamanaka,
56
Conflicts of interest
Ac

IEEE- 24th International Conference on Thermoelectrics (2005).

57
[24] Takeyuki Sekimoto, Ken Kurosaki, Hiroaki Muta and
58
There are no conflicts to declare. Shinsuke Yamanaka, Materials Transactions, 46, 7 (2006).
59
60

13
 AUTHOR SUBMITTED MANUSCRIPT - MRX-112122.R2 Page 14 of 14

Journal XX (XXXX) XXXXXX Joshi et al

1
2
3 [25] Jiong Yang, Huanming Li, Ting Wu, Wenqing Zhang, [55] Jiong Yang, Huanming Li, Ting Wu, Wenqing Zhang,
4 Lidong Chen and Jihui Yang, Adv. Funct. Mater., 18, 2880–2888, Lidong Chen and Jihui Yang, Adv. Funct. Mater., 18, (2008).
5 (2008). [56] Bo Yuan, Bo Wang, Lihong Huang, Xiaobo Lei, Lidong
6 [26] Mal-Soon Lee, Ferdinand P. Poudeu and S. D. Mahanti, Zhao, Chao Wang and Qinyong Zhang, Journal of Electronic
7 Phy. Rev. B, 83, 085204 (2011). Materials, 46, 5 (2017).
8 [27] Junichiro Shiomi, Keivan Esfarjani and Gang Chen, Phy. [57] Mal-Soon Lee, Ferdinand P. Poudeu and S. D. Mahanti,

pt
Rev. B, 84, 104302 (2011). Phy. Rev. B, 83, 085204 (2011).
9
[28] C. Coban, Y. O. Ciftci and K. Colakoglu, Indian J Phys, [58] C Coban, Y O Ciftc and K Colakoglu, Indian J Phys., 90,
10
90, 11 (2016). 11 (2016).
11 [29] Gaili Sun, Yuanyuan Li, Xinxin Zhao, Yiming Mi1 and Lili [59] Z.W. Huang, Y.H. Zhao, H. Hou and P.D. Ha, Physica B,
12 Wang, Journal of Materials Science and Chemical Engineering, 3 407 (2012).
13 (2015). [60] A. Yakoubi, O. Baraka and B. Bouhafs, Results in

cri
14 [30] P. Hohenberg and W. Kohn, Phy. Rev., 136 (1964). Physics, 2 (2012).
15 [31] W. Kohn and L. J. Sham, Phy. Rev., 140 (1965). [61] Jianhua Ma, Vinay I. Hegde, Kamaram Munira, Yunkun
16 [32] P. Blaha, G.K.H. Madsen, D. Kvasnicka, J. Luitz, Xie and Sahar Keshavarz, Phys. Rev. B, 95, 024411 (2017).
17 WIEN2K, an augmented plane wave plus local orbitals program for [62] J. Heyd, G. E. Scuseria and M. Ernzerhof, J. Appl. Phys.,
18 calculating crystal properties (Vienna, Austria) 2008. 118, 8201 (2003).
19 [33] J. Perdew, K.P. Burke and M. Ernzerhoff, Phys. Rev. Let., [63] M Flrez, J M Recio, E Francisco, M A Blanco and A M
20 77, 3865 (1996). Penda, Phy. Rev. B, 664112 (2002).

us
21 [34] C. S. Wang, W. E. Pickett, Phy. Rev. Lett., 51, 597 (1983). [64] G K H Madsen, P Blaha, K Schwarz, E Sjo ¨stedt and L
22 [35] R. O. Jones and O. Gunnarsson, Rev. Mod. Phys., 61, 689 Nordstrom, Phys. Rev. B, 64, 195134 (2001).
(1989). [65] L. Chaput, P. P´ echeur, J. Tobola, and H. Scherrer, Phys.
23
[36] S. Baci, B. G. Yalcin, H. A. Rahnamaye Aliabad, S. Rev. B., 72, 085126 (2005).
24
Duman and B. Salmankurt, RSC Adv., DOI 10.1039/C6RA02981G [66] H. M. A. Yamamoto and T. Takeuchi, J. Appl. Phys., 115,
25 (2016). 023708 (2014).
26
27
28
29
[37] Atsushi Togo and Isao Tanaka, Scripta Materialia, 108
(2015).
[38] Atsushi Togo and Isao Tanaka, Phy. Rev. B, 87, 184104
an [67] C.L. Julian, Phys. Rev., 137, A128 (1965).
[68] L. Andrea, G. Hug, L. Chaput, ONERA-DLR Aerospace
Symposium, Jun 2014, COLOGNE, Germany.
(2013). [69] H Joshi, D P Rai, E Deligoz, H B Ozisik and R K Thapa,
30 [39] M A Blanco, E Francisco and V. Luana, Comput. Phys. Mater. Res. Express., 4, 105506 (2017).
31 Comm., 158, 57 (2004). [70] Takeyuki Sekimoto, Ken Kurosaki, Hiroaki Muta and
32 [40] A Otero-de-la-Roza, D Abbasi-Pe´rez and V Luana, Shinsuke Yamanaka, Japanese Journal of Applied Physics, 46, 27
dM
33 Comput. Phys. Comm., 182, 2232 (2011). (2007).
34 [41] G K H Madsen and D J Singh, Comput. Phys. Commun.,
35 17 (2006).
[42] J. Bardeen and W. Schockley, Phy. Rev., 80 (1950).
36
[43] J. Xi, M. Long, L. Tang, D. Wang and Z. Shuai
37
Nanoscale, 4, 4348 (2012).
38 [44] Saurabh Singh, R K Maurya and Sudhir K Pandey, J.
39 Phys. D: Appl. Phys. 49, 425601 (2016).
40 [45] D. T. Morelli and G. A. Slack, High Lattice Thermal
41 Conductivity of Solids, Springer, New York, USA, (2006).
42 [46] Yu. Stadnyk, Yu. Gorelenko, A. Tkachuk, A. Goryn, V.
43
pte

Davydov and O. Bodak, J. Alloys and Compd., 329 (2001).

44 [47] L. L. Wang, L. Miao, Z. Y. Wang, W. Wei, R. Xiong, H.
45 J. Liu, J. Shi, and X. F. Tang, Journal of Applied Physics, 105,
46 013709 (2009).
47 [48] G. Melnyk, E. Bauer, P. Rogl, R. Skolozdra and E. Seidl,
48 J. Alloys and Compd., 296, (2000).
49 [49] Laila Offernes, P. Ravindran and A. Kjekshus, J. Alloys
and Compd., 439 (2007).
50
[50] Kulwinder Kaur and Rajan Kumar, Creamics
51
ce

International, 43, (2017).

52 [51] F. D. Murnaghan, Proc. Natl. Acad. Sci., USA 30, 5390
53 (1994).
54 [52] Bo Yuan, Bo Wang, Lihong Huang, Xiaobo Lei, Lidong
55 Zhao, Chao Wang and Qinyong Zhang, Journal of Electronic
56 Materials, 46, 5 (2017).
Ac

57 [53] Philipp Haas, Fabien Tran and Peter Blaha, Phy. Rev. B,
58 79, 085104 (2009).
59 [54] Agostino Zoroddu, Fabio Bernardini and Paolo
60 Ruggerone, Phy. Rev. B, 64, 045208 (2001).

14