(Sang Jun Park)
1
(Il-Ho Kim)
1*
Copyright © 2023 The Korean Institute of Metals and Materials
Key words(Korean)
thermoelectric, skinnerite, mechanical alloying, hot pressing, doping
1. INTRODUCTION
Thermoelectric energy conversion technology offers a solution to the energy crisis,
as it can directly convert heat into electricity and harvest thermal energy from low-grade
heat sources [1-4]. The thermoelectric conversion efficiency depends on the performance of the thermoelectric
material and the Carnot efficiency of the power generation system. The thermoelectric
figure of merit is expressed as Z = α2σκ-1 [5], where the numerator term α2σ is the power factor (α: Seebeck coefficient, σ: electrical conductivity, and κ:
thermal conductivity). Because all three parameters (α, σ, and κ) are temperature-dependent,
the dimensionless index (ZT) is used to evaluate the thermoelectric properties by
multiplying the absolute temperature (T). Therefore, to improve the performance of
thermoelectric materials, it is necessary to increase their power factor at the application
temperature and/or decrease their thermal conductivity.
Recently, Cu-Sb-S compounds have been attracting interest as eco-friendly and cost-effective
energy materials. Four types of compounds are candidates for p-type thermoelectric
materials: Cu3SbS4 (famatinite), Cu12Sb4S13 (tetrahedrite), Cu3SbS3 (skinnerite), and CuSbS2 (chalcostibite). Except for Cu3SbS4, these compounds are known to have low lattice thermal conductivity because the Cu
atoms vibrate because of the unshared electron pairs (lone-pair electrons) of Sb.
Among the Cu-Sb-S compounds, Cu3SbS3 has the largest SSb-S bonding angle, and thus the lowest lattice thermal conductivity
[6]. In addition, it has a high absorption coefficient and a suitable bandgap energy
(1.40−1.84 eV) for photovoltaic applications [7-10]. However, there have been few studies on the thermoelectric performance of Cu3SbS3.
Skinnerite is known to have three crystal structures [11,12]: orthorhombic (space group P212121, <263 K), monoclinic (space group P21/c, 263–395 K), and orthorhombic (space group Pnma, >395 K). These structural phase
transitions arise because of the disorderly motion of Cu atoms as the temperature
increases. Skinnerite synthesized by the conventional melting method has a monoclinic
structure at room temperature, but when the temperature is elevated to 395 K or higher,
the phase is transformed from monoclinic to orthorhombic; thus, its application as
a thermoelectric material may be limited [13]. In addition to these three known structures, skinnerite may form a cubic structure,
which is predicted to be similar to tetrahedrite (Cu12Sb4S13 or Cu3SbS3.25) [14]. Cubic skinnerite maintains lone-pair electrons in a tetrahedron at the Sb site
but has no S atom at the center of the unit cell and is missing a S atom at the CuS3
triangular vertex. This structure is unstable, which can have a significant effect
on the vibrational mode of the triangulated Cu atoms and Sb atoms with lone-pair electrons
[13]. Du et al. [14] stabilized the cubic skinnerite phase by partially replacing Sb with Fe. The effect
of Fe doping on the crystal structure has also been reported for Cu-Fe-Sb-S quaternary
systems [15,16].
In our previous studies [17,18], we synthesized a cubic skinnerite Cu3SbS3 phase stable at 323−623 K using mechanical alloying (MA) followed by hot pressing
(HP), and a good thermoelectric performance (ZT = 0.51 at 623 K) was obtained. In
this study, we prepared Fe-doped Cu3Sb1-yFeyS3 using the MA-HP process and evaluated the effect of the Fe doping level on the thermoelectric
properties.
2. EXPERIMENTAL PROCEDURE
To synthesize Cu3Sb1-yFeyS3 (y = 0.02, 0.04, and 0.06), elemental powders of Cu (purity 99.9%, <45 μm), Sb (purity
99.999%, <75 μm), Fe (purity 99.9%, <53 μm), and S (purity 99.99%, <75 μm) were weighed
to stoichiometric compositions, and the mixed raw powders and stainless-steel balls
were loaded into a hardened steel jar. MA was conducted at 350 rpm for 18 h in an
Ar atmosphere using a planetary ball mill (Fritsch, Pulverisette5). The synthesized
powder was loaded into a graphite mold and HP was performed under a pressure of 70
MPa at 623 K for 2 h. The optimal MA-HP process conditions were determined in our
previous studies [17,18].
The phases of the MA powder and HP sintered bodies were analyzed using X-ray diffraction
(XRD; Bruker, D8-Advance) with Cu Kα radiation. The Rietveld analysis program (TOPAS)
was used to calculate the lattice constants. The microstructure of the sintered specimens
was observed using scanning electron microscopy (SEM; FEI, Quanta400) in the backscattered
electron mode (BSE). Elemental line scanning and mapping analyses were performed according
to the energy level of each element using energy-dispersive Xray spectroscopy (EDS;
Bruker, XFlash4010). Carrier concentration and mobility were evaluated by measuring
the Hall coefficient using the van der Pauw method (Keithley 7065). The Seebeck coefficient
and electrical conductivity were measured using ZEM-3 equipment (Advance Riko) in
a He atmosphere, and the thermoelectric power factor was examined. The thermal diffusivity,
specific heat, and density were measured using a TC-9000H system (Advance Riko) in
a vacuum atmosphere, and the thermal conductivity was estimated. The temperature dependence
of the thermoelectric characteristics was examined over the temperature range of 323–623
K using the measured Seebeck coefficient, electrical conductivity, and thermal conductivity.
Finally, the dimensionless figure of merit was evaluated.
3. RESULTS AND DISCUSSION
Figure 1(a) shows the XRD patterns of the Cu3Sb1-yFeyS3 powders synthesized by MA. Cubic skinnerite (space group I m: PDF#01-075-1574) was
obtained after MA at 350 rpm for 18 h. The MA powder was confirmed to have a single
cubic skinnerite phase. Figure 1(b) shows the XRD patterns of the Cu3Sb1-yFeyS3 specimens sintered by HP. The cubic skinnerite phase remained even after HP at 623
K under 70 MPa, without phase decomposition or phase transitions. The lattice constants
calculated by Rietveld analysis are summarized in Table 1, confirming the substitution of Fe at the Sb sites. The lattice constant of undoped
cubic Cu3SbS3, reported by Lee [17,18] was 1.03740 nm. However, in this study, the lattice constant was significantly reduced
by Fe doping. This is because the ionic radius of Fe is smaller than that of Sb. Shannon
[19] reported ionic radii of 90, 76, and 63 pm for Sb3+, Fe2+, and Fe3+, respectively. Zhang et al. [13] reported that Fe (117 pm) had a smaller covalent radius than Sb (140 pm). In this
study, as the Fe content increased, the lattice constant decreased from 1.03370 to
1.03310 nm, indicating the successful substitution of Fe at the Sb sites.
For comparison, the data [17,18] for undoped Cu3SbS3 are shown in Table 1 and Figs 1–9. Du et al. [14] detected a secondary phase of Cu2-xS and could not obtain single-phase cubic Cu3SbS3 by heat treatment without the addition of a proper dopant, although cubic Cu3SbS3 was synthesized by MA and spark plasma sintering (SPS). Pfitzner [12] synthesized stable orthorhombic Cu3SbS3 (space group Pnma) at temperatures higher than 394 K by vacuum melting of Cu2S and Sb2S3, but cubic Cu3SbS3 could not be synthesized. Daniel et al. [20] reported that a Cu3SbS3 thin film produced by thermal evaporation had a monoclinic structure (space group
P21/c). Even after annealing at temperatures up to 573 K, the crystallinity improved
but there was no change in the crystal structure. Wang et al. [21] reported that Cu3SbS3 nanowires synthesized by a solvothermal route exhibited an orthorhombic structure.
Atri et al. [22] obtained cubic skinnerite (space group I m) via co-thermal deposition and wet-chemical
synthesis for nonstoichiometric (Cu-poor or Cu-rich) compositions. However, they found
the crystal structure changed to orthorhombic in the case of Cu3Bi0.5Sb0.5S3, when Bi was substituted at the Sb site, due to the expansion of the unit cell since
Bi3+ has a larger ionic radius than Sb3+.
Figure 2 presents SEM photographs and EDS analysis results of the sintered specimens of Cu3Sb1-yFeyS3. High sintered densities (relative densities) of 98.3%–98.8% were obtained (Table 1) compared to the theoretical density (5.1 gcm-3) of Cu3SbS3 [12]. Figure 2(a) shows the polished surfaces of Cu3Sb0.98Fe0.02S3 with the elemental line scans. Figures 2(b)−2(d) show the fractured surfaces of sintered Cu3Sb1-yFeyS3 (y = 0.02−0.06). No specific changes in the microstructure are observed with respect
to the Fe content. Figures 2(e)−2(h) present the elemental distributions in the specimen with y = 0.02 and show that all elements are homogeneously distributed.
Figure 3 shows the charge transport properties (carrier concentration and mobility) of Cu3Sb1-yFeyS. Hall coefficient measurements confirmed that all specimens were p-type semiconductors
with positive Hall coefficients; this data is not presented. Cu vacancies with low
formation energies are easily formed in Cu chalcogenide compounds and contribute to
the p-type conduction behavior [23]. Undoped cubic Cu3SbS3 has a carrier (hole) concentration of (3.3−3.9) × 1018 cm-3 and mobility of 146−153 cm2V-1s-1 [17,18]. However, in this study, Fe doping increased the carrier concentration to (1.2−1.7)
× 1019 cm-3, and the mobility was in the range of 248−392 cm2V-1s-1 on average. In skinnerite, Sb exists as Sb3+, and the dopant Fe may exist as Fe2+ or Fe3+. Considering the increase in the hole concentration due to Fe doping, we conclude
that Sb3+ is substituted with Fe2+. Zhang et al. [13] obtained a hole concentration of 6.5 × 1019 cm-3 and mobility of 2.4 cm2V-1s-1 for cubic Cu3SbS3, and a hole concentration of 1.9 × 1020 cm-3 and mobility of 5.8 cm2V-1s-1 for cubic Cu3Sb0.9Fe0.1S3. Du et al. [14] reported a hole concentration of 9.1 × 1019 cm-3 and mobility of 3.7 cm2V-1s-1 for orthorhombic Cu3SbS3 (in reality, orthorhombic skinnerite + cubic tetrahedrite), and a hole concentration
of 3.2 × 1020 cm-3 and mobility of 16 cm2V-1s-1 for cubic Cu3Sb0.9Fe0.1S3, similar to values obtained for cubic tetrahedrite.
Figure 4 shows the electrical conductivity of Cu3Sb1-yFeyS3. Cubic Cu3SbS3 exhibits an electrical conductivity of (0.95–1.92) × 104 Sm-1 at 323–623 K, but Fe doping increased this to (2.55–3.62) × 104 Sm-1 at 323 K and (4.48–6.47) × 104 Sm-1 at 623 K, indicating non-degenerate semiconductor behavior. However, as the Fe content
increases, the temperature dependence of the electrical conductivity decreases. Chen
et al. [6] obtained a very low electrical conductivity of approximately 10 Sm-1 at 300 K for orthorhombic Cu3SbS3, owing to the low carrier concentration. Zhang et al. [13] reported an electrical conductivity of (0.25– 2.00) × 104 Sm-1 for cubic Cu3SbS3 and (1.75–3.57) × 104 Sm-1 for cubic Cu3Sb0.9Fe0.1S3 at 323–623 K. Du et al. [14] achieved (0.54–1.63) × 103 Sm-1 for orthorhombic Cu3SbS3 and (0.81–2.00) × 104 Sm-1 for cubic Cu3Sb0.9Fe0.1S3 at 300– 625 K.
Figure 5 shows the Seebeck coefficients of Cu3Sb1-yFeyS3. Lee [17,18] obtained a Seebeck coefficient of 140–178 μVK-1 at 323–623 K for cubic Cu3SbS3. However, in this study, Fe doping lowered the Seebeck coefficient to 97–106 μVK-1 at 323–623 K for Cu3Sb0.94Fe0.06S3. This was attributed to the increase in the carrier concentration caused by Fe doping.
When the temperature increased in the measurement temperature range, the Seebeck coefficient
gradually increased; an intrinsic transition did not occur. The positive Seebeck coefficient
confirms that the Fe-doped skinnerite acts as a p-type semiconductor. Zhang et al.
[13] reported Seebeck coefficients of 177–180 μVK-1 for cubic Cu3SbS3 and 114–162 μVK-1 for cubic Cu3Sb0.9Fe0.1S3 at 300–623 K. Du et al. [14] obtained values of 165–220 μVK-1 for orthorhombic Cu3SbS3 and 165–180 μVK-1 for cubic Cu3Sb0.9Fe0.1S3 at 300–625 K, which were higher than the 115–145 μVK-1 reported for tetrahedrite at the same temperatures.
Figure 6 shows the temperature dependence of the thermoelectric power factor of Cu3Sb1-yFeyS3. Cubic Cu3SbS3 exhibited a power factor of 0.17–0.61 mWm-1K-2 at 323–623 K; however, this was greatly improved by doping with Fe to 0.28–0.33 mWm-1K-2 at 323 K and 0.79–1.17 mWm-1K-2 at 623 K. This results from a significant increase in electrical conductivity, although
there was a decrease in the Seebeck coefficient because of Fe doping. Cu3Sb0.98Fe0.02S3 exhibited a maximum power factor of 1.17 mWm-1K-2 at 623 K. Du et al. [14] reported a low power factor of 0.077 mWm-1K-2 at 625 K for orthorhombic Cu3SbS3 but obtained a 10-fold increased value of 0.627 mWm-1K-2 at 625 K for cubic Cu3Sb0.9Fe0.1S3. As a result, Fe doping at skinnerite was suggested to improve the power factor;
however, this value is lower than 1.04 mWm-1K-2 at 625 K for the tetrahedrite produced by the same MA-HP method.
Figure 7 shows the thermal conductivity of Cu3Sb1-yFeyS3. Cubic Cu3SbS3 exhibits a thermal conductivity of 0.64–0.74 Wm-1K-1 at 323–623 K; however, cubic Cu3Sb1-yFeyS3 (y = 0.02–0.06) reached a higher thermal conductivity of 0.88– 1.05 Wm-1K-1 at 323 K and 1.07–1.20 Wm-1K-1 at 623 K. Zhang et al. [13] obtained a thermal conductivity of 0.45– 0.70 Wm-1K-1 for cubic Cu3SbS3 and 0.55–0.95 Wm-1K-1 for cubic Cu3Sb0.9Fe0.1S3 at 300–623 K. Du et al. [14] reported 0.30–0.25 Wm-1K-1 for orthorhombic Cu3SbS3; Fe doping increased this to 0.45–0.55 Wm-1K-1 for cubic Cu3Sb0.9Fe0.1S3 at 300–625 K. However, these values are lower than the 0.85–0.95 Wm-1K-1 obtained for tetrahedrite at the same temperatures.
Figure 8 shows the lattice and electronic thermal conductivities of Cu3Sb1-yFeyS3. Assuming no bipolar conduction, thermal conductivity can be expressed as the sum
of the lattice thermal conductivity (κL) by phonon transport and the electronic thermal conductivity (κE) of the charge carriers. Here, the electronic thermal conductivity was calculated
using the Wiedemann–Franz relation (κE = LσT). The temperature-dependent Lorenz number (L) was calculated using the formula
L = 1.5 + exp (–|α|/116) [24]. Lee [17,18] reported κL = 0.53–0.59 Wm-1K-1 and κE = 0.05– 0.20 Wm-1K-1 for cubic Cu3SbS3 at 323–623 K. In this study, cubic Cu3Sb1-yFeyS3 (y = 0.02–0.06) exhibited κL = 0.65– 0.87 Wm-1K-1 and κE = 0.16–0.23 Wm-1K-1 at 323 K, and κL = 0.48–0.62 Wm-1K-1 and κE = 0.50–0.72 Wm-1K-1 at 623 K. Zhang et al. [13] reported κL = 0.45–0.40 Wm-1K-1 for cubic Cu3SbS3 and κL = 0.61–0.36 Wm-1K-1 for cubic Cu3Sb0.9Fe0.1S3 at 300–623 K. Du et al. [14] obtained κL = 0.30–0.20 Wm-1K-1 for orthorhombic Cu3SbS3 and κL = 0.40–0.25 Wm-1K-1 for cubic Cu3Sb0.9Fe0.1S3 at 300–625 K, which is lower than κL = 0.75–0.40 Wm-1K-1 for tetrahedrite at 300–625 K.
Figure 9 shows the temperature dependence of the ZT values for Cu3Sb1-yFeyS3. The ZT value increased with temperature because of the temperature dependence of
the power factor and thermal conductivity. The ZT values of the Fe-doped cubic skinnerites
Cu3Sb1-yFeyS3 (y = 0.02–0.06) ranged from 0.09–0.12 at 323 K to 0.59–0.45 at 623 K. In our previous
studies [17,18], cubic Cu3SbS3 prepared by the MA-HP process exhibited a maximum ZT of 0.51 at 623 K. However, in
this study, Fe doping stabilized the cubic skinnerite phase and improved the thermoelectric
performance; as a result, ZT = 0.59 was achieved at 623 K for Cu3Sb0.9Fe0.02S3. Zhang et al. [13] reported a ZT = 0.57 for cubic Cu3SbS3 and ZT = 0.62 for cubic Cu3Sb0.9Fe0.1S3 produced by the MA-HP-annealing-quenching process at 623 K. Du et al. [14] obtained a low value of ZT = 0.1 at 623 K for orthorhombic Cu3SbS3 produced by the MA-HP method, but the thermoelectric performance was greatly improved
by Fe doping, resulting in ZT = 0.25–0.70 at 623 K for cubic Cu3Sb1-yFeyS3 (y = 0.02–0.1).
4. CONCLUSIONS
Cubic skinnerites Cu3Sb1-yFeyS3 (0.02 ≤ y ≤ 0.06) were successfully synthesized using MA and HP. From the results of the XRD
phase identification and SEM–EDS analysis, Fe doping was effective in creating a single
skinnerite phase with a cubic structure. However, no skinnerite phases with monoclinic
or orthorhombic structures were produced. The HP was performed at 623 K to obtain
homogeneous and densely sintered bodies. The cubic skinnerite phase remained after
HP without changes in the crystal structure, phase transition, or phase decomposition.
The Fe-doped skinnerites behaved like p-type semiconductors; Cu3Sb0.98Fe0.02S3 and Cu3Sb0.96Fe0.04S3 exhibited non-degenerate semiconductor behavior, whereas Cu3Sb0.94Fe0.06S3 exhibited degenerate semiconductor behavior with minimal temperature dependence on
electrical conductivity. Fe doping at the Sb sites increased the carrier concentration,
resulting in an increase in electrical conductivity and a decrease in the Seebeck
coefficient. An increase in the electronic thermal conductivity by Fe doping resulted
in an increase in thermal conductivity, and a thermal conductivity of 0.85–1.20 Wm-1K-1 was obtained at 323–623 K. Cu3Sb0.98Fe0.02S3 exhibited a maximum power factor of 1.16 mWm-1K-2 and the highest ZT of 0.59 at 623 K. In summary, a single Fe-doped cubic skinnerite
phase was prepared by the MA-HP process without subsequent annealing, and the partial
doping of Fe at the Sb sites was effective in stabilizing the skinnerite phase and
improving thermoelectric performance. We conclude that skinnerite has potential as
a thermoelectric material for mid-temperature applications.
Acknowledgements
This study was supported by the Basic Science Research Capacity Enhancement Project
(National Research Facilities and Equipment Center) through the Korea Basic Science
Institute funded by the Ministry of Education (Grant No. 2019R1A6C1010047).
REFERENCES
Zhou X., Yan Y., Lu X., Zhu H., Han X., Chen G., Ren Z., Mater. Today,21, 974 (2018)

Ando O. H., Maran A. L. O., Henao N. C., Renew. Sustain. Energy Rev,91, 376 (2018)

Powell A. V., J. Appl. Phys,126, 100901 (2019)

Baday C., Yurddaskal M., Ozgul M., Zor M., Celik E., J. Mater. Sci,28, 17468 (2017)

Lu X., Morelli D. T., Xia Y., Zhou F., Ozolins V., Chi H., Zhou X., Uher C., Adv.
Energy Mater,3, 342 (2012)

K. Chen, Synthesis and Thermoelectric Properties of Cu-Sb-S Compounds, Ph.D. Thesis
(United Kingdom: Queen Mary University of London, 2016),
Maiello P., Zoppi G., Miles W. R., Pearsall N., Forbes I., Sol. Energy Mater. Sol.
Cells,113, 186 (2013)

Ramasamy K., Sims H., Butler W. H., Gupta A., Chem. Mater,26, 2891 (2014)

Nefzi K., Rabhi A., Kanzari M., J. Mater. Sci.: Mater. Electron,27, 1888 (2016)

Hussain A., Ahmed R., Ali N., Shaari A., Luo J. T., Fu Y. Q., Surf. Coat. Technol,319,
294 (2017)

Whitfield H. J., Sol. Sta. Commun,33, 747 (1980)

Pfitzner A., Z. Kristallogr,213, 228 (1998)

Zhang J., Wang L., Liu M., Wang J., Sun K., Yang Y., Hu B., Xu J., Su T., Du B., J.
Mater. Sci.: Mater. Electron,32, 10789 (2021)

Du B., Zhang R., Liu M., Chen K., Zhang H., Reece M. J., J. Mater. Chem. C,7, 394
(2019)

Tatsuka K., Morimoto N., Econ. Geol,72, 258 (1977)

Raghavan V., J. Phase Equilib. Diffus,25, 459 (2004)

G. E. Lee, Thermoelectric Properties of Cu3Sb(S/Se)4 and Cu3Sb(S/Se)3 Prepared by
Solid-state Synthesis, Ph.D. Thesis (Republic of Korea: Korea National University
of Transportation, 2021),
Lee G. E., Kim I. H., Korean J. Met. Mater,60, 455 (2022)

Shannon R. D., Acta Crystallogr. A,32, 751 (1976)

Daniel T., Mohanraj K., Sivakumar G., J. Mater. Sci.: Mater. Electron,29, 9251 (2018)

Wang M. X., Yue G. H., Fan X. Y., Yan P. X., J. Cryst. Grow,310, 3062 (2008)

Atri S., Gusain M., Kumar P., Uma S., Nagarajan R., J. Chem. Sci,132, 132 (2010)

Yang C., Huang F., Wu L., Xu K., J. Phys. D: Appl. Phys,44, 295404 (2011)

Kim H. S., Gibbs Z. M., Tang Y., Wang H., Snyder G. J., APL Mater,3, 041506 (2015)

Figures and Table
Fig. 1.
XRD patterns of (a) synthetic powders and (b) sintered specimens of Cu3Sb1-yFeyS3.
Fig. 2.
SEM images of the (a) polished surface with EDS elemental line scans, (b-d) fractured
surfaces of Cu3Sb1-yFeyS3, and (e-h) EDS elemental maps of Cu3Sb0.98Fe0.02S3.
Fig. 3.
Charge transport properties of Cu3Sb1-yFeyS3.
Fig. 4.
Temperature dependence of the electrical conductivity for Cu3Sb1-yFeyS3.
Fig. 5.
Temperature dependence of the Seebeck coefficient for Cu3Sb1-yFeyS3.
Fig. 6.
Temperature dependence of the power factor for Cu3Sb1-yFeyS3.
Fig. 7.
Temperature dependence of the thermal conductivity for Cu3Sb1-yFeyS3.
Fig. 8.
Separation of the (a) lattice thermal conductivity and (b) electronic thermal conductivity.
Fig. 9.
Dimensionless figure of merit for Cu3Sb1-yFeyS3.
Table 1.
Relative densities and lattice constants of Cu3Sb1-yFeyS3.
|
Specimen
|
Relative density [%]
|
Lattice constant [nm]
|
|
Cu3SbS3 |
99.7
|
1.03740
|
|
Cu3Sb0.98Fe0.02S3 |
98.8
|
1.03370
|
|
Cu3Sb0.96Fe0.04S3 |
98.3
|
1.03366
|
|
Cu3Sb0.94Fe0.06S3 |
98.6
|
1.03310
|