3.1. Structure and composition analysisFigure 1(a) shows the XRD patterns of Bi
Sb
Te
obtained after adding X mol% of MnSb
Se
(from (a) to (e), X = 0, 1, 2, 3, and 4), and those of Bi
Sb
Te
, obtained by adding 2 mol% of Mn
Cu
Sb
Se
(Fig. 1(a), trace (f)). As seen from the XRD patterns, the XRD diffraction patterns for the doped samples are identical to that of the Bi
Sb
Te
phase, and no secondary phases of MnSb
Se
or Mn
Cu
Sb
Se
were detected using our equipment. Meanwhile, some Mn and Se atoms form MnSe alloys in the host, but this cannot be resolved by XRD owing to its trivial volume ratio in the composite. The above speculation is further supported by TEM images shown below. The XRD pattern of Bi
Sb
Te
can be indexed by the space group of R-3mH (No. 166). Figure 1(b) shows the lattice parameters (
) of the doped Bi
Sb
Te
, as a function of the MnSb
Se
content, for various X. Doping of the Bi
Sb
Te
samples increased the values of the a and b lattice parameters, compared with the undoped sample, while the values of the c lattice parameter was reduced. Meanwhile, the cell volume was reduced by the doping. Figure 1(c) and 1(d) show the low- and high-magnification SEM images of the doped Bi
Sb
Te
(by adding 2 mol% of MnSb
Se
, respectively, indicating that the sample exhibits uniform-size grain and no preferred texture.
Figure 2 shows the TEM and EDX images of Bi
Sb
Te
with added 2 mol% of MnSb
Se
(Figs. 2(a1)–2(a3)) and 2 mol% of Mn
Cu
Sb
Se
(Figs. 2(b1)–2(b3)), respectively. The low-magnification TEM image of Bi
Sb
Te
with MnSb
Se
(Fig. 2(a1)) reveals uniformly sized grains and a second phase at the grain boundary, which is more clearly revealed in the amplified TEM image in Fig. 2(a2). The corresponding EDX mapping images of Bi, Sb, Te, Mn, and Se are shown in the right column in Figs. 2(a1)–2(a3). It can be seen that Bi is abundantly present at the grain boundary, which arises owing to the decomposition of a small amount of Bi
Sb
Te
in the presence of hightemperature and high pressure. Elemental Bi is in the liquid form and is expelled to the grain boundary. Figure 2(a3) shows an HRTEM image of Bi
Sb
Te
with 2 mol% of MnSb
Se
. In the image, the calculated inter-planar distance is ∼0.31 nm, which is consistent with the (015) plane of Bi
Sb
Te
(JCPDS 49-1713).
Figure 2(b1)–2(b4) shows the microstructural features of Bi
Sb
Te
with 2 mol% of Mn
Cu
Sb
Se
. The grains are uniformly sized, and the second phase of nanoparticles is clear, too. This is more clearly revealed by the magnified TEM images in Figs. 2(b2) and 2(b3). The elemental EDX mapping in Fig. 2(b3) is shown in the left column. Cu, Sb, and Te were all uniformly distributed in the host matrix. However, Mn and Se were enriched in nanoparticles, suggesting the formation of the MnSe foreign phase. Figure 2(b4) shows a high-magnification TEM image of the sample, with a clearly identifiable lattice structure of MnSe. We suggest that alloying or replacing reactions occur in these samples. We assume that addition of MnSb
Se
or Mn
Cu
Sb
Se
to the composites results in the decomposition during the preparation process, following which Mn atoms can be incorporated into the BiSb
Te
lattice by replacing the atoms at BiSb sites, with Se atoms replacing Te atoms, leading to the formation of the (BiSbMn)
(TeSe)
alloy with the same lattice structure as that of (BiSb)
Te
. The EDX and HRTEM results indicate that MnSe nanoparticles emerge as the foreign second phase in the host matrix of Bi
Sb
Te
.
3.2. Thermal and electrical transport propertiesThe temperature dependence of thermal conductivity, for temperatures in the 300–550 K range, for a series of samples, is shown in Fig. 3 (the data reported by Poudel et al. are provided for comparison[10]). As shown in Fig. 3(a), the overall thermal conductivity increases with increasing temperature. It is clear that all thermal conductivities of the doped samples are higher than that of Bi
Sb
Te
for temperatures under 400 K, which can be rationalised by considering their higher electrical thermal conductivities, shown in Fig. 3(c). The thermal conductivities of the composite samples are lower than that of the undoped Bi
Sb
Te
for
K. The addition of MnSb
Se
or Mn
Cu
Sb
Se
increases the carrier concentration and suppresses the bipolar diffusion.[15] Figure 3(b) shows the Lorenz numbers L of the samples, which were calculated using a single parabolic band model (SPB),[16–19] as follows:[20]
| (1) |
where
L is in
K
−2 and
S is in
V/K. Figure
3(c) and
3(d) show the temperature dependence of electrical thermal conductivities and lattice conductivities for the samples. The carrier thermal conductivity
can be obtained by using the Wiedemann–Franz law,
, and the lattice thermal conductivity
can be calculated by subtracting
from the overall thermal conductivity
(where
L,
σ, and
T are the Lorenz number, the electrical conductivity, and the temperature, respectively). Figure
3(c) shows that the
values for the doped samples are higher than for the Bi
Sb
Te
, owing to the improvement in the carrier concentration. Nevertheless, the lattice thermal conductivities for the doped samples are lower than that for the undoped sample. This is mainly owing to the increased scattering from the second phase and the increased presence of structural defects in the host matrix,
[21,22] as shown in Fig.
3(d). The overall thermal conductivities for the Bi
Sb
Te
alloys with MnSb
Se
or Mn
Cu
Sb
Se
are lower than that of the nano-crystalline BiSbTe sample below 400 K, as reported by Poudel
et al.Figure 4 shows the temperature dependence of the electrical resistivity ρ, the Seebeck coefficient S, the power factor
, the ZT value, the room temperature Hall carrier concentration (n), and the carrier mobility (μ), respectively. As shown above, the electrical resistivities of the doped samples are lower than that of the undoped Bi
Sb
Te
sample. The can be ascribed to the substitution of Bi
by Mn
in MnSb
Se
that contributed free vacancies, which is similar to Djieutedjeu's report.[23] The doping process can be described as follows:
| (2) |
In addition, the lowest electrical resistivity was achieved for Bi
Sb
Te
with 2 mol% of Mn
Cu
Sb
Se
. This can also be ascribed to the substitution of Bi
by Cu
, resulting in two more free vacancies in the sample. This process can be described as follows:
| (3) |
Thus, it is expected that the carrier concentration in the Bi
Sb
Te
sample doped with Mn
Cu
Sb
Se
is higher than those in other samples, which is supported by the data in Fig.
4(c). From Fig.
4(b), the Seebeck coefficients of the doped samples are lower than that of the undoped sample below 450 K, and the Seebeck coefficient of the undoped Bi
Sb
Te
decreases faster than those of other samples with increasing the temperature, and this may be owing to the increasing hole concentrations for the doped samples, as shown in Fig.
4(c). Figure
4(c) shows the room temperature Hall carrier concentration and carrier mobility for the samples (the independent points of different carrier concentrations and carrier mobility represent Bi
Sb
Te
with 2 mol% of Mn
Cu
Sb
Se
. The reference Bi
Sb
Te
exhibits a low carrier concentration (
n
cm
), while the carrier concentration increases with the addition of MnSb
Se
. The highest carrier concentration is obtained after adding 2 mol% of MnSb
Se
(
n
cm
), and the highest carrier mobility (
μ ∼152 cm
V
s
) is achieved for the same sample. The carrier concentration for Bi
Sb
Te
with 2 mol% of Mn
Cu
Sb
Se
reaches a maximum of
cm
and the carrier mobility is 140 cm
V
s
. However, the hole mobility is weakly dependent on the amount of MnSb
Se
, as shown in Fig.
4(c), because the hole moves close to the Te position, but Mn and Cu substitute for the Bi position and therefore cause weak scattering at the hole.
[24]Figure 4(d) shows the power factors (
) of all samples, and it is obvious that the power factors of all of the doped samples are improved. The PF of Bi
Sb
Te
with 2 mol% of Mn
Cu
Sb
Se
is higher than others, constituting an average
improvement over that of the undoped Bi
Sb
Te
. Owing to a high electrical resistivity, the power factor of the sample with 2 mol% of Mn
Cu
Sb
Se
is lower than that of the nano-crystalline (NC) BiSbTe sample.[10] Figure 4(e) shows the temperature-dependent ZTs of the samples, revealing a significant improvement of the ZT values of doped samples. In particular, the ZT value for Bi
Sb
Te
with 2 mol% of Mn
Cu
Sb
Se
is the most striking, and the maximal ZT value of ∼1.43 at 375 K reveals a
improvement over the value for the reference Bi
Sb
Te
. Overall, incorporating a small quantity of MnSb
Se
or Mn
Cu
Sb
Se
into Bi
Sb
Te
alloys significantly improves their thermoelectric performance.
Figure 5 shows the cumulative temperature dependence of thermoelectric performance, for all samples. To evaluate the energy conversion efficiency (η) and output power density (
) of the TE material, the engineering dimensionless figure of merit (
) and the engineering power factor (
) have been developed by Kim et al.[26]
can be expressed as follows:
| (4) |
where
,
, and
are the hot-side temperature, the cold-side temperature, and
–
, respectively. The output power density (
) can be obtained from the engineering power factor (
), as follows:
[26] | (5) |
where
L is the Lorenz number and
is defined as
.
Figure 5(a) shows the dependence of
on
for the doped samples, clearly showing that the
values of the doped samples are substantially enhanced with respect to that of the undoped sample. In particular, a
value of 0.62 (at
K) for BiSbTe with 2 mol% of Mn
Cu
Sb
Se
is higher than that for the other samples, which reveals a
improvement over the value for the undoped sample, mostly owing to a higher
, as shown in Fig. 5(c). Figure 5(b) shows the energy conversion efficiency (η) for all of the samples. The cumulative temperature-dependent maximal efficiency is defined as[26]
| (6) |
where
is the Carnot efficiency and
is a dimensionless intensity factor of the Thompson effect (which plays an important role in increasing the efficiency
[25]) defined as
in which
) is the Seebeck coefficient at the hot-side temperature
. The efficiency of Bi
Sb
Te
with 2 mol% of Mn
Cu
Sb
Se
approaches 10.24% at
K, which is
higher than that of the undoped Bi
Se
Te
sample. Figure
5(c) and
5(d) show the engineering power factor (
and the output power density (
), for all of the samples. The
and
for all of the doped samples have been improved, especially for the sample with Mn
Cu
Sb
Se
. Thus, a high value of engineering
(0.62) and a high conversion efficiency (
η) of
(at
K) were obtained for the Bi
Sb
Te
alloy by adding Mn
Cu
Sb
Se
.