The influence of argon working pressure during magnetron sputtering on thermoelectric properties has been investigated on p-type Bi0.5Sb1.5Te3 flexible films deposited at various working pressures in the range from 2 to 5 Pa. The microstructure and orientations, atomic compositions, and carrier concentration could be regulated by adjusting the working pressure, due to the size-dependent inhibition of the deposition of the sputtered Bi, Sb, and Te atoms from argon ions. Profiting from the occurrence of the (006) orientation, the nearest stoichiometric ratio, the highest carrier concentration and mobility, and the quantum confinement effect, the film deposited at 4 Pa displays the maximum power factor of 1095 μW m−1 K−2 at 360 K. These results suggest that the electrical transport properties of the sputtered flexible thermoelectric thin films can be synergistically optimized by selecting an appropriate working pressure.

Thermoelectric (TE) materials have attracted a large amount of scientific research due to their unique ability to directly convert waste heat into electrical energy. A dimensionless figure of merit ZT, defined as ZT = S2σT/κ, is usually used to evaluate the energy conversion efficiency of TE materials, where S, T, σ, and κ are the Seebeck coefficient, absolute temperature, electrical conductivity, and thermal conductivity, respectively.1,2 The power factor, defined as PF = S2σ, represents the electrical transport performance. Obviously, optimization in the ZT relies on increasing the PF or reducing the κ.

Bi2Te3-based alloys are most widely used for power generation and refrigeration in low temperature ranges (<600 K).3 In recent years, a traditional middle-temperature (500–800 K) material GeTe has been well advanced in TE performance within a wide temperature range of 300–723 K.4,5 However, thermal stability and manufacturing cost limit their industrial applications. In comparison with their bulk counterparts, Bi2Te3-based films offer a more lightweight, comfortable, and low-cost solution for wearable or miniature TE devices.6 Furthermore, a low dimensional thin film probably has a higher Seebeck coefficient due to the quantum confinement effect, and lower thermal conductivity due to stronger phonon scattering at the interfaces and boundaries, presumably leading to the promotion of the ZT.7,8 Realizing that the (00l) orientation has been proven an effective strategy to improve the PF of the Bi2Te3-based films.9–11 Moreover, the substrate types12–14 and subsequent annealing conditions15–19 reveal distinct influences on the TE properties of the deposited films. Recently, we realized the carrier energy filtering effect in Ag-doped Bi0.5Sb1.5Te3 films deposited by magnetron sputtering, and significantly improved their electrical transport properties.20 

The magnetron sputtering technique helps to form highly uniform nanocrystal domains in fabricated films, which could construct coherent interfaces and appropriate potential barriers. In fact, being convenient, scalable, and cost-competitive, magnetron sputtering has been frequently used to prepare TE thin films.21–23 Most of the technical parameters can be readily modulated during sputtering, leading to a great possibility to optimize the TE performance of the deposited films through regulating their microstructure and stoichiometry. However, to the best of our knowledge, most of the investigations on Bi2Te3-based films fabricated through magnetron sputtering are limited in hard substrates. The effects of sputtering parameters on the TE properties of flexible films are scarcely investigated,24–28 especially against the p-type Bi0.5Sb1.5Te3 films.

In this work, we fabricated the flexible Bi0.5Sb1.5Te3 films at different working pressures during magnetron sputtering to investigate their influence on the TE performance. Due to the synergistically optimized orientation, atomic compositions, and carrier concentration, the film deposited at 4 Pa owns the highest power factor 1095 μW m−1 K−2 at 360 K, much better than those of the films deposited at other work pressures.

Bi0.5Sb1.5Te3 thin films were prepared on flexible polyimide (PI) substrates by magnetron sputtering with a Bi0.5Sb1.5Te3 target (99.95% purity). PI pieces of 10 × 10 × 1 mm3 size were cleaned with alcohol, acetone, de-ionized water, dried with nitrogen gas gun blowing, and then installed into the substrate bracket. The base pressure of the sputtering chamber was less than 9.9 × 10−4 Pa, and the working pressure was set to 2, 3, 4, and 5 Pa, respectively. The flow rate of Ar gas (99.999%) was 40 sccm (standard cubic centimeters per minute). During the deposition process, the substrate bracket is heated at 100 °C and rotated around the vertical axis of the chamber at 30 rpm to improve the uniformity of the obtained films. All Bi0.5Sb1.5Te3 films were prepared by RF magnetron sputtering of the Bi0.5Sb1.5Te3 target with a power of 100 W. Among them, four types of thin films were obtained by sputtering at 2 and 3 Pa for 1 h, at 4 Pa and 5 Pa for 2 h, which were labeled as BST2, BST3, BST4, and BST5, respectively. Finally, all the deposited films were annealed at 300 °C under Ar gas flow for 1 h. The annealed films were marked as BST2a, BST3a, BST4a, and BST5a in sequence.

The phase of the thin films was analyzed by a Rigaku Miniflex 600 X-ray diffractometer (XRD) with Cu Kα irradiation. The surface microstructure and composition analysis were performed by a combination of Zeiss Supra 55 scanning electron microscopy (SEM) and Bruker X-flash 6130 energy dispersive spectroscopy (EDS). The electrical conductivity and Seebeck coefficient were measured simultaneously under vacuum conditions from 300 to 575 K by a JouleYacht China thin films thermoelectric measurement system. The carrier concentration n and mobility μ were determined using the Van der Pauw method on an East Changing ET9000 system. The film thickness was determined by a Kla-Tencor XP-2 profiler. The uncertainties of the measurements and the calculated parameters are 5% for electrical conductivity, 7% for the Seebeck coefficient, and 10% for the power factor. The schematic diagram of the electrical conductivity and Seebeck coefficient measurements and the image of one typical Bi0.5Sb1.5Te3 film on the PI substrate are provided in the supplementary information (Fig. S1).43 

Figure 1(a) shows the XRD patterns of the BST2a-5a films prepared at different working pressures. All the major peaks can be indexed to Bi0.5Sb1.5Te3 (PDF#49-1713) without a second phase. The noisy patterns indicate low crystallinity due to the low deposition temperature. It is obvious that the annealing process improved the crystallinity for all the deposited films, as revealed in the supplementary information (Fig. S2).43 All the annealed films have two major diffraction peaks corresponding to the (015) and (1010) planes, whose relative intensities increase as the working pressures increase from 2 to 5 Pa. Thus, the BST4a and BST5a films have the best crystallinities. It is worth mentioning that a small peak corresponding to the (006) plane occurs in BST4a films. The crystallite size D has been calculated for all the annealed films using Debye Scherrer's equation: D = 0.9 λ/(βcosθ), where λ is the wavelength of x-ray radiation (0.154056 nm), β is the full width half maximum, and θ is the diffraction angle. The D value increases gradually from 18.3 nm for BST2a to 31.5 nm for BST5a. This size is typical in sputtered films on the PI substrate,13,28 and should be slightly bigger than the actual value due to the instrumental and strain-induced broadening in β.29–32 Based on the XRD patterns, the lattice parameters were calculated. Their dependence on the working pressure is plotted in Fig. 1(b). Slight variations in a, b, and c might originate from the changing compositions. Figure 1(c) depicts the atomic compositions of the BST2a-5a films from EDS analysis. As the working pressure increases from 2 to 5 Pa, the contents of Bi and Sb decrease from 11.09% to 10.05%, and 29.47% to 28.86%, respectively. Meanwhile, the Te content gradually increases from 59.44% to 61.09%. These variations might originate from a size-dependent inhibition of the sputtered Bi, Sb, and Te atoms during their deposition process. During sputtering, plenty of argon ions are distributed in the chamber, which have a higher possibility to collide with the biggest Bi atoms and bigger Sb atoms, in comparison with the Te atoms. This trend could be intensified as the working pressure increases. These results indicate that the working pressure affects the structure, orientation, and atomic compositions of the fabricated Bi0.5Sb1.5Te3 films. Among all the samples, the stoichiometric composition deviation in the BST4a film is the smallest.

FIG. 1.

(a) XRD patterns of the BST2a-5a films. (b) Lattice parameters a, b, and c as a function of the working pressure. (c) Atomic compositions as a function of the working pressure.

FIG. 1.

(a) XRD patterns of the BST2a-5a films. (b) Lattice parameters a, b, and c as a function of the working pressure. (c) Atomic compositions as a function of the working pressure.

Close modal

Figures 2(a)2(d) present the SEM surface topography of the BST2-5 films. It can be found that nanoscale grains with flake shape stack irregularly in all the as-deposited films, implying an island-growth model. As the working pressure increases from 2 to 5 Pa, these grains gradually grow larger, consistent with the previous report.33 After annealing, the aggregation of these nanoscale grains leads to a porous structure, which is constructed by a large number of grains owning a wide size range roughly from 10 to 100 nm, as shown in Figs. 2(e)2(h). This porosity was found to decrease as the working pressure increased, accompanied with the reduction of the surface roughness. Relatively, the BST4a film is flatter and denser than the other films, implying fewer grain boundaries and better electrical transport properties. These porous structures and hierarchical grain sizes are beneficial to reduce thermal conductivity. Recently, Zhao et al. achieved almost the lowest lattice thermal conductivity for (Bi,Sb)2Te3 bulk materials by intentionally producing porous structures, where large-scale size distribution of the nanograins and nanopores can effectively scatter a wide spectrum of phonons.34 

FIG. 2.

SEM surface topography of (a) BST2, (b) BST3, (c) BST4, (d) BST5, (e) BST2a, (f) BST3a, (g) BST4a, and (h) BST5a.

FIG. 2.

SEM surface topography of (a) BST2, (b) BST3, (c) BST4, (d) BST5, (e) BST2a, (f) BST3a, (g) BST4a, and (h) BST5a.

Close modal

Figure 3(a) shows the temperature dependence of electrical conductivity σ for theBST2a-5a films. For the BST2a, BST3a, and BST5a films, σ increases as the temperature increases, while that of the BST4a film decreases with the increasing temperature. Increasing the working pressure from 2 to 4 Pa can significantly increase σ. At room temperature, the BST4a film has the highest σ exceeding 3.2 × 104 S/m, approximately five times bigger than the lowest σ in the BST5a film. In principle, σ depends on the carrier concentration n and the carrier mobility μ as σ = neμ, where e is the electron charge. Figure 3(b) illustrates the variations of room temperature n and μ as the working pressure. The BST2a film has the lowest n of 6.81 × 1018/cm3 and μ of 16.3 cm2 V−1 s−1 due to the lowest crystallinity, which could be confirmed by the XRD patterns and SEM images. As the working pressure increased from 2 to 4 Pa, n and μ were improved significantly to 3.47 × 1019/cm3 and 72.8 cm2 V−1 s−1, respectively. The remarkable increase in n might originate from enhanced crystallinity.9,35 Similar behavior has been investigated in Bi0.5Sb1.5Te3 films fabricated through magnetron sputtering. Mu et al. observed synchronous increments in n and μ with increasing substrate temperatures, from 473 to 723 K, and attributed the increased n by 52% to the remarkably improved crystallinity (from 40% to 99%).9 Herein, the simultaneously increased μ with n should be correlated with less carrier scattering as the grain aggregation and the occurrence of (006) orientation as the working pressure increases to 4 Pa. In comparison with those of the BST4a film, the n and μ of the BST5a film decline distinctly. The former probably comes from the excess Te content, which might suppress the formation of the antisite defects and then reduce the p-type carriers. The latter should be correlated with the deteriorated microstructure and the reduction of the (006) orientation.

FIG. 3.

Electrical transport properties of the BST2a-5a films. (a) Temperature-dependent electrical conductivity. (b) Carrier concentrations and mobilities at room temperature. (c) Temperature-dependent Seebeck coefficient. (d) Temperature-dependent power factor.

FIG. 3.

Electrical transport properties of the BST2a-5a films. (a) Temperature-dependent electrical conductivity. (b) Carrier concentrations and mobilities at room temperature. (c) Temperature-dependent Seebeck coefficient. (d) Temperature-dependent power factor.

Close modal
Figure 3(c) gives the temperature dependence of the Seebeck coefficient S for the BST2a-5a films. All the positive S values indicate p-type conducting. As the temperature increases, except that of the BST3a film, the S initially increases and then decreases at higher temperatures, showing the peak Smax at a certain temperature TSmax due to the bipolar effect.36 For the BST2a film, the Smax ∼ 188 μV/K appears at 355 K. For the BST4a and BST5a films, the Smax is slightly increased to ∼191 μV/K at 380 K. This value is relatively high and comparable with those reported previously for Bi0.5Sb1.5Te3 films, which could be attributed to the quantum confinement effect.37,38 Moreover, in the whole testing temperature range, their S values are superior to that of the BST2a film. Generally, the Seebeck coefficient S can be expressed as
S = 8 π 2 k B 2 3 e h 2 m T ( π 3 n ) 2 / 3 ,
where kB, h, and m* are the Boltzmann constant, Planck constant, and effective mass of carriers, respectively. Then, we obtained the m* values 0.30m0, 0.75m0, 0.86m0, and 0.71m0 for BST2a-5a thin films, respectively. The variation in m* suggests the modulation of the energy band structure. In fact, low crystallinity in the case of low substrate temperature during magnetron sputtering leads to many localized energy levels in the band gap, which would be eliminated gradually with the improved crystallinity. Therefore, the working pressure indirectly affects the band gap Eg. Theoretically, the S is inversely proportional to the carrier concentration n as S n 2 / 3, opposite to the σ as σ = neμ. Thus, an enhanced σ is normally accompanied by a declined S. In this work, the simultaneously optimized S with the σ could be attributed to the following reasons: (i) the enhanced crystallinity alters the band structure; (ii) the working pressure can minimize stoichiometric deviation.

Figure 3(d) presents the temperature dependence of the power factor PF for the BST2a-5a films. Benefited from the highest σ and the suboptimal S, the BST4a film has the best PF ∼ 1095 μW m−1 K−2 at 360 K, nearly 5 times that for the BST5a film and double that for the BST2a film. Table I summarizes some representative reports of Bi0.5Sb1.5Te3 films prepared by magnetron sputtering and includes our results for comparison. It can be found that high PF is correlated with the dominant (006) orientation in sputtered Bi0.5Sb1.5Te3 films, especially on the PI substrates. Since (015) is the preferred orientation for Bi0.5Sb1.5Te3 films because of their low formation energy, a high nucleation rate under high sputtering power promotes island growth, favoring the growth along the (015) orientation. Theoretically, the island density N can be described by the formula: N = N 0 e ( φ / k T ), where N0 is the site density of adsorbed incident atoms on the film surface during sputtering; φ is the energy barrier height of atom migration between sites.39 Thus, N is inversely proportional to the substrate temperature T, i.e., the thermal energy provided by the substrate temperature facilitates the in-plane growth along the c-axis orientation. Although the PF value for the BST4a film is comparable with some reports previously, low crystallinity, nonstoichiometric composition, and weak (00l) orientations deteriorate the electrical transport properties. The subsequent improvement for this work will be focused on the synergetic regulation of the sputtering power, the substrate temperature, and the working pressure during magnetron sputtering to realize the dominant (00l) orientations for better carrier mobility and electrical transport performance.

TABLE I.

Comparison of the TE performance of Bi0.5Sb1.5Te3 films at room temperature prepared by magnetron sputtering.

MaterialSubstrateOrientationσ (104 S m−1)S (μV/K)PF (μW m−1 K−2)
Bi0.5Sb1.5Te3 (Ref. 40SiO2 (015) 7.3 249 4530 
Bi0.5Sb1.5Te3 (Ref. 41SiO2 (015) 5.0 249 3149 
Bi0.5Sb1.5Te3 (Ref. 34Glass (006) 4.4 190 2000 
Bi0.5Sb1.5Te3 (Ref. 42PI (006) (0015) 2.1 196 790 
Bi0.5Sb1.5Te3 (Ref. 11PI (006) 3.7 177 1150 
Bi0.5Sb1.5Te3 (Ref. 10PI (1010) (006) 6.8 230 3600 
This work PI (015) 3.2 179 1057 
MaterialSubstrateOrientationσ (104 S m−1)S (μV/K)PF (μW m−1 K−2)
Bi0.5Sb1.5Te3 (Ref. 40SiO2 (015) 7.3 249 4530 
Bi0.5Sb1.5Te3 (Ref. 41SiO2 (015) 5.0 249 3149 
Bi0.5Sb1.5Te3 (Ref. 34Glass (006) 4.4 190 2000 
Bi0.5Sb1.5Te3 (Ref. 42PI (006) (0015) 2.1 196 790 
Bi0.5Sb1.5Te3 (Ref. 11PI (006) 3.7 177 1150 
Bi0.5Sb1.5Te3 (Ref. 10PI (1010) (006) 6.8 230 3600 
This work PI (015) 3.2 179 1057 

This work fabricated flexible p-type Bi0.5Sb1.5Te3 using the magnetron sputtering technique at different argon working pressures. The structure, morphology, composition, and electrical transport property analysis manifest that an appropriately selected working pressure can induce the (006) orientation of the deposited film, optimize the microstructure after the annealing, modulate the atomic compositions, and then, the carrier concentration. Finally, owing to the simultaneously optimized S and σ, the BST4a film gets the highest PF ∼ 1095 μW m−1 K−2 at 360 K, much better than those fabricated under the same conditions except the working pressure. These results show that for better TE performance, the technique parameters during magnetron sputtering need systematic optimization against different substrates and different TE systems.

This research was supported by the Fundamental Research Funds for the Central Universities (Grant No. 21622419) and the Science and Technology Planning Project of Guangzhou (Grant No. 201605030008).

The authors have no conflicts to disclose.

Ding Hu: Investigation (lead); Writing – original draft (equal). Shaojun Liang: Investigation (supporting); Writing – original draft (equal). Yichun He: Investigation (supporting). Rensheng Zhang: Investigation (supporting). Song Yue: Conceptualization (lead); Funding acquisition (lead); Supervision (lead); Writing – review & editing (lead).

The data that support the findings of this study are available within the article.

2.
C.
Gayner
and
K. K.
Kar
,
Prog. Mater. Sci.
83
,
330
(
2016
).
3.
Q.
He
,
W.
Zhu
,
Y.
Du
,
P.
Wei
,
X.
Nie
,
W.
Zhao
, and
Q.
Zhang
,
J. Power Sources
603
,
234403
(
2024
).
4.
L. C.
Yin
et al,
Adv. Funct. Mater.
33
,
2301750
(
2023
).
5.
D. Z.
Wang
et al,
J. Am. Chem. Soc.
146
,
1681
(
2024
).
6.
H.
Yu
,
Z.
Hu
,
J.
He
,
Y.
Ran
,
Y.
Zhao
,
Z.
Yu
, and
K.
Tai
,
Nat. Commun.
15
,
2521
(
2024
).
7.
M. S.
Dresselhaus
,
G.
Dresselhaus
,
X.
Sun
,
Z.
Zhang
,
S. B.
Cronin
, and
T.
Koga
,
Phys. Solid State
41
,
679
(
1999
).
8.
C. H.
Suen
et al,
J. Materiomics
3
,
293
(
2017
).
9.
X.
Mu
,
H.
Zhou
,
D.
He
,
W.
Zhao
,
P.
Wei
,
W.
Zhu
,
X.
Nie
,
H.
Liu
, and
Q.
Zhang
,
Nano Energy
33
,
55
(
2017
).
10.
G.
Qiu
,
J.
Li
,
Y.
Ling
,
G.
Dong
,
J.
Feng
,
P.
Zhang
, and
R.
Liu
,
J. Alloys Compd.
950
,
169916
(
2023
).
11.
F.
Ma
,
D.
Ao
,
X.
Liu
, and
W. D.
Liu
,
Ceram. Int.
49
,
18584
(
2023
).
12.
B. G.
Kim
,
S. H.
Bae
,
J.
Byeon
,
C.
Lee
, and
S. M.
Choi
,
Mater. Lett.
270
,
127697
(
2020
).
13.
T.
Kurokawa
,
R.
Mori
,
O.
Norimasa
,
T.
Chiba
,
R.
Eguchi
, and
M.
Takashiri
,
Vacuum
179
,
109535
(
2020
).
14.
J.
Zhang
,
H.
Xu
,
Z.
Zheng
,
C.
Wang
,
X.
Li
,
F.
Li
,
P.
Fan
, and
Y. X.
Chen
,
Nanomaterials
13
,
257
(
2023
).
15.
W.
Zhu
,
Y.
Deng
,
Y.
Wang
,
B.
Luo
, and
L.
Cao
,
Thin Solid Films
556
,
270
(
2014
).
16.
J. M.
Lin
,
Y. C.
Chen
,
C. P.
Lin
,
H. C.
Chien
,
C. Y.
Wen
,
J. Y.
Chang
, and
Z. Y.
Zhan
,
Microelectron. Eng.
148
,
51
(
2015
).
17.
Z. H.
Zheng
,
D.
Yang
,
P. C.
Zhang
,
F.
Li
,
Y. X.
Chen
,
G. X.
Liang
, and
P.
Fan
,
Mater. Lett.
275
,
128143
(
2020
).
18.
O.
Norimasa
,
T.
Kurokawa
,
R.
Eguchi
, and
M.
Takashiri
,
Coatings
11
,
38
(
2021
).
19.
B. G.
Kim
,
K. H.
Seo
,
C. H.
Lim
, and
S. M.
Choi
,
J. Mater. Res. Technol.
15
,
606
(
2021
).
20.
S.
Liang
,
Y.
Shi
,
D.
Hu
,
H.
Zhu
, and
S.
Yue
,
Ceram. Int.
49
,
35309
(
2023
).
21.
L.
Cao
,
H.
Gao
, and
M.
Miao
,
CrystEngComm
22
,
7790
(
2020
).
22.
X.
Tao
,
B. W.
Stuart
,
K.
Wang
,
J. W.
Murray
,
E.
Bilotti
, and
H. E.
Assender
,
ACS Appl. Mater. Interfaces
13
,
10149
(
2021
).
23.
M.
Naumochkin
,
G. H.
Park
,
K.
Nielsch
, and
H.
Reith
,
Phys. Status Solidi
16
,
2100533
(
2022
).
24.
K.
Singkaselit
,
A.
Sakulkalavek
, and
R.
Sakdanuphab
,
Adv. Nat. Sci.: Nanosci. Nanotechnol.
8
,
035002
(
2017
).
25.
D.
Kong
,
W.
Zhu
,
Z.
Guo
, and
Y.
Deng
,
Energy
175
,
292
(
2019
).
26.
P.
Jitthamapirom
,
P.
Wanarattikan
,
P.
Nuthongkum
,
R.
Sakdanuphab
, and
A.
Sakulkalavek
,
Ferroelectrics
552
,
64
(
2019
).
27.
S.
Kianwimol
,
R.
Sakdanuphab
,
N.
Chanlek
,
A.
Harnwunggmoung
, and
A.
Sakulkalavek
,
Surf. Coat. Technol.
393
,
125808
(
2020
).
28.
N.
Somdock
,
S.
Kianwimol
,
A.
Harnwunggmoung
,
A.
Sakulkalavek
, and
R.
Sakdanuphab
,
J. Alloys Compd.
773
,
78
(
2019
).
29.
A.
Khorsand Zak
,
W. H.
Abd. Majid
,
M. E.
Abrishami
, and
R.
Yousefi
,
Solid State Sci.
13
,
251
(
2011
).
30.
N. K.
Gupta
et al,
Thin Solid Films
756
,
139355
(
2022
).
31.
L.
Pandey
et al,
Phys. Rev. Mater.
6
,
044203
(
2022
).
32.
L.
Pandey
et al,
J. Phys.: Condens. Matter
35
,
355702
(
2023
).
33.
P.
Fan
,
R.
Li
,
Y. X.
Chen
,
Z.
Zheng
,
F.
Li
,
G.
Liang
, and
J.
Luo
,
J. Eur. Ceram. Soc.
40
,
4016
(
2020
).
35.
Y. X.
Chen
,
K. D.
Shi
,
F.
Li
,
X.
Xu
,
Z. H.
Ge
, and
J.
He
,
J. Am. Ceram. Soc.
102
,
5989
(
2019
).
36.
K.
Kim
,
G.
Kim
,
S. I.
Kim
,
K. H.
Lee
, and
W.
Lee
,
J. Alloys Compd.
772
,
593
(
2019
).
37.
M.
Tan
,
Y.
Deng
, and
Y.
Hao
,
Phys. Status Solidi
210
,
2611
(
2013
).
38.
B.
Du
,
J.
Wu
,
X.
Lai
,
Y.
Deng
,
S.
Wang
,
H.
Liu
,
J.
Liu
, and
J.
Jian
,
Ceram. Int.
45
,
3244
(
2019
).
39.
40.
M.
Tan
,
X. L.
Shi
,
W. D.
Liu
,
M.
Li
,
Y.
Wang
,
H.
Li
,
Y.
Deng
, and
Z. G.
Chen
,
Adv. Energy Mater.
11
,
2102578
(
2021
).
41.
M.
Lan
,
S.
Sun
,
S.
Liu
,
G.
Li
,
Z.
Wang
, and
Q.
Wang
,
J. Eur. Ceram. Soc.
43
,
7493
(
2023
).
42.
Q.
Zou
et al,
Appl. Phys. Lett.
120
,
023903
(
2022
).
43.
See the supplementary material online for the schematic diagram of the thermoelectric measurement apparatus, the image of a flexible Bi0.5Sb1.5Te3 film, and the XRD patterns of all prepared films.

Supplementary Material