⇦Chinese Physics Letters, 2021, Vol. 38, No. 9, Article code 097201 Simultaneous Optimization of Power Factor and Thermal Conductivity towards High-Performance InSb-Based Thermoelectric Materials Wang Li (李旺), Tian Xu (许天), Zheng Ma (马征), Abubakar-Yakubu Haruna, Qing-Hui Jiang (姜庆辉), Yu-Bo Luo (罗裕波)*, and Jun-You Yang (杨君友)* Affiliations State Key Laboratory of Materials Processing and Die & Mould Technology, Huazhong University of Science and Technology, Wuhan 430074, China Received 1 August 2021; accepted 13 August 2021; published online 2 September 2021 Supported by the National Natural Science Foundation of China (Grant Nos. 52002137, 51772109, 51872102, and 51802070), the Fundamental Research Funds for the Central Universities (Grant Nos. 2021XXJS008 and 2018KFYXKJC002), and Graduates' Innovation Fund of Huazhong University of Science and Technology (Grant No. 2020yjsCXCY022).
*Corresponding authors. Email: luoyubo@hust.edu.cn; jyyang@mail.hust.edu.cn Citation Text: Li W, Xu T, Ma Z, Haruna A Y, and Jiang Q H et al. 2021 Chin. Phys. Lett. 38 097201    Abstract Thermoelectric performance of InSb is restricted by its low Seebeck coefficient and high thermal conductivity. Here, CuCl is employed to optimize simultaneously the electrical and thermal transport properties of InSb. The substitution of Cl for Sb results in enhanced electron effective mass, leading to high Seebeck coefficient of $-159.9$ µV/K and high power factor of 31.5 µW$\cdot$cm$^{-1}$$\cdot$K$^{-2}$ at 733 K for InSb + 5 wt% CuCl sample. In addition, CuCl doping creates hierarchical architectures composed of Cu$_{9}$In$_{4}$, Sb, Cu$_{2}$Sb in InSb, leading to a strengthened phonon scattering in a wide wavelength (i.e., nano to meso scale), thus a low lattice thermal conductivity of 2.97 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ at 733 K in InSb + 5 wt% CuCl. As a result, a maximum $ZT$ of 0.77 at 733 K has been achieved for the InSb + 5 wt% CuCl sample, increasing by $\sim $250% compared to pristine InSb. DOI:10.1088/0256-307X/38/9/097201 © 2021 Chinese Physics Society Article Text Thermoelectric (TE) materials, which can convert heat to electricity or vice versa, become more and more important in energy conservation. The conversion efficiency of a TE device is determined by thermoelectric figure-of-merit $ZT = S^{2}T/\rho \kappa =S^{2}T/\rho (\kappa_{\scriptscriptstyle {\rm L}} + \kappa_{\rm e})$, where $S$, $\rho$, $\kappa_{\scriptscriptstyle {\rm L}}$, $\kappa_{\rm e}$ are the Seebeck coefficient, electrical resistivity, the lattice part and electronic part contributing to thermal conductivity, respectively.[1–3] Due to the close relation among $S$, $\rho$ and $\kappa_{\rm e}$ via carrier concentration, it is of great challenge to obtain a significant $ZT$ improvement through only one optimization of three parameters. Numerous efforts have been made to tackle this tough problem, such as first principle calculation,[4,5] low dimension strategies[6–8] and new materials synthesis methods.[9] Several classical high performance TE materials have been developed and many encouraging advances have been made in the past decades.[10] Unfortunately, materials used for thermoelectric generator at medium temperature are dominant by Pb based compounds, which are harmful to our environment and unsuitable for large scale applications. To date, many works have been focused on exploring candidates with environmental-friendly elements.[11–16] InSb, a direct bandgap compound with the NiAs-type crystal structure (with a space group of $F\bar{4}3m$). It has been reported recently as a great potential in medium temperature thermoelectric application.[17] However, the InSb compound has a high thermal conductivity ($\sim $17 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ at room temperature[18]), which largely limits its thermoelectric performance. To reduce the thermal conductivity, many works have been carried out. Su et al. reported a 16.6% decrease of $\kappa_{\scriptscriptstyle {\rm L}}$ by fabricating the InSb compound with the melt spinning method.[19] Moreover, Ga doping[20] and Te doping[21] in InSb have also been proved to be effective for reducing $\kappa_{\scriptscriptstyle {\rm L}}$, which results in $\sim $63% and $\sim $74% reduction in $\kappa_{\scriptscriptstyle {\rm L}}$ due to the extra phonon scattering from Ga-poor nano-particles and In vacancies, respectively. Our previous works revealed a sharp decrease in $\kappa_{\scriptscriptstyle {\rm L}}$ stemmed from the eutectic remelting at 753–773 K.[22,23] Even so, the $\kappa_{\scriptscriptstyle {\rm L}}$ is still large ($> 2$ W$\cdot$m$^{-1}$$\cdot$K$^{-1}$) when $T < 723$ K. In addition, the Seebeck coefficient of InSb is lower ($|S| < 120$ µV/K at 303–723 K) compared with other classical high performance TE materials.[24–26] Therefore, reducing the thermal conductivity and enhancing the Seebeck coefficient of InSb are highly desirable to optimize its thermoelectric performance. In this Letter, CuCl is used as an effective additive to improve the electrical and thermal transport properties of InSb. On the one hand, the substitution of Cl for Sb results in an increase in carrier concentration owing to the electron donor nature of Cl$_{\rm Sb}^{2+}$ point defects, and the substitution of Cl for Sb enhanced the Seebeck coefficient as a result of enhancement in electron effective mass induced by less dispersive electron band. On the other hand, hierarchical architectures composed of mesoscale Cu$_{2}$Sb, nanoscale Sb ($\sim $50 nm) and Cu$_{9}$In$_{4}$ (10–20 nm) result in an obvious reduction in lattice thermal conductivity of InSb due to the strengthened phonon scattering at 650–700 K. As a consequence, the InSb + 5 wt% CuCl sample achieves a maximum $ZT$ of 0.77 at 723 K, which is about 250% higher than that of the pristine InSb. InSb samples were prepared by high temperature (923 K) melting. The obtained ingots were crushed into fine powders and passed through 200-mesh sieve. Then, CuCl powders were added into InSb with different weight ratios $x$% ($x = 0,\, 3,\, 5,\, 8$). The powder mixtures were then subjected to spark plasma sintering (SPS) under a pressure of 60 MPa in vacuum atmosphere to get densified pellets. The characterization methods and specific equipment information are given in the Supplementary Material. Figure 1 displays the powder x-ray diffraction (XRD) patterns of the InSb + $x$ wt% CuCl ($x = 0,\, 3,\, 5,\, 8$) samples. The main Bragg diffraction peaks of all the samples match well with the NiAs-type InSb (PDF 06–0208). It is noted that no CuCl peak can be observed within the detection limit of the XRD measurements, but Bragg peaks of elemental Sb (PDF 35–0732) appear in the samples when the weight fraction of CuCl is more than 3%. The contents of Sb in the InSb + $x$ wt% CuCl ($x = 3,\, 5,\, 8$) samples are estimated by Rietveld refinement (Fig. S1, Table S1), which increases with the increasing $x$, i.e., 0.3%, 3.1% and 6.0% for the $x = 3,\, 5,\, 8$ samples.
cpl-38-9-097201-fig1.png
Fig. 1. Powder XRD patterns of the InSb + $x$ wt% CuCl ($x=0,\, 3,\, 5,\, 8$) samples.
Electron probe microanalysis (EPMA) was carried out to better understand the microstructure of the CuCl doped InSb, i.e., InSb + 8 wt%CuCl. Back-scattering electronic (BSE) image [Fig. 2(a)] shows some dark particles in the grey InSb matrix defined by energy dispersive x-ray spectroscopy (EDS) spot-1 results in Fig. 2(b). Wavelength dispersive spectrometer (WDS) mapping images of Cu and Sb in Figs. 2(c) and 2(d) further confirm that the grey particles mainly consist of Cu and Sb elements. EDS spot-2 results reveal that the composition of such grey particles is Sb : Cu = $31.61\!:\! 67.76$, closed to $1\!:\!2$. Therefore, we speculate such grey particles to be Cu$_{2}$Sb. In addition, considerable Sb precipitates [marked in blue circles in Fig. 2(e)] sized in nanometers can be observed in the sample. It is worth noting that the Cl atoms distribute uniformly in the sample [Fig. 2(f)], indicating the successful substitution of Cl for Sb sites; while the Cu atoms cannot dissolve into InSb according to our WDS measurement, see Fig. 2(c).
cpl-38-9-097201-fig2.png
Fig. 2. (a) Back-scattering electronic (BSE) image of polished surface; (b) energy dispersive x-ray spectroscopy (EDS) analysis of spots in panel (a); (c)–(f) wavelength dispersive x-ray spectroscopy (WDS) mapping results of panel (a).
To further reveal the structure and composition of the InSb + 8 wt%CuCl sample, transmission electrical microscopy (TEM) characterization was carried out and the results are shown in Fig. 3. Dislocations and nanoscale Sb particles ($\sim $50 nm) embedded in InSb matrix can be well observed [Fig. 3(a)], in line with our EPMA results. The corresponding selected area electron diffraction (SAED) pattern indicates that the precipitates are Sb secondary phase (red dots) taken along the [010] axis, as shown in Fig. 3(b). More importantly, large amounts of precipitates sized in several nanometers are observed in InSb matrix, as shown in Fig. 3(c). High-angle annular dark field (HAADF), Fig. 3(d), image and EDS mapping indicate that such nanoprecipitates are mainly composed of Cu and In elements. To further characterize its nature, fast Fourier transform (FFT) was carried out and the result reveals that such nanoprecipitates should be the Cu$_{9}$In$_{4}$ compound with a cubic structure, which is also consistent with the spot composition analysis (Fig. S2). According to the above XRD, EPMA and TEM results, one can see that CuCl doping results in the formation of mesoscale Cu$_{2}$Sb and nanoscale Sb ($\sim $50 nm) and Cu$_{9}$In$_{4}$ (10–20 nm), namely hierarchical architectures. Therefore, we speculate the following reaction occurring during the SPS process of CuCl doped InSb: $$\begin{align} 15x{\rm CuCl}+(5+4x){\rm InSb}={}&5{\rm InSb}_{1-x}{\rm Cl}_{3x} +x{\rm Cu}_{9}{\rm In}_{4}\\ & +3x{\rm Cu}_{2}{\rm Sb}+6x{\rm Sb}.~~ \tag {1} \end{align} $$
cpl-38-9-097201-fig3.png
Fig. 3. Transmission electrical microscopy (TEM) observation of the dense bulk sample for $x = 8$. (a) Low-magnification TEM image of the sample; (b) HRTEM of enlarged image for the area highlighted by yellow dotted line and the corresponding SEAD pattern; (c) a representative area of the matrix, and the inset picture is FFT from the green dotted rectangle; (d) HAADF image of selected area highlighted by red dotted line in (c); (e)–(h) EDS mapping images of (d).
cpl-38-9-097201-fig4.png
Fig. 4. (a) Temperature dependent $\rho$ for InSb + $x$ wt% CuCl ($x = 3,\, 5,\, 8$). (b) Hall carrier concentration ($n_{\scriptscriptstyle {\rm H}}$) and mobility ($\mu_{\scriptscriptstyle {\rm H}}$) vs $x$. (c) Temperature dependent $S$ for InSb + $x$ wt% CuCl. (d) Temperature dependent power factor (PF) for InSb + $x$ wt% CuCl ($x = 3,\, 5,\, 8$).
Figure 4 displays the $\rho$ and $S$ versus temperature for InSb + $x$ wt% CuCl ($x = 0$ 3, 5, 8). The $\rho$ for all the samples decreases with the temperature increasing, indicating a semiconductor behavior. For instance, the $\rho$ of InSb decreases from 17.5 $µ\Omega$$\cdot$m at 313 K to 8.34 $µ\Omega$$\cdot$m at 733 K. In addition, CuCl doping results in an increase in the electrical resistivity of InSb. For instance, the room-temperature $\rho$ of InSb increases from 17.5 $µ\Omega$$\cdot$m for the pristine sample to 306 $µ\Omega$$\cdot$m for the $x = 8$ sample. Our Hall effect measurements show that the main reason for such an increase in $\rho$ is the reduction in carrier mobility resulted from CuCl doping, as shown in Fig. 4(b). The Seebeck coefficients of InSb + $x$ wt% CuCl ($x = 0,\, 3,\, 5,\, 8$) are negative, reflecting their n-type conduction nature. The absolute value $S$ for pristine InSb increases from 76 µV/K at 313 K to 116 µV/K at 600 K and then decreases to the 107 µV/K at 733 K, which is due to the well-known bipolar diffusion, as shown in Fig. 4(c). After CuCl doping, the room temperature $S$ of InSb deceases with the increasing CuCl content, which can be explained by the increase in carrier concentration owing to the substitution of Cl for Sb. In addition, the $S$ of InSb + $x$ wt% CuCl ($x = 0,\, 3,\, 5,\, 8$) is higher than the pristine InSb at 600–733 K. We ascribe this enhancement to the increased electron effective mass[27] which is estimated with the carrier concentration and $S$ value at 313 K, according to the SPB model.[25] The calculation details are given in the Supplementary Material. The electron effective masses are 0.025 m$_{\rm e}$, 0.026 m$_{\rm e}$, 0.034 m$_{\rm e}$, and 0.034 m$_{\rm e}$ for the $x = 0,\, 3,\, 5,\, 8$ samples, respectively, which are closed to that of previous research on InSb.[28] Such an increase in effective mass results from the less dispersive electron band due to Cl doping.[29] Figure 4(d) shows the power factor (PF) of the InSb + $x$ wt% CuCl ($x = 0,\, 3,\, 5,\, 8$) samples. Specifically, for the $x = 5$ sample, the PF is lower than that of the pristine one at 313–600 K due to its higher $\rho$ than InSb, but is higher than the pristine InSb at 600–733 K due to the increase in Seebeck coefficient. Finally, a high value of 31.5 µW$\cdot$cm$^{-1}$$\cdot$K$^{-2}$ is reached at 733 K for the $x = 5$ sample, 123% higher than the pristine one (14.1 µW$\cdot$cm$^{-1}$$\cdot$K$^{-2}$). Figure 5(a) shows the thermal conductivity $\kappa$ for the InSb + $x$ wt% CuCl ($x = 0,\, 3,\, 5,\, 8$) samples. The $\kappa$ of InSb decreases from 10.3 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ at 313 K to 4.6 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ at 733 K. By CuCl doping, the $\kappa$ of $x = 3,\, 5,\, 8$ samples are higher than that of the pristine InSb at 313–650 K, but is lower than the pristine InSb at 650–733 K. For example, the $\kappa$ values are 11.64 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$, 11.8 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ and 10.6 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ for the $x = 3,\, 5,\, 8$ samples at 313 K, and are 2.93 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$, 2.99 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ and 2.63 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ for the $x = 3,\, 5,\, 8$ samples at 733 K, respectively. The electronic thermal conductivity values of all the samples were calculated by $\kappa_{\rm e} = LT/\rho$, where $L$ was estimated by experimental $S$ data according to the single parabolic band (SPB) model,[25] as shown in Fig. S3(a). The $\kappa_{\rm e}$ increases with temperature for all the samples owing to the decrease in electrical resistivity with increasing temperature [Fig. S3(b)].
cpl-38-9-097201-fig5.png
Fig. 5. (a) Temperature-dependent thermal conductivity $\kappa$ and (b) lattice thermal conductivity $\kappa_{\scriptscriptstyle {\rm L}}$ of the InSb + $x$ wt% CuCl samples ($x = 0,\, 3,\, 5,\, 8$).
cpl-38-9-097201-fig6.png
Fig. 6. (a) Temperature dependent $ZT$ values for InSb + $x$ wt% ($x=0,\, 3,\, 5,\, 8$) CuCl; (b) peak $ZT$ of InSb single crystal,[18] InSb made by melt spinning (MS) method,[19] In$_{0.9}$Ga$_{0.1}$Sb[20] compounds and InSb + 5 wt% CuCl.
The lattice thermal conductivity $\kappa_{\scriptscriptstyle {\rm L}}$ of the InSb + $x$ wt% CuCl ($x = 0,\, 3,\, 5,\, 8$) samples, obtained by subtracting $\kappa_{\rm e}$ from the $\kappa$, are shown in Fig. 5(b). Similar to the thermal conductivity $\kappa$, the $\kappa_{\scriptscriptstyle {\rm L}}$ of all the samples decreases with increasing temperature. In the temperature range 313–650 K, the $\kappa_{\scriptscriptstyle {\rm L}}$ of all the samples follows well with the tendency of $\kappa_{\scriptscriptstyle {\rm L}} \sim T^{-1}$, indicating that the phonon scattering is dominated by long acoustic phonons scattering.[23] However, In the temperature range of 313–650 K, the $\kappa_{\scriptscriptstyle {\rm L}}$ of InSb + $x$ wt% CuCl ($x = 3,\, 5,\, 8$) deviates from the $T^{-1}$ tendency, indicating extra phonon scattering occurred. Combined with our XRD, EPMA and TEM results, we speculate it to the strengthened phonon scattering by hierarchical architectures at high temperature, including the atomic Cl$_{\rm Sb}^{2+}$ point defects, nanoscale Cu$_{9}$In$_{4}$ and Sb precipitates, as well as the mesoscale Cu$_{2}$Sb particles. In addition, it is noted that the InSb + $x$ wt% CuCl ($x = 0,\, 3,\, 5,\, 8$) samples are stable at 600–733 K according to our $C_{p}$ results, as shown in Fig. S4. Figure 6(a) shows the $ZT$ of InSb + $x$ wt% CuCl samples. Due to the simultaneous optimization in PF and $\kappa$ at 733 K, the $ZT$ value of InSb was greatly improved, which increases from 0.22 for InSb to 0.77 for the $x = 5$ sample. Such an enhanced $ZT$ is also much higher than other InSb based compounds reported in Refs. [18–20], as shown in Fig. 6(b). In summary, we have demonstrated a simultaneous optimization in electrical and thermal properties of InSb by CuCl doping. It is found that CuCl can react with InSb during the SPS process, resulting in Cl to substitute Sb sites, and the formation of mesoscale Cu$_{2}$Sb, nanoscale Sb ($\sim $50 nm) and Cu$_{9}$In$_{4}$ (10–20 nm). Consequently, CuCl doping enables us to achieve a high power factor of 31.5 µW$\cdot$cm$^{-1}$$\cdot$K$^{-2}$ at 733 K and a low thermal conductivity of 2.99 W$\cdot$m$^{-1}$$\cdot$K$^{-1}$ at 733 K for the InSb + 5 wt% CuCl sample. Eventually, due to simultaneous optimization in electrical and thermal properties, the peak $ZT$ reaches a high value of 0.77 at 733 K for the InSb + 5 wt% CuCl sample, which is 250% higher than that of pristine InSb. References Complex thermoelectric materialsHigh-performance bulk thermoelectrics with all-scale hierarchical architecturesUltralow thermal conductivity and high thermoelectric figure of merit in SnSe crystalsElectronic Structures and Thermoelectric Properties of ZnSb Doped with Cd and In from First Principles CalculationsZintl Phase BaAgSb: Low Thermal Conductivity and High Performance Thermoelectric Material in Ab Initio CalculationDirectional Design of Materials Based on Multi-Objective Optimization: A Case Study of Two-Dimensional Thermoelectric SnSeTuning Thermal Conductivity in Si Nanowires with Patterned StructuresBipolar Thermoelectrical Transport of SnSe Nanoplate in Low Temperature *High-Pressure Synthesis and Thermal Transport Properties of Polycrystalline BAs x Power generation and thermoelectric cooling enabled by momentum and energy multiband alignmentsImprovement of Thermoelectric Performance in BiCuSeO Oxide by Ho Doping and Band ModulationPredicting the Potential Performance in P-Type SnS Crystals via Utilizing the Weighted Mobility and Quality FactorHigh Entropy Semiconductor AgMnGeSbTe 4 with Desirable Thermoelectric PerformanceHigh Thermoelectric Performance in the New Cubic Semiconductor AgSnSbSe 3 by High-Entropy EngineeringUltralow Thermal Conductivity and Thermoelectric Properties of Rb 2 Bi 8 Se 13High-Performance Thermoelectrics from Cellular Nanostructured Sb2Si2Te6InAs and InSb as Thermoelectric MaterialsThermoelectric properties and figure of merit of a Te-doped InSb bulk single crystalEffects of melt spinning process on microstructure and thermoelectric properties of n-type InSb compoundsEnhanced thermoelectric performance in In1−xGaxSb originating from the scattering of point defects and nanoinclusionSelf-compensation induced vacancies for significant phonon scattering in InSbNew insight into InSb-based thermoelectric materials: from a divorced eutectic design to a remarkably high thermoelectric performanceAn in situ eutectic remelting and oxide replacement reaction for superior thermoelectric performance of InSbCrossref OpenURL ResolverHigh Performance Thermoelectrics from Earth-Abundant Materials: Enhanced Figure of Merit in PbS by Second Phase NanostructuresSubtle Roles of Sb and S in Regulating the Thermoelectric Properties of N‐Type PbTe to High PerformanceInertial effective mass as an effective descriptor for thermoelectrics via data-driven evaluationEffective masses of electrons and heavy holes in InAs, InSb, GaSb, GaAs and some of their ternary compoundsThermoelectric properties and chlorine doping effect of In 4 Pb 0.01 Sn 0.03 Se 2.9 Cl x polycrystalline compounds
[1] Snyder G J and Toberer E S 2008 Nat. Mater. 7 105
[2] Biswas K, He J, Blum I D, Wu C I, Hogan T P, Seidman D N, Dravid V P, and Kanatzidis M G 2012 Nature 489 414
[3] Zhao L D, Lo S H, Zhang Y, Sun H, Tan G, Uher C, Wolverton C, Dravid V P, and Kanatzidis M G 2014 Nature 508 373
[4] Zhou K, Zhang T, Liu B, and Yao Y J 2020 Chin. Phys. Lett. 37 017102
[5] Wang S F, Zhang Z G, Wang B T, Zhang J R, and Wang F W 2021 Chin. Phys. Lett. 38 046301
[6] Yan S, Wang Y, Gao Z, Long Y, and Ren J 2021 Chin. Phys. Lett. 38 027301
[7] Zhu G P, Zhao C W, Wang X W, and Wang J 2021 Chin. Phys. Lett. 38 024401
[8] Zhou L Y, Zheng Q, Bao L H, and Liang W J 2020 Chin. Phys. Lett. 37 017301
[9] Gao L, Liu Q, Yang J, Wu Y, Liu Z, Qin S, Ye X, Jin S, Li G, Zhao H, Long Y 2020 Chin. Phys. Lett. 37 066202
[10] Qin B C, Wang D Y, Liu X X, Qin Y X, Dong J F, Luo J F, Li J W, Liu W, Tan G J, Tang X, Li J, He J, and Zhao L 2021 Science 373 556
[11] Feng B, Li G Q, Hu X M, Liu P H, Li R S, Zhang Y L, Li Y W, He Z, and Fan X A 2020 Chin. Phys. Lett. 37 037201
[12] He W, Qin B, and Zhao L D 2020 Chin. Phys. Lett. 37 087104
[13] Ma Z, Xu T, Li W, Cheng Y, Li J, Zhang D, Jiang Q, Luo Y, and Yang J 2021 Adv. Funct. Mater. 31 2103197
[14] Luo Y, Hao S, Cai S, Slade T J, Luo Z Z, Dravid V P, Wolverton C, Yan Q, and Kanatzidis M G 2020 J. Am. Chem. Soc. 142 15187
[15] Cai S, Hao S, Luo Y, Su X, Luo Z Z, Hu X, Wolverton C, Dravid V P, and Kanatzidis M G 2020 Chem. Mater. 32 3561
[16] Luo Y, Cai S, Hao S, Pielnhofer F, Hadar I, Luo Z Z, Xu J, Wolverton C, Dravid V P, Pfitzner A, Yan Q, and Kanatzidis M G 2020 Joule 4 159
[17] Bowers R, Ure R W, Bauerle J E, and Cornish A J 1959 J. Appl. Phys. 30 930
[18] Yamaguchi S, Matsumoto T, Yamazaki J, Kaiwa N, and Yamamoto A 2005 Appl. Phys. Lett. 87 201902
[19] Su X L, Tang X F, and Han L 2010 Acta Phys. Sin. 59 2860 (in Chinese)
[20] Zhang Q, Xiong Z, Jiang J, Li W, Xu G, Bai S, Cui P, and Chen L 2011 J. Mater. Chem. 21 12398
[21] Mao J, Niedziela J L, Wang Y, Xia Y, Ge B, Liu Z, Zhou J, Ren Z, Liu W, Chan M K Y, Chen G, Delaire O, Zhang Q, and Ren Z 2018 Nano Energy 48 189
[22] Cheng Y, Yang J, Jiang Q, He D, He J, Luo Y, Zhang D, Zhou Z, Ren Y, and Xin J 2017 J. Mater. Chem. A 5 5163
[23] Xin J, Jiang Q, Wen Y, Li S, Zhang J, Basit A, Shu L, Li X, and Yang J 2018 J. Mater. Chem. A 6 17049
[24] Luo Y, Yang J, Jiang Q, Li W, Zhang D, Zhou Z, Cheng Y, Ren Y, and He X 2016 Adv. Energy Mater. 6 1600007
[25] Zhao L D, Lo S H, He J, Li H, Biswas K, Androulakis J, Wu C I, Hogan T P, Chung D Y, Dravid V P, and Kanatzidis M G 2011 J. Am. Chem. Soc. 133 20476
[26] Tan G, Stoumpos C C, Wang S, Bailey T P, Zhao L D, Uher C, and Kanatzidis M G 2017 Adv. Energy Mater. 7 1700099
[27] Suwardi A, Bash D, Ng H K, Gomez J R, Repaka D V M, Kumar P, and Hippalgaonkar K 2019 J. Mater. Chem. A 7 23762
[28] Bouarissa N and Aourag H 1999 Infrared Phys. & Technol. 40 343
[29] Hee K J, Jae K M, Oh S, Rhyee J S, Park S D, and Ahn D 2015 Dalton Trans. 44 3185