ABSTRACT
The influence of bismuth substitution on the thermoelectric properties of AgSbTe2 compounds were investigated and compared with the undoped AgSbTe2. The addition of Bi dopants not only resulted in a reduction in thermal conductivity, but also significantly increased the thermopower in Ag(Sb1-xBix)Te2 series. It is clear that additional phonon scatterings were created by bismuth doping and leaded to the reduction of thermal conductivity. Meanwhile the thermal power was also enhanced, this was attributed to the electron filtering effects caused by the nanoscaled microstructures. Due to the extremely low thermal conductivity (~ 0. 48 Wm-1K-1) and moderate power factor were found in AgBi0. 05Sb0. 95Te2, the maximum ZT value 〜 1. 0 had been reached at 570 K, representing a 10% enhancement with respect to an undoped AgSbTe2, this results shows a promising thermoelectric properties in the medium temperature range.
Introduction
Energy and the environment have become the most critical issues in recent years. The urgent need for new sources of energy other than fossil fuels, as well as the most efficient use of current fossil fuel supply, has kindled significant researches on alternative energy sources and different types of energy conversion technologies. One of the types of energy conversion technologies that have gained renewed attention is thermoelectric energy conversion, where heat is converted directly into electricity using a class of materials known as thermoelectric materials. 1-7 Thermoelectric (TE) materials are becoming increasingly important in the field of electrical generation and environmental friendly refrigeration. 1, 2 The performance of TE devices is assessed by the dimensionless figure of merit (ZT), defined as ZT = α2σT/, where α, σ, T and are the Seebeck coefficient, the electrical conductivity, the absolute temperature and the thermal conductivity, respectively. Based on the above relation, the best performance TE materials should have high electrical conductivity, large Seebeck coefficient and low thermal conductivity. 1AgSbTe2 has been known as a potential thermoelectric material for a long period 8-16 due to its relatively low thermal conductivity (0. 6 ~ 0. 8 Wm-1K-1). 17, 18 AgSbTe2 is widely identified as the disordered NaCl type (Fm-3m) where Ag and Sb randomly occupy the Na sites. 19 According to the previous studies, 11 the disordered lattice structure dominantly contributes to low lattice thermal conductivity through Umklapp and intrinsic phonon-phonon scattering processes without significant reduction in the electrical conductivity. Recently, the AgSbTe2 compound has attracted much attention to construct the so called bulk nanostructured TE materials such as (AgSbTe2)1−x(GeTe)x (also called TAGS) and (AgSbTe2)1−x(PbTe)x (also called LAST) with excellent TE properties. 20-25Doping is a possible approach to optimize the thermoelectric properties of p-type AgSbTe2 by reducing its thermal conductivity and adjusting its carrier concentration. In this study, trivalent Bi3+ ions were selected to substitute Sb3+ ions in a p-type Ag(Sb1-xBix)Te2 systems in order to suppress lattice thermal conductivity dramatically while simultaneously contributing to the thermopower enhancement. The BiTe-AgSbTe2 solid solution has already been studied26 but its TE properties in the middle temperature range have so far not yet reported. We investigated the medium temperature thermoelectric properties of Ag(Sb1-xBix)Te2 (x = 0, 0. 03, 0. 05, 0. 07 and 0. 1) at 300-600 K. We found that the sample with x = 0. 05 exhibits a maximum ZT value〜1 at 570 K and therefore it’s a potential candidate for use as a p-type TE material.
Experimental
Synthesis
Polycrystalline Ag(Sb1-xBix)Te2 (x = 0, 0. 03, 0. 05, 0. 07 and 0. 1) were prepared by solid state reaction. High-purity starting elements of Ag (99. 995%, filament), Sb (99. 9999%, shot), Bi (99. 999%, chunks) and Te (99. 999%, shot), were melted in carbon coated quartz tubes with diameter of 1. 6 cm under the vacuum at 800 °C for 6 h, then slowly cooled to 475 °C, followed by a rapid quenched in water. The obtained ingots were pulverized to powder, the mixed powder then sealed in quartz tubes under a vacuum of about 10-4 Torr after multiple argon gas purging cycles, pretreated at 500 °C for 18 h in a box furnace, and furnace cooled to room temperature. Crystals were grown with a vertical Bridgman furnace starting from the pretreated powder and vacuum sealed in carbon coated quartz tubes of 10 cm length and 1. 6 cm in diameter, the temperature profile of the Bridgman furnace used for the whole series was maintained at 450–700 °C within 25 cm region. Initial complete melting was achieved at 700 °C for 24 h to ensure complete reaction and mixing. The temperature gradient of 1°Ccm-1 was programed around the solidification point near 555 °C, and the quartz tube was then slowly lowered into the cooling zone at a rate of ∼0. 5 mmh-1. Highly dense ingots with dark silvery metallic shine were obtained as shown in Fig. 1a. These ingots were stable in water and air. The obtained crystalline ingots were cut and polished into rectangular shapes of approximately 3x3x12 mm3, circular discs of a diameter of 12 mm and a thickness of 12 mm for later physical properties measurements as shown in Fig. 1b. The density of the ingots measured by the Archimedes method varied from 7. 11∼7. 12 gcm-3 which is ~ 99. 9% of the theoretical density. Fig. 1 (a) Bi doped AgSbTe2 crystal grown by Bridgman method and (b) its various cuts for thermo-electric measurements.
Structural and Morphology analysis
X-ray diffraction experiments were conducted for phase identifications using a powder X-ray diffractometer (X’Pert PRO-PANalytical, CuKα radiation) at 2θ angles of 20-80°. With these data, lattice parameters of Ag(Sb1-xBix)Te2 (x = 0, 0. 03, 0. 05, 0. 07 and 0. 1) samples were calculated by the Rietveld refinement program. Fractured surface morphology was characterized with a field emission scanning electron microscopy (FESEM, Hitachi, S-4800). The chemical composition of the as-prepared ingots was determined by wavelength dispersive X-ray fluorescence spectrometry (WD-XRF). Mirror-like polished specimens of Ag(Sb1-xBix)Te2 samples extracted from various locations of the ingot were analyzed using a Rigaku ZSX primus Ⅱ X-ray fluorescence spectrometer.
Thermoelectric Characterization
The Electrical conductivity (σ) and Seebeck coefficient (α) were measured simultaneously by commercial equipment (ZEM-3, ULVAC-RIKO, Japan) under a helium atmosphere from 300 to 600 K. The Hall Effect was measured at room temperature under 0. 55T with a four probe configuration of the ECOPIA HMS-5000 system using the van der Pauw method. The thermal conductivity (L) was determined as a function of temperature from room temperature to 600 K using the laser flash apparatus (NETZSCH, LFA 457). The front face of a small disk-shaped sample (diameter∼12 mm; thickness∼2 mm) coated with a thin layer of graphite is irradiated by a short laser burst, and the resulting rear face temperature rise is recorded and analyzed. Thermal conductivity κ was calculated using the equation κ= DCpd from the thermal diffusivity D obtained by a laser flash apparatus (NETZSCH, LFA 457), specific heat Cp determined by a differential scanning calorimetry method (NETZSCH, STA 449), and the density d obtained by the Archimedes method.
Results and discussion
The samples of Ag(Sb1-xBix)Te2 were obtained as crystalline ingots, they were then cut and polished for the transport property measurements presented here. Fig. 2a represents powder X-ray diffraction (XRD) patterns of samples with composition Ag(Sb1-xBix)Te2 (x = 0, 0. 03, 0. 05, 0. 07 and 0. 1). All diffraction peaks can be indexed into the face centered cubic (FCC) AgSbTe2 structure (reference code: 01-089-3671). For the undoped sample, weak diffraction peaks due to the Sb7Te impurity (reference code: 00-046-1068) that are often reported in the literature9 are detected. This indicates that the undoped sample consists of the major phase AgSbTe2 and the precipitated Sb7Te minor phase. However, as the substitution of Bi for Sb for x = 0 to 0. 1, the weak impurity diffraction peaks disappear completely in the background shown in (Fig. 2a). It suggests that, upon substituting Sb with Bi, the tendency to form the impurity phases of Sb7Te is completely suppressed. The lattice parameter as a function of Bi fraction is displayed in the inset of Fig. 2a, the lattice parameter increases with x increase, as expected based on the difference between the metallic radii of Bi (156 pm) and Sb (140 pm). The linear dependence of the lattice parameter versus x indicates that Bi is substituting Sb in the crystal lattice. To figure out the phase composition of the obtained samples accurately, the DSC heat flow curves were measured and are plotted in Fig. 2b. We note the presence of an endothermic peak that appears in undoped AgSbTe2 sample at 620 K. As for Bi-AST samples, the endothermic peak vanishes completely. Therefore, on the basis of the XRD and DSC analyses, we conclude that the undoped sample contains tiny amount of Sb7Te impurity, whereas Bi-AST samples are homogeneous single phase solid solutions. These results indicate that a small amount of Bi doping can inhibit the emergence of Sb7Te impurity phase quite effectively. Fig. 2 (a) Powder XRD patterns for AgBixSb1-xTe2 samples. The inset of (a) shows lattice parameter of Ag(Sb1-xBix)Te2 and (b) DSC curves of Ag(Sb1-xBix)Te2 for x = 0, 0. 03, 0. 05, 0. 07 and 0. 1. Fig. 4 (a) and (b) shows the FESEM photograph of the free fracture surface of the AgBi0. 05Sb0. 95Te2 sample. The magnified photo (Fig. 4b) displays highly compacted density, which is in very good agreement with the density measurements. Fig. 4 (a) SEM photographs of the bulk AgBi0. 05Sb0. 95Te2 and (b) its magnified portion of the square area in Fig. 4a. The average compositions of the Ag(Sb1-xBix)Te2 series obtained from wavelength dispersive X-ray fluorescence analysis were closely consistent with the nominal compositions (Table 1). The temperature dependence of the electrical conductivity σ measured in the range of 300-600 K is shown in Fig. 3a. For all samples, the electrical conductivity decreases with the increasing temperature and then increases again, indicating the semi-conducting transport behavior. The Bi-AST samples possess lower σ compared with the undoped AgSbTe2 compound at room temperature (table 1). Table 1 shows the properties used to describe the electron transport characteristics of Ag(Sb1-xBix)Te2 (x = 0, 0. 03, 0. 05, 0. 07 and 0. 1) compounds at room temperature. Compared with undoped sample, the Bi-AST samples display lower electrical conductivity, carrier concentration but higher mobility. For x = 0. 05, the value of electrical conductivity is higher than that of reported BiTe-AgsbTe2. 26The Seebeck coefficients of all the specimens are positive, as shown in Fig. 3b, indicating the p-type conduction. At room temperature, the Seebeck coefficients α for Bi-AST samples span from 172 to 245 µVK-1, higher than that of the undoped AgSbTe2 sample. For x = 0. 03, the value of Seebeck coefficient is comparable to that of reported BiTe-AgsbTe226 at room temperature. The Seebeck coefficient for metals or degenerate semiconductors, with the assumption of a parabolic band and energy independent scattering, can be expressed as27where kB is Boltzmann’s constant, q is the electronic charge, h is the Planck constant, n is the carrier concentration and m* is the effective mass. The effective masses were calculated with Eq. 1 and reported in table 1. In our work, the minimum Seebeck coefficient value ~ 164 µVK-1 was obtained for undoped AST sample with maximum effective mass value of 2. 08m0 at room temperature. So the increase in Seebeck coefficient value is not because of the increase of effective mass. The possible reason of the increase in Seebeck coefficient may be connected to the electron filtering effects caused by the nanoscaled microstructures. 11, 28The power factor P. F. = α2σ curves are plotted in Fig. 3c, the power factor initially increases, reaches maximum and then decreases with increasing temperature. The power factor for undoped AgSbTe2 sample appears to have the largest values over the full temperature range whereas Bi-AST samples have lower power factor except sample with x = 0. 05 has achieved moderate value which is comparable to that of reported BiTe-AgsbTe2. 26 Fig. 3 Temperature dependence of electrical transport properties for AgBixSb1-xTe2 samples (a) electrical conductivity, (b) Seebeck coefficient and (c) power factor. Table 1 Carrier concentration n, Hall mobility µH, electrical conductivity σ, Seebeck coefficient α, effective mass m*/mo and XRF composition of all samples at room temperatureNominal compositionXRF compositionn ( 1019cm-3)µH (cm2V-1s-1)σ(104Sm-1)α (µVK-1)m*/mo. AgSbTe2Ag0. 98Sb1. 016Te213. 816. 23. 61642. 08AgBi0. 03Sb0. 97Te2AgBi0. 024Sb0. 982Te22. 32320. 851720. 56AgBi0. 05Sb0. 95Te2AgBi0. 043Sb0. 967Te27. 229. 41. 102441. 0AgBi0. 07Sb0. 93Te2AgBi0. 067Sb0. 918Te23. 717. 472. 022011. 7AgBi0. 1Sb0. 9Te2AgBi0. 089Sb0. 83Te21. 6821. 101. 262221. 34Fig. 4a shows the temperature dependence of thermal diffusivity D and measured specific heat capacity Cp of Ag(Sb1-xBix)Te2 samples. From the specific heat data, the results closely matches the literature value (Cp ~ 0. 205 Jg-1K-1) of AgSbTe2. 17 Total thermal conductivity is calculated as a product of the measured specific heat at constant pressure Cp, thermal diffusivity D and material density d. Fig. 5b shows the temperature dependence of the total thermal conductivity of the Ag(Sb1-xBix)Te2 samples. For all samples the first decreases and then increases with the increasing temperature and the magnitude spans from 0. 48 to 0. 91 Wm-1K-1. The thermal conductivity of the sample with x = 0. 05 exhibits lowest thermal conductivity than all other samples over the temperature range from 300 to 600 K with the minimum 0. 48 Wm-1K-1 at 510 K. For the compositions of x = 0. 03 and x = 0. 05, the thermal conductivity at middle temperature range is significantly lower than that of reported BiTe-AgsbTe2. 26 Near room temperature, bulk of the heat is conducted by long-wavelength acoustic phonons8 and the rapid increase in thermal conductivity at high temperatures may be related to an enhancement in ambipolar thermal conductivity. 29 Using the Wiedmann–Franz law, e = LσT, where L is the Lorentz constant, the lattice thermal conductivity L can be estimated by subtracting the electronic thermal conductivity e from the total thermal conductivity, L = – LσT. For a degenerate semiconductor L = 0. 7L0 (L0 = π2/3(kB/e)2 = 2. 45×10-8 V2K-2). 29 As shown in Fig. 5c, all Bi-AST samples exhibit lower lattice thermal conductivity compared with undoped sample. It is noted that the sample with x = 0. 05 exhibits lowest lattice thermal conductivity than all samples with the minimum ~ 0. 38 Wm-1K-1 over the temperature range from 470 to 550 K, closing to the minimum theoretical thermal conductivity ~ 0. 3 Wm-1K-1 reported by Cahill et al. 30 From Fig 5b and 5c, it is obvious that the total thermal conductivity of this system is about 70-90% dominated by the lattice thermal conductivity, especially for the undoped samples. The significant decrease in the lattice thermal conductivity achieved in the samples with x = 0. 03 and x = 0. 05 compared to undoped AgSbTe2 can be attributed to large mass mismatch generated from the substitution of bismuth (209 amu) for antimony (121. 8 amu), as the bismuth atoms act as point defect scattering sites and interfere phonon wave propogation. 31Fig. 6 Temperature dependence of thermal transport properties for Ag(Sb1-xBix)Te2 samples (a) Diffusivity and specific heat, (b) total thermal conductivity and (c) lattice thermal conductivity. The TE figure of merit ZT is calculated from the measured values of σ, α and using the equation ZT = α2σT/. Fig. 6 shows the temperature dependence of ZT of all Ag(Sb1-xBix)Te2 (x = 0, 0. 03, 0. 05, 0. 07 and 0. 1) compounds. For all samples, ZT increases with increasing temperature. The highest figure of merit is observed for AgBi0. 05Sb0. 95Te2 where it reaches ZT 1. 0 at 570 K. This value is 10% higher than that of the undoped AgSbTe2 at the same temperature. We have also achieved ZT value 70% higher than of reported BiTe-AgsbTe226 at 420 K for AgBi0. 05Sb0. 95Te2 sample, these results shows a significant improvement in the TE figure of merit ZT for Bismuth doped AST samples. Fig. 6 Temperature dependence of thermoelectric figure of merit ZT of Ag(Sb1-xBix)Te2 samples.
Conclusion
The TE properties of Bi-doped Ag(Sb1-xBix)Te2 compounds have been investigated. XRD and DSC analysis indicate that single phase material crystallizing in a cubic NaCl- type structure in Bi-doped AgSbTe2 samples. The Seebeck coefficient increased and the electrical conductivity decreased with increasing Bi doping. The thermal conductivity was reduced remarkably by Bi doping and the lattice thermal conductivity reduction was mainly responsible for the total decrease in thermal conductivity. The lattice contribution had a larger effect on the total thermal conductivity than the electronic contribution at all temperatures examined. This result is due to additional phonon scatterings were created by bismuth doping. The maximum ZT value had been reached ~ 1. 0 for the sample AgBi0. 05Sb0. 95Te2 at 570 K, which is 10% higher than that of undoped AgSbTe2, showing promising thermoelectric properties in the medium temperature range.
Acknowledgement
This work was supported by Academia Sinica and National Science Council, Taiwan, Republic of China, under Grant No. NSC100-2112-M-001-019-MY3.