Mechanical and Thermoelectric Properties of Bulk AlSb Synthesized by Controlled Melting, Pulverizing and Subsequent Vacuum Hot Pressing

Featured Application

Potential thermoelectric material for conversion of waste heat to electricity.

Abstract

Aluminum antimonide is a semiconductor of the Group III-V order. With a wide indirect band gap, AlSb is one of the least discovered of this family of semiconductors. Bulk synthesis of AlSb has been reported on numerous occasions, but obtaining a single phase has always proven to be extremely difficult. This work reports a simple method for the synthesis of single-phase AlSb. Subsequently, consolidation was done into a near single-phase highly dense semiconductor in a form usable for thermoelectric applications. Further, the thermoelectric properties of this system are accounted for the first time. In addition, the mechanical properties of the intermetallic compound are briefly discussed for a possibility of further use.

1. Introduction

The semiconductor aluminum antimonide is an intermetallic alloy compound and is known because of its high band gap energy. Both aluminum and antimony are abundant on the earth’s crust and are known as eco-friendly materials. The semiconductor has a large band gap of 1.69 eV and the nature of the band gap is indirect [1]. Consequently, there has not been much interest in it in terms of thermoelectric properties. This work, for the first time, assesses the possibility of the semiconductor for use in a thermoelectric device. The constituents of the compound are also very cheap in comparison to the thermoelectric devices in study today.

The largest problem in the study of the thermoelectric properties of AlSb is the synthesis of a bulk single-phase alloy. Aluminum is a very reactive metal and gets readily oxidized in air to form aluminum (III) oxide. Antimony, on the other hand, has a high vapor pressure and tends to escape from its surface in any physical form. AlSb is a line compound, meaning any change in stoichiometry can result in remnant elements remaining in the alloy, thus disrupting the formation of a single phase [2]. Owing to this, the synthesis of bulk AlSb has proven to be very difficult in the past. Mechanical alloying (MA) has been used in the past to synthesize the intermetallic compound, but not much success has been achieved in terms of obtaining a single phase [3,4,5,6,7,8]. This could be largely due to the reasons stated above. Trichês et al. showed in their work that as milled powders show an excess of antimony alongside the AlSb synthesized [4]. This could be because aluminum is a very ductile material and can get stuck on the wall of the vial and balls, resulting in a change of stoichiometry of the vial constituents. On the other hand, casting is a very basic method, but it always has the possibility of oxidation of aluminum and sublimation of antimony particles. To avoid oxidation, Karati et al. used the vacuum arc melting method to melt and synthesize the alloy, but the 50-50 at % composition showed an excess of aluminum [8]. So, after adding three at % excess of Sb to compensate for antimony sublimation, they succeeded in giving single-phase AlSb. In this work, melting was done in a controlled atmosphere of argon gas. Being an intermetallic compound, AlSb is very brittle in nature, and so, in order to be used or tested for thermoelectric applications, they need to be compacted. Another problem associated with melting is the uneven morphology within the solid structure. In order to prevent that and facilitate energy conversion use, powder metallurgy steps like pulverizing and sintering by a vacuum hot press could be effective. AlSb has been extensively studied as an anode material for Li-ion batteries, but they have not been studied as a potential thermoelectric material in the bulk form before.

Thermoelectric property of a material system is the efficiency of the system to convert heat into electricity. With much potential in the future, research in the field of thermoelectricity is inspired by the huge amount of energy wasted during all and any industrial processes. Thermoelectric properties are measured and understood by a dimensionless figure of merit called ZT [9]. The higher the value of ZT is, the better the thermoelectric property. ZT is defined by the following equation:

$$\mathrm{ZT}=\frac{{\mathrm{S}}^2\mathsf{\sigma }}{\mathsf{\kappa }}\mathrm{T}$$

(1)

where S is the Seebeck coefficient, σ is the electrical conductivity, τ is the total thermal conductivity and T is the absolute temperature. Increasing the value of ZT is more complicated than it looks from Equation (1). Seebeck coefficient and electrical conductivity, both directly proportional to ZT, are related to one another by the carrier concentration of the material. The higher the carrier concentration, the higher the electrical conductivity and the lower the Seebeck coefficient. So, an optimum value for S and σ are needed to give a high ZT value. A measure of such optimization is called the power factor and is given by

$$\mathrm{PF}={\mathrm{S}}^2\mathsf{\sigma }$$

(2)

Intrinsic AlSb is known to have p-type conductivity [10]. Formerly, some dopant studies showed that Be- and Si-doped AlSb continue to show p-type behavior [11] while Se- and Te-doped AlSb have shown n-type conductive characteristics [12]. However, nowhere before was the ZT or any other thermoelectric property measure reported for bulk AlSb.

Intermetallic compounds, such as AlSb, generally have poor fracture toughness [13]. As a measure of the mechanical properties, Vickers micro-hardness test was done. The test is conducted by exerting a certain amount of load on the surface of the sample using a diamond tip [14]. The size of mark the load leaves and the amount of load used are used to calculate the H_V value, a measure of the mechanical strength of the material. Intermetallic compounds are known to demonstrate H_V values within a few hundred.

2. Materials and Methods

Aluminum (99.9% purity, Aldrich) and antimony shots (99.999% purity, Kojundo) were weighed in a glove box with inert (Ar gas) atmosphere. A graphite crucible with stopper was used for the controlled melting. The inner walls were coated with a thick layer of boron nitride, and an argon gas environment was maintained inside. The prepared crucible with sample was heated to 1273 K in a vacuum furnace. The cooled ingot was then pulverized either by mortar-pestle (MP) or by a high energy vibratory mill (HEVM; KMTech TMM-70, Korea) using zirconia vial and balls (5 mm). The ball to sample ratio was maintained at 10:1. A vibratory mill was used for an hour at 1080 rpm to ensure better pulverization. Then the samples were sieved using a 325-mesh sieve. The samples were consolidated by Vacuum hot pressing (VHP) at 80 MPa pressure and 1173 K temperature for 6 h. BN coated graphite die of 10 mm diameter was used for this process.

Post-synthesis, phase transformation was analyzed using Cu-Kα radiation of an X-ray diffractometer (XRD; BRUKER AXS Advance D-8, Germany) and surface morphology was investigated using a scanning electron microscopy (SEM; Quanta-400, Netherlands). A cylindrical sample of 3 × 3 × 10 cm³ was used to measure Seebeck coefficient and electrical conductivity over the range of 300-873 K by ZEM-3 (ULVAC-RIKO, Japan), which runs by the four-probe method. Thermal diffusivity was measured by laser flash method using TC-9000H (ULVAC-RIKO, Japan). The equation κ = ρ × C_p × d was used to calculate thermal conductivity, where ρ is the density calculated using the Archimedes principle, C_p is the specific heat capacity, and d is thermal diffusivity. The measured values were then put into the Equation (1) to give the dimensionless figure of merit. VHPed samples were also mounted into a polymer matrix and polished to test for mechanical properties. Vickers micro-hardness was measured using PMT-X7B (MATSUZAWA, Japan). A load of 50 g-force (490.33 mN) was applied.

3. Results and Discussion

Figure 1a shows the XRD peaks of the as-casted samples. When fifty-fifty atomic percent of aluminum and antimony were maintained, the melting shows an excess of antimony in the XRD image. Since the melting was done in a controlled condition, no sublimation of antimony was observed. Yet, some amount of remnant oxygen, in considerably small quantities, might have oxidized aluminum, leading to the deficit of aluminum. Most XRDs have a detection limit of up to 5% by weight [15], which is why trace amounts of alumina could have remained undetected. Another explanation could be that the difference of stoichiometry may be caused by the difference of purity of the two elemental shots used. The extra high purity of Sb could have resulted in a deficit of aluminum, which needed to be compensated. In order to compensate for this loss, two at % excess of aluminum were added and the melting was run again. XRD peaks with two at % excess of Al as-casted samples showed a single-phase bulk AlSb. Despite that, the intermetallic semiconductor was extremely brittle, and no thermoelectric properties could be measured. Consequently, the samples were pulverized using mortar and pestle and consolidated by VHP. After VHP, two at % excess Al sample showed remnant peaks of excessive Sb in the matrix. Figure 1b shows the XRD data for vacuum hot-pressed samples. The discrepancy might be due to the formerly mentioned limitation of X-ray diffraction techniques. To further compensate, three at % of aluminum was added, and the process was repeated. Thereby, both the as-casted and VHPed samples showed single-phase bulk AlSb. In order to highlight minor peaks in this AlSb system further, the y-axis in the XRD data was modified to a logarithmic scale. Figure 1c,d are the respective logarithmic scale data for as casted and VHPed samples. After the log has been added, a small peak of remnant Sb can be seen in Figure 1d, which was absent in Figure 1b.

The relative densities of the VHPed samples in both the cases were ~94%, as shown in

Table 1. Relative densities of the VHPed samples under different pulverization techniques.

Sample	Description	Relative Density (%)
S12 *	Pulverized by MP and then VHP	94
S23 *	Pulverized by MP and then VHP	94
S33 *	Pulverized by HEVM and then VHP	99

* S1_x: x denotes x at. % excess of Al used.

. SEM images in Figure 2 show the existence of voids in the morphology. In order to further increase the density and reduce such voids, the casted sample was pulverized by HEVM. X-ray diffraction data show broadening of the peaks, which could be due to a decrease in particle size [3]. The milled powders were then consolidated by vacuum hot pressing, and relative density of ~99% was achieved. SEM images confirm that after vibratory milling, the shapes of the particles changed to near round shapes as opposed to the irregular shapes found by manual pulverizing.

Due to the relatively large band gap, the electrical resistivity of the single-phase bulk AlSb material is very high. Consequently, the thermoelectric properties of such material could not be calculated by the ZEM-3 machine. This could be because a large amount of energy is needed for the electrons at the valence band to be promoted to the conduction band. However, the semiconductor showed enough electrical conductivity for thermoelectric measurements when small amounts of remnant antimony would remain within the matrix. This electrical conductivity could be due to point defects in the compound. Aluminum vacancies may be responsible for the increased carrier movement. The transport properties of the sample S1₂ from Table 1 are shown in Figure 3a–d. The highest Seebeck coefficient of the S1₂ sample was found to be 370 µV/K at 855 K. This high Seebeck coefficient might have been caused by a very low carrier concentration, which could also explain the low electrical conductivity. The highest electrical conductivity was 6.6 S/cm at 660 K. The electrical conductivity of the sample increased with increasing temperature as shown in Figure 3b, but after 660 K, the conductivity decreased with temperature. The conductivity initially increased because, with higher temperature, more electrons gained the energy to jump from the valence to conduction band. However, after 660 K, interactions of the carriers with vibrating atoms and phonons caused scattering of the electrons, thus reducing the overall conductivity, as is the case for metals [16]. In addition, the slight decrease in electrical conductivity could also be a result of the minority carriers across the band gap. At a high temperature, the carrier concentration of the minority carriers would also rise, thus increasing the overall electrical resistivity of the semiconductor.

The thermal conductivity of the AlSb was found to be decreasing with increasing temperature. The lowest thermal conductivity was measured to be 4.7 W/mK at 873 K, as shown in Figure 3c. This thermal conductivity is rather high for applications in thermoelectricity and needs considerable reduction. The dimensionless figure of merit, ZT, was found to be the highest at 873 K and the value was 0.015. The ZT value was significantly low and might be largely due to very low electrical conductivity and high thermal conductivity. However, this was the first time that such properties of bulk aluminum antimonide were elucidated. With further engineering, the thermoelectric properties of this cheap and environmentally friendly semiconductor can be improved further.

The thermoelectric properties of InSb and GaSb have been extensively studied. Pristine InSb is known to exhibit a ZT of 0.25, while improvements have resulted in ZT value as large as 1.28 at 773 K [17]. On the other hand, the thermoelectric properties of GaSb have been improved up to 0.23 at 873 K [18]. In comparison to the other antimony compounds of group III, the thermoelectric properties of AlSb are very low. It can be assumed that doping and nanostructuring in the AlSb system matrix might improve such properties of AlSb. As a continuation of this study, doping and subsequent improvement of thermoelectric properties are being planned.

The as-casted AlSb was found to be very fragile. In fact, no electrical or thermal properties could be measured from it. The VHPed samples were also very brittle, but their superior microstructure, particularly after HEVM, made them strong enough for measurement of such properties. The mechanical properties were measured in terms of Vickers pyramid number (H_V). The mechanical properties of the highly dense sample S3₃ from Table 1 were measured in terms of Vickers pyramid number (H_V). Figure 4 shows a scanning electron microscopy image of the dent made by microhardness test. An H_V value of 350 ± 3.7 was found for the VHPed sample, showing moderately brittle behavior.

4. Conclusions

In an attempt to establish the true thermoelectric power of bulk AlSb, this study outlines a method for the synthesis and subsequent consolidation of a single-phase compound. Numerous studies have been attempted in the past to synthesize AlSb. A highly dense, bulk near single-phase AlSb has been synthesized in this work. Though the thermoelectric property of single-phase AlSb could not be measured with the available resources, the presence of a small amount of excess antimony in the system allowed for such measurements and for the first time, a thermoelectric figure of merit for AlSb is reported. The value is very low and can be deemed insignificant, yet the material system is extremely inexpensive and environmentally friendly. Possible nanoengineering and use of suitable dopants may be able to increase the thermoelectric ability of AlSb to a much greater extent. In addition, this work also reports the first practical results of the mechanical strength of the III-V semiconductor.