All chemicals are analytical grade products purchased from Shanghai Chemical Reagent Company and were used as received without further purification.

In a typical synthesis process, 0.3071 g (0.8 mmol) of In(NO₃)₃ and 0.2712 g (1.2 mmol) of Na₂TeO₃ were put into a Teflon-lined stainless steel autoclave of 50 mL capacity and dissolved in ethylene glycol (EG) (35.56 mL) under vigorous magnetic stirring to form a clear solution at room temperature for 1 h. Then, EDA (4.44 mL) was added into the mixed solution. The solution was stirred for 30 min again. Then the autoclave was closed and maintained at 200°C for 3, 6, 12, 18, and 24 h. After the treatment, the autoclave was cooled to room temperature naturally. The black flocculating product was collected from the solution by centrifugation, washed several times with absolute ethanol and ultrapure water with resistivity of 18 MΩ-cm, and then dried at 60°C in vacuum for 10 h; as a result, the black powders were obtained.

The structural properties of the as-prepared products were analyzed by power X-ray diffractometer, which were obtained using Bruker D8 Advance diffractometer operating at 40 kV and 40 mA (Cu Kα radiation, λ = 0.154178 nm). The morphology of the as-prepared products was analyzed by field-emission scanning electron microscopy (FE-SEM, Sirion 200, 10 kV). Transmission electron microscope (TEM) images and energy dispersive X-ray spectroscopy (EDX) were obtained at 200 kV using a JEM-2010 microscope by dropping a dilute ethanol solution of the powders onto the ultrathin carbon-coated copper grids.

The obtained powders were pressed under a pressure of 58 MPa for 5 min and further pressed with a pressure of 460 MPa for 30 min at room temperature to decrease the porosity of the bulk pellet. Therefore, a rectangular bar of the powders with dimensions 15.06 mm × 5.03 mm × 1.07 mm was obtained for electrical conductivity and Seebeck coefficient measurement. A four-probe method was adopted for electrical conductivity measurement illustrated in Figure 1b. Silver pastes dropped in two ends of the thermoelectric pellet were used as electrical contacts of the electrodes to the sample. The set up of Seebeck coefficient was illustrated in Figure 1d. To decrease the effect of contact resistance on experimental results, a temperature difference of about 1 to 4 K between cool and hot ends of the bulk pellet was used for Seebeck coefficient measurement. The temperature gradient was established in the sample when the electrical power was applied by a ceramic heater. The temperature differences (ΔT) were determined by the nickel chromium-nickel silicon thermocouples. The two thermocouples were contacted with two ends of the pellet to determine the temperature change. Electrodes fixed on two ends of the thermoelectric pellet were used to monitor the Seebeck voltage drop between the two electrodes. Seebeck coefficients were determined from the slope of plots of sample voltage versus ΔT: S = -ΔV/ΔT.

The morphology and size of the as-synthesized products were characterized by FE-SEM on a Cu substrate. The FE-SEM images show a large number of hierarchical structures, each of which consists of arrays of small nanoplatelets which is similar to nanostring-cluster structure. Figure 2a shows a FE-SEM image of the In₂Te₃ hierarchical structures with diameter from 100 nm to 2 μm and length from several to over 100 μm, under a typically solvothermal reaction with In(NO₃)₃ and Na₂TeO₃ as the reactants, and EDA as the reductant and complexing agent at 200°C for 24 h. Figure 2b shows that a typically well-oriented nanoplatelet in the hierarchical structures possesses an edge length of approx. 700 nm and a thickness of approx. 150 nm. The phase purity and crystallographic structure of the products were determined by X-ray powder diffraction (XRD) with Cu Kα radiation. Figure 3a shows the transformation from t-Te nanowires to In₂Te₃ hierarchical structures occurred with increasing the reaction duration. As shown in Figure 3a, the as-synthesized In₂Te₃ hierarchical structures at 200°C for 24 h exhibited three broad peaks at 2θ = 25.02, 41.40, and 49.02, which were assignable to diffractions of the (511), (822), and (933) planes, respectively. In addition, five weak peaks at 2θ = 27.58, 28.92, 59.92, 66.02, and 75.46 were assignable to diffractions of the (440), (600), (1200), (993), and (1266), respectively. The diffraction peaks can be indexed to the purely fcc phase of In₂Te₃ (space group: F4\overline 3 m, no. 216) with lattice constants of a = b = c = 1.848 nm (JCPDS card: 33-1488). No other peaks from any other phases of indium telluride were detected. The crystal structure, as seen in Figure 3b, shows strong covalent bonding within each layer and a weak van der Waals force between the layers. Thus, the appearance of nanoplatelets on the wires is understandable due to the planar sheetlike nature of the building blocks. In other words, growth of the nanoplatelets is actually the outward embodiment of the internal crystal structure in this case.

The detailed structure of the products was further characterized by TEM. Figure 4a exhibits the TEM image of a typical α-In₂Te₃ hierarchical structure. Reliable electron diffraction patterns and high-resolution TEM images are difficult to obtain because of the partial melting of the samples under electron-beam irradiation at 200 kV, which could be obviously seen in Figure 4a. However, XRD pattern (Figure 3a) and FE-SEM images (Figure 2) show that the obtained hierarchical structures are polycrystalline. EDX spectrum reveals an atomic ratio of In:Te = 42.81:57.19 (Figure 4b), close to the stoichiometric In₂Te₃ within experimental errors. The signals for Cu, Cr peak were originated from the substrate.

To understand the growth process of the In₂Te₃ hierarchical structures, the effects of the reaction duration, temperature, volume ratio of EDA/EG, concentration of In(NO₃)₃ on the resulting products were systematically investigated (Figure 3, 4, 5, 6 and 7). (1) Time-dependent experiments in EDA at 200°C were performed to gain insight into the formation process of the hierarchical structures. The obtained product was nearly pure hexagonal t-Te nanowires within 3 h at 200°C (Figures 3a and 5a). Some flower-like In₂Te₃ nanoparticles appeared on the surface of the nanowires when the reaction duration was increased to 6 h (Figure 5b). Prolonging the reaction duration to 12 h, the same results as that of 6 h were produced (Figures 3a and 5c). Extending the reaction duration to 18 h caused smooth nanowires to grow into the wires with rough surface composed of nanoparticles (Figure 5d). The above results show that the intermediate morphology composed of the Te nanowires and In₂Te₃ nanoplates can be produced in 6 and 12 h. Well-defined In₂Te₃ hierarchical structures were obtained when prolonging the reaction duration up to 24 h (Figures 2 and 3a). Besides the ordered nanoplates, some nanowires were detected in the nanostructure, which suggests that the nanoplates nucleate and grow out of the nanowires. (2) When aged at 160°C over 48 h, the nearly pure hexagonal t-Te nanowires were only formed (Figures 6b and 7b), but the FE-SEM images (Figure 6c) shows that some flower-like In₂Te₃ nanoparticles appeared on the surface of the nanowires. When aged at 220°C over 24 h, the mixture of nanowires and nanoplates was produced, as shown in Figure 6d. Figure 7c exhibits the mixture mainly composed of t-Te nanowires. The indexed Te peaks are in good agreement with the standard literature data (JCPDF card number: 36-1452) in Figure 7. (3) The volume ratio of EDA/EG also plays an important role in structure and phase control of the resulting products (not shown). When pure EDA or EG was used in the reaction, pure cubic In₂Te₃ could not be got (not shown). Additionally, when the volume ratios of EDA/EG were 1:2, 1:12, or 1:16, pure cubic In₂Te₃ could also not be obtained. (4) When the other conditions remained the same, the concentration of In(NO₃)₃ was 13.25 mM, the obtained products were a mixture of In₂Te₃ and t-Te (not shown).

On the basis of the above experimental results, a diffusion-limited reaction mechanism can be reasonably used to explain the formation of the In₂Te₃ hierarchical structures. The chemical reactions to form the hierarchical structures are formulated as follows:

In the reaction system, acetaldehyde is produced by dehydration of EG at high temperature, as shown in Equation 1. The EG acted as both solvent and reductant in this case, and ions were reduced by EG to form elemental Te at the beginning of the reaction, which is shown in Equation 2. The reaction (2) was a quick step, which determined the formation of t-Te nanowires at the beginning of the reaction. The EDA plays an important role in the transformation process from the t-Te nanowires to the hierarchical In₂Te₃ nanowires because strongly basic solvents, such as ammonia and EDA, can reduce Te into Te^2- with a low speed (Equation 3). This leads to the formation of nanoporous structures on the surface of the nanowires (Figure 5b, c); meanwhile, a complex is formed by the reaction of the solvent EDA molecule and the metal In³⁺ ions. In this complex, each In³⁺ is surrounded by six NH₂ group, similar to Co³⁺ 34 . Although In³⁺ is combined with the t-Te and reduced into indium by the EDA in situ near the surface of the template t-Te nanowires at high temperature and high pressure, as Equation 4, In³⁺ has little solubility in EDA at room temperature and atmosphere 35 . Thus, indium atoms can quickly diffuse into the t-Te nanowires. With increasing the reaction duration, the molar ratio of In:Te is close to the stoichiometric composition of In₂Te₃ (Figure 4b). The transformation mechanism of the t-Te nanowires to the In₂Te₃ hierarchical structures is shown in Figure 8.

To evaluate thermoelectric properties of the In₂Te₃ hierarchical structures, the room-temperature electrical conductivity and Seebeck coefficient of the bulk pellets were measured, which were pressed using the as-prepared powders at 460 MPa for 30 min. The Seebeck voltage measured between cool and hot ends of the bulk pellet varies linearly with the temperature difference. The slope yields a Seebeck coefficient of -300 μV/K (Figure 1c). The negative sign indicates that the bulk pellet behaves as a n-type semiconductor, which is consistent with the EDX result (Figure 4b). The Seebeck coefficient of the bulk pellet is about 43% enhancement over 210 μV/K of the reported corresponding best thin film sample (seen in ref. 25 ). Additionally, the bulk pellet exhibits a linear current-voltage (I-V) curve (see Figure 1a) that is symmetric about the origin, indicating that the contacts are ohmic. The slope yields a resistance of 436 Ω. An electrical conductivity σ of 6.42 Ω^-1 m^-1 can be obtained by using an average bulk pellet thickness in a 1D electrical transport model, which is one to two orders of magnitude higher than that of the corresponding thin film sample (0.66 Ω^-1 m^-1) 26 36 , and four orders of magnitude higher than that of the corresponding p-type bulk sample 37 . In addition, bulk In₂Te₃ has very small thermal conductivity about 1.4 W (m K)^-1 25 . The thermal conductivity will further decrease by nanostructuring corresponding thermoelectric materials 2 38 .