Introduction

Thermoelectric technology enables a direct conversion between electricity and thermal energy for both power generation and cooling1. The performance is strongly dependent on the material’s figure of merit zT = S2T/ρκ, in which S, ρ, T, and κ are Seebeck coefficient, electrical resistivity, absolute temperature, and thermal conductivity respectively.

While progress has been notable in mid- and high-temperature materials, such as IV–VI compounds including PbTe2, GeTe3, and SnSe4, skutterudites5,6, Mg2Sn7, and half-Heuslers8, advancements in near-room-temperature materials have been relatively slow. Bi2Te3 has stood out as the sole commercialized thermoelectric materials9,10,11,12 for longer than half a century. Because of the strong layered structure, Bi2Te3 largely relies on texturization to utilize its high in-plane zT13, leaving a substantial challenge in toughness.

Recently, several promising alternatives to commercial Bi2Te3 have been reported10,11,14,15,16, such as Mg3Sb210,11,14, MgAgSb14,15, and Ag2Se17,18 which exhibit compatible materials’ zT and offer greater sustainability. This motivated a few studies reporting impressive power generation efficiency and/or cooling performance using these materials, including n-Mg3Sb2/p-Bi2Te311, p-MgAgSb/n-Mg3.2Bi1.5Sb0.514, n-Mg3(BiSb)2/p-MgAgSb19, and n-Mg3Sb2/p-CdSb20 combinations.

Ag2Se was initially investigated as a thermoelectric material in the 1960s21 and has since been improved to show a zT above 0.7 near room temperature22,23,24,25. Ag2Se undergoes a phase transition from the low-temperature orthorhombic phase to the high-temperature cubic phase at ~406 K26. The high zT was usually realized in the orthorhombic phase of Ag2Se and was mainly attributed to its high carrier mobility and low lattice thermal conductivity27. However, phase transitions are typically undesirable as they may result in volume variations, which could lead to structural damage either within the material itself or at the interface between the material and electrodes during service. This somewhat limited the research on Ag2Se to focus on exploring its material properties23,25,28 and fabricating film devices specifically designed to operate at room temperature17,18,29,30,31. There are few reports on the power generation and cooling performance of Ag2Se bulk modules. This motivates the current work to focus on exploring device properties of bulk Ag2Se bellowing its phase transition temperature.

In addition to thermoelectric performance, mechanical properties are of equal importance to withstand loading. However, most inorganic thermoelectrics32,33,34,35, including Bi2Te336 are intrinsically brittle of their strong bond ionicity and/or covalency, which is therefore challenging for durable serviceability. Fortunately, Ag2Se26,37 was found to show plasticity, which indicates a great advantage for tough thermoelectric applications near room temperature.

This work focuses on bulk Ag2Se devices below its phase transition temperature. Ag2Se bars with a high average zT of 0.7 within 300–380 K were fabricated using a one-step hot-pressing process, to enable a contact resistivity as low as 12 µΩ cm2 using Ni and Ag as electrodes. These electrode-bonded bars were then paired with commercial p-Bi2Te3 legs for fabricating modules (7 pairs, 12 × 12 mm), to enable a power generation efficiency ηmax of >1% at ΔT = 50 K and a maximum cooling temperature ∆Tmax of >50 K. In addition to offering the above device performance that is highly competitive to commercial Bi2Te3 modules, Ag2Se has a distinct advantage in toughness indicated by its large fracture strain by order of magnitude. This offers the great potential of the Ag2Se for efficient and durable thermoelectric applications near room temperature.

Results and discussion

The details of material synthesis, module fabrication (dimensions in Table S1), characterizations, and property measurements (including the setup in Fig. S1) are given in the Methods. Three cylinders, with or without Ni/Ag electrodes, were hot-pressed in this work. As shown in Fig. 1, the hot-pressed cylinder without electrode is cut parallel (for XRD, S, ρ, κ measurements, bending and compression tests) and perpendicular (for XRD, S, ρ, κ measurements, and bending test) to the pressure direction. The cylinders with electrodes are sliced along the pressure direction for module fabrication.

Fig. 1: Schematic of the fabrication and slicing.
figure 1

a Schematic of the one-step hot-pressing, schematic slicing diagram of b the cylinder for property measurements and c the cylinder with electrodes for module assembly. Photograph of Ag2Se legs and the contact structure are also shown (c).

Full size image

X-ray diffraction (XRD) patterns for hot-pressed Ag2Se pellets cut along directions perpendicular (⊥) and parallel (//) to that of pressure applied during hot pressing are shown in Fig. S2a. All the diffraction peaks can be well indexed to Ag2Se of an orthorhombic structure, with no observable impurity peaks, thus affirming the purity of the samples. In addition, the similar relative intensities of diffraction peaks observed in Ag2Se (⊥) and Ag2Se (//) samples suggest the absence of obvious texture, indicating the transport-property isotropy in these polycrystalline materials. The scanning electron microscopy (SEM) images and the corresponding energy dispersive spectroscopy (EDS) mapping (Fig. S2b) further corroborate the high purity and homogeneity. The measured differential scanning calorimetry (DSC) curve (Fig. S2c) showing an endothermic peak at 406 K agrees well with the reported phase transition temperature of Ag2Se38.

Temperature-dependent thermoelectric transport properties for Ag2Se (⊥) and Ag2Se (//) samples were measured and shown in Fig. S3. Note that all measurements in this work were carried out under 380 K to avoid the phase transition at 406 K, which ensures the measurement repeatability. Both electronic and thermal transport properties of Ag2Se (⊥) and Ag2Se (//) are measured to be isotropic. The transport properties can be well described by a single parabolic band (SPB) model with an acoustic scattering (Fig. S3b, c). An average zT of 0.7 within 300–380 K was achieved in Ag2Se (//), which is comparable to available results of polycrystalline Ag2Se17,22,23,26,28,39,40 (Fig. S3f) and to that of commercial n-type Bi2Te341,42.

Vickers hardness, three-point bending, and compression tests were conducted on both commercial Bi2Te3 by an extrusion technique and as-fabricated Ag2Se at room temperature to evaluate their mechanical properties. The Vickers hardness of Ag2Se is measured to be ~35 kgf mm−1 (Fig. S4a) along both perpendicular (⊥) and parallel (//) directions, which is consistent with the literature results26,28. The Vickers hardness of n- and p-type Bi2Te3 along the parallel (//) direction is measured to be 55 kgf mm−1 and 48 kgf mm−1, respectively, which is slightly higher than that measured along the perpendicular (⊥) direction. The difference can be understood by the orientation preference of the extruded materials as confirmed by XRD results and the calculated orientation factor F(110), which is 0.18 for n-type and 0.10 for p-type Bi2Te3, as shown in Fig. S5.

While the hardness of Ag2Se may be inferior to that of Bi2Te3, its bending strength and compressive strength are significantly higher than those of Bi2Te3. This largely benefits machining and device operation. As shown in Fig. 2a, b, both n- and p-type commercial Bi2Te3 are unable to withstand bending strains above 0.5% or compressive strains above 2.5% and exhibit brittle fractures. Note that although zone-melted n-type Bi2Te3 has inferior mechanical properties compared to the p-type43, the commercial Bi2Te3 used in this work was prepared by the hot extrusion technique. It has been proven that the mechanical properties of n-type Bi2Te3 are better than those of p-type43. In strong contrast, Ag2Se allows additional large plastic deformation (Fig. 2c) to enable an ultimate bending strain of 4% at 128 MPa. Similarly, during the compression test (Fig. 2d), Ag2Se exhibits an elastic strain of 2% before yielding at ~50 MPa, followed by plastic deformation to enable a final strain as large as 40% at 273 MPa. The significant difference in mechanical properties can also be clearly reflected by the microstructures of the fracture surface (Fig. S6). Intergranular crack paths are observed at the fracture surface of Bi2Te3 during both bending and compression tests, indicating brittle fracture. In the case of Ag2Se, although macroscopic cracks are observed at the surface, the bulk sample is not completely fractured, demonstrating excellent toughness.

Fig. 2: Mechanical properties.
figure 2

a, c Stress–strain curves for bending and b, d for compression tests for commercial Bi2Te3 by extrusion and as-fabricated Ag2Se by hot-pressing at room temperature.

Full size image

To optimize the module performance, minimizing contact resistance is essential, which ensures a low device internal resistance (Rin), thereby maximizing both the output power (Pmax) and conversion efficiency (η) of the device. In this work, both Ni and Ag were found to facilitate low contact resistance when directly bonded to Ag2Se by hot-pressing. The electrical contact resistance (Rc) at Ag2Se/Ni (module 1) and Ag2Se/Ag (module 2) joints are estimated to be 0.3 mΩ (Figs. S7a and 7b). This corresponds to an interfacial contact resistivity (ρc) as low as 12 µΩ cm2, leading the total contact resistance to be ≤10% of the Rin of the leg. A robust bonding without any cracks is confirmed by SEM observations taken after the hot pressing and welding processes (Fig. S8). The total resistances of these modules are shown in Fig. S9.

With a fixed cold side temperature of ~285 K for power generation, the open-circuit voltage VOC, maximum output power Pmax, heat flow Q, and maximum conversion efficiency ηmax versus different temperature gradients (ΔT) are shown in Fig. 3 for both modules and commercial Bi2Te3 module. The measured voltage V, P, and Q as a function of current I for these modules at different ΔT are presented in Fig. S10. It can be seen that both modules in this work enable quite competitive performance with that of the commercial one, particularly within a ΔT of 50 K. Long-term efficiency measurements at a ΔT of ~85 K are performed on the modules to check the thermal stability. Although the linear coefficient of thermal expansion (CTE) of Ag2Se differs from p-Bi2Te3 at 300–373 K (Fig. S11), no obvious degradation in ηmax, Pmax, VOC, and Rin is observed for module 1 (Ni electrode) after continuous measurement for 30 days (Fig. 3e). However, the Rin of module 2 (Ag electrode) increases obviously as the measurement time progresses (Fig. S12), leading to a notable degradation in both output power and efficiency. This indicates that using Ni as electrodes for Ag2Se enables superior long-term stability of the module compared to using Ag as electrodes.

Fig. 3: Power generation performance.
figure 3

a Open-circuit voltage (VOC), b maximum output power (Pmax), c heat flow (Q), and d conversion efficiency (ηmax) as a function of different temperature gradients (ΔT) for n-Ag2Se/p-Bi2Te3 modules and commercial Bi2Te3 one. Literature results are included for comparison9,10,15,16,20,42,47,48,49, e ηmax, Q, Pmax, VOC, internal resistance Rin and ΔT of module 1 during continuous measurements for 30 days at ΔT of ~85 K.

Full size image

The cooling capability of both modules at a fixed hot-side temperature of ~300 K is shown in Fig. 4. The measured cold-side temperatures (Tc) under different input current I are provided in Fig. S13. The maximum cooling temperature difference ΔTmax reaches ~56 Κ, which is quite comparable to that of a commercial one. Similar comparability (Fig. 4a–c) can also be seen from the measurements on current dependent maximum cooling power and coefficient of performance (COP), as well as on ΔT dependent maximum coefficient of performance (COPmax), indicating equivalent energy consumption while pumping similar amount of heats. Although module 2 (Ag electrode) showed unsatisfactory stability during long-term power generation measurements, its cooling performance was found to be quite stable after 34 days of measurements (Fig. 4e). Note that module 1 performs better than module 2 in both power generation and cooling applications. Since the contact resistivity of Ag2Se/Ni and Ag2Se/Ag are very close (Fig. S7), the reason for the better performance of module 1 is presumed to be that the Ag2Se in module 1 has a slightly lower carrier concentration that is closer to the optimal value. This is evidenced by the higher voltage output and internal resistance of module 1 (Fig. S10). In addition, the device performance still has room for further improvement through geometric optimization according to the numerical simulation (Fig. S14).

Fig. 4: Cooling performance.
figure 4

a measured maximum cooling temperature difference (ΔTmax), b current dependent maximum cooling power (Qcmax), and c coefficient of performance (COP), as well as d maximum COP as a function of different ΔT for the modules50, e cold-side temperature (Tc), hot-side temperature (Th) and corresponding cooling temperature difference (ΔT) of n-Ag2Se/p-Bi2Te3 module 2 at a given current I of 4.3 A during duration measurements for 34 days, showing good stability. During the cooling duration tests, we run the test once per day for a continuous period of 22 h.

Full size image

In summary, bulk Ag2Se is ensured to show quite comparable near-room-temperature thermoelectric properties to commercial n-Bi2Te3, at both material and device levels for both power generation and cooling applications. Long-term measurements for over one month demonstrate that Ni is a better choice than Ag as electrodes for Ag2Se, ensuring good thermal stability. Most importantly, mechanical evaluations demonstrate the much more superior toughness in Ag2Se, highlighting the great capability to address the historical challenge for durable and large-scale thermoelectric applications near room temperature.

Methods

Synthesis

Polycrystalline Ag2Se was synthesized using high purity (>99.99%) of Ag and Se granules, weighted according to the stoichiometric ratio of Ag:Se = 2:1, loaded into the quartz tubes and sealed under vacuum. The raw materials were heated to 1273 K in 7 hours and kept at this temperature for 8 hours, then cooled down to 773 K in 5 hours and kept 773 K for 48 h followed by furnace cooling to room temperature. The obtained ingots were grinded and then densified by hot pressing. Dense Ag2Se pellets, Ni/Ag2Se/Ni, and Ag/Ag2Se/Ag cylinders with ~12 mm in diameter were obtained by hot pressing at 573 K for 20 minutes under a uniaxial pressure of ~60 MPa. As shown in Fig. 1, the Ni/Ag2Se/Ni and Ag/Ag2Se/Ag legs were sliced from the obtained Ni/Ag2Se/Ni and Ag/Ag2Se/Ag cylinders with a dimension of 2 × 2 × 4 mm3, the same size of legs in commercial Bi2Te3 modules. We used low-temperature solder (In52Sn48, with a melting point of 391 K) to avoid phase transition. The soldering process was conducted at 391 K, which does not reach the phase transition temperature of Ag2Se (~406 K).

Characterization and transport-property measurements

XRD and transport properties were measured on samples sliced along the directions parallel and perpendicular to that of pressure applied during hot pressing, as shown in Fig. 1b. The orientation factor F of (110) was calculated from the ratios of the integral intensities of the (110) planes to the intensities of the (hkl) planes for preferentially and for randomly orientated samples according to the Lotgering method44. The phase transition of Ag2Se was confirmed by a differential scanning calorimetry (DSC) measurement system (Netzsch DSC 3500 Sirius). The phase composition and microstructures of the materials were characterized by X-ray diffraction (XRD, DX-2700) and a scanning electron microscope (SEM, Phenom Pro, and Zeiss Sigma 300VP) equipped with an energy dispersive spectrometer (EDS).

Resistivity, Hall mobility, carrier concentration, and Seebeck coefficient were simultaneously measured at various temperatures under helium. The details of the measurements can be found in our previous work45. The thermal conductivity was calculated by κ = dCpλ, where d is the density measured by Archimedes drainage method, Cp is the specific heat, λ is the thermal diffusivity measured using laser flash technique (Netsch LFA 467). Sound velocities were measured at room temperature using a pulse receiver (Olympus-NDT) equipped with an oscilloscope (Keysight). Water and Shear gel (Olympus) was used as couplant during the measurements of longitudinal and transverse sound velocity (vL and vS), respectively.

Mechanical property

Vickers hardness test was carried out using Vickers hardness tester (DHV-1000) under the load of 2.94 N holding for 10 s. Mechanical tests, including three-point bending and compression tests, were performed on hot-pressed Ag2Se and commercially extruded Bi2Te3 (Xiamen X-Meritan Technology Co., Ltd.) using a micro-computer controlled electronic universal test machine [Lishi (Shanghai) Instruments Co., Ltd., P. R. China] with loading rates of 0.03 mm/min and 0.5 mm/min, respectively. During three-point bending tests, the span of the fixture was kept at 6 mm, while the sheet was 2.5 mm in width and 1 mm in thickness. Bulks with the size of 2 × 2 × 4 mm3 were used for compression tests, the same direction and dimensions as device legs (Fig. 1). The yielding strength (σ0.2) in bending and compression were the bending and compressive 0.2% offset stress from the bending and compressive stress-strain curve, respectively. After bending and compression tests, the surfaces and fracture surfaces were characterized by SEM with a secondary electron (SE) detector (Zeiss Sigma 300VP).

Conversion efficiency and cooling performance

The commercial Bi2Te3 module used in this work is from Xiamen X-Meritan Technology Co., Ltd. (7 pairs, 12 × 12 mm2), the detailed parameters of which are listed in Table. S1. The n-Ag2Se/p-Bi2Te3 modules in this work were fabricated by replacing the n-type Bi2Te3 legs of commercial Bi2Te3 modules with n-type Ag2Se legs.

The contact resistance (Rc) was measured by a four-probe technique at a constant electric current of ~200 mA. The interfacial contact resistivity (ρc) was estimated by ρc = Rc × A, where A is the cross-section area of the leg. The total resistance of the module was obtained from the slope of the AC voltage vs. current between the two copper wires of the modules within 0–0.2 A.

A homemade power-generation measurement system was used to measure conversion efficiency (Fig. S1a). The cold-side temperature (Tc) was maintained by a water-cooling system. K-type thermocouples were used to measure the hot side (Th) and cold side (Tc) temperatures of modules and the temperature difference of the heat flow meter (∆TCu). The output power P, heat flow Q, and conversion efficiency η of the modules under different temperature gradients were measured in a vacuum. P is determined by P = IV, where I is the current, and V is the output voltage. Q is obtained by Q = (κCuACu∆TCu)/LCu, where κCu, ACu, ∆TCu, and LCu are thermal conductivity, cross-section area, temperature difference, and distance between thermocouples of the heat flow meter. The average κCu used for determining Q is ~386 W m-1 K−146. ACu and LCu are 12 × 12 mm2 and 25 mm, respectively. Therefore, η is determined by η = P/(P + Q). The maximum η at a given Th can be obtained by varying the load resistance in the circuit. We measured each parameter (including temperature, voltage and current) for 30 times to minimize the system error.

For cooling performance measurement (Fig. S1b), the thermoelectric module was attached to Cooper Block with a built-in cooling water circulation system using thermal silicone grease. Two K-type thermocouples were used to record temperatures (Th for hot-side, and Tc for cold-side temperature). A ceramic heater with a similar size (12 × 12 mm2) was attached to the top of the module using the same silicone grease to measure the cooling power QC at various temperature reductions ∆T = Th − Tc. The minimum temperature of the cold side (Tcmin) can be obtained by varying the input current I in the circuit, and the maximum cooling temperature difference (∆Tmax = Th- Tc) was estimated at equilibrium when the heater did not work. All measurements were carried out under the vacuum of <1 Pa with Th of ~297-300 K. The coefficient of performance (COP) is determined by COP = QC/Pin, where QC is cooling power, and Pin is total input power. QCmax is the maximum cooling power under a given current Iq when ∆Tmax = 0 K.