Thermoelectric generators (TEGs) produce electricity from heat energy1–3, with conventional TEGs connecting n- and p-type materials in series via metal electrodes4–7. Placing one end of a TEGs on a heat source induces a temperature gradient in the materials, with heat flowing in the direction along the temperature gradient. Additionally, electrons (n-type materials) and holes (p-type materials) move simultaneously in the direction of heat flow. Thus, current passes through the TEGs, generating electric power.
Recently, TEGs have been proposed as power supplies for the wireless sensors and wearable devices that constitute the Internet of things (IoT)8–11. These applications require TEGs that are small, lightweight, and mechanically flexible, without excessively high power caused by a large temperature gradient. Thin-film TEGs on a flexible substrate are the most promising candidate to satisfy these requirements12–15. To date, bismuth telluride-based alloys have proved a popular choice for thin-film TEGs owing to their impressive thermoelectric properties near room temperature16–19. Much research has been devoted to increasing the thermoelectric performance and optimizing the structure of TEGs20–23. However, current TEGs suffer from two major weaknesses. First, a heat source is required to create the temperature gradient in the TEGs. Second, heat flows vertically to the heat source and the direction cannot be controlled; therefore, p- and n-type materials are needed to produce electric power efficiently. Overcoming these weaknesses will allow the use of TEGs to expand unabated.
One promising TEG structure for surmounting these obstacles is water-floating carbon nanotube (CNT) TEGs (see Fig. 1). Carbon nanotubes are lightweight and possess both flexibility and mechanical strength24–26. In addition, single-wall carbon nanotubes (SWCNTs) with specific chirality exhibit relatively high thermoelectric properties near room temperature27–30. When a bundle of SWCNT films floats on water, the water passes through the gaps in the SWCNT bundle and reaches the surface via capillary action. When the water evaporates, heat is absorbed, and the surface temperature drops via evaporation-induced cooling. Moreover, this cooling effect can be enhanced by exposing the SWCNT films to sunlight and wind flow. Designing the SWCNT films to have permeable and nonpermeable areas establishes a temperature gradient between these areas, enabling thermoelectric power to be generated via the Seebeck effect. This phenomenon indicates that TEGs can be fabricated using either p- or n-type materials alone. Typically, SWCNTs exhibit stable p-type properties31. In contrast, synthesizing air-stable n-type SWCNTs is extremely challenging32–35. Therefore, water-floating TEGs based on p-type SWCNT films are promising candidates to supply power to wireless environmental sensors used to measure the quality, quantity, and temperature of water. In addition, the TEGs have potential as wearable sensors where body sweat represents the water source.
Here, we report the fabrication and successful testing of water-floating TEGs using only p-type SWCNT films. The flexible substrate comprises square holes arranged in a staggered pattern, with each hole covered by an edge of a rectangular SWCNT film. The ends of adjacent SWCNT films are connected by metal electrodes. The temperature distribution in the TEGs was monitored as they floated on water. Thus, we verified that the heat flow can be controlled by evaporation-induced cooling and that the temperature gradient is stable. In addition, the thermoelectric power generated by the TEGs was measured in response to variations in water temperature, sunlight exposure, and wind exposure.
Fabrication and testing of water-floating SWCNT film TEGs
The process used to fabricate the water-floating SWCNT film TEGs is illustrated in Fig. 2a. Briefly, carbon nanotubes synthesized by the super-growth method (SG-CNTs) (ZEONANO SG101, ZEON Co.) were used as the starting material36, 37. Powdered SG-CNTs were dispersed in ethanol to prepare an SG-CNT dispersion solution with a concentration of 0.2 wt%. Next, an ultrasonic homogenizer (SONICS 85, AZONE Co.) was used to disperse the SG-CNT powder completely. The output power of the homogenizer was 60 W, and the dispersion time was 20 min. Because the vibrational energy during dispersion generates heat, the dispersion was conducted in a cold-water bath. The SWCNT films were prepared by a vacuum filtering method. A membrane filter (PTFE, 90 mm diameter: ADVANTEC) was placed in a filter holder in a suction bottle, and the dispersion solution was filtered by reducing the pressure in the suction bottle using a rotary pump to extract the material in the solution. A CNT-dispersed solution (40 mL) was released drop-by-drop onto the filter, and aspirated for 1 h to produce SWCNT films with a diameter of approximately 80 mm. After drying for 24 h in air, the SWCNT films were removed from the membrane filter. To assemble the TEGs, the SWCNT films were cut into four pieces, each measuring 45 mm in length and 15 mm in width. The substrate (80 mm × 60 mm, 125 µm thick) was a polyimide sheet (Kapton, DuPont) in which four rectangular holes (20 mm × 10 mm) were drilled in a staggered arrangement. The four sections of the SWCNT films were bonded to the substrate with silver paste such that adjacent films each covered half a hole in the polyimide. The SWCNT films were connected in series using thin copper wires. The cross-plane SEM image shows that the film was deposited almost uniformly, with a film thickness of approximately 50 µm (Fig. 2b). The surface SEM image shows the stacking of numerous winding SWCNT bundles with various diameters, including the presence of gaps between the bundles (Fig. 2c). These gaps allow water to reach the surface via capillary action. The TEM image of the SWCNTs shows that their diameters measure several tens of nanometers, with the SWCNTs arranged in a one-dimensional linear structure (Fig. 2d).
We conducted our experiments assuming typical natural environment conditions. The SWCNT film TEGs were floated on a 450 mL volume of water at initial temperatures of approximately 20°C and 80°C. Wind was applied to the TEGs using a compact circulator (PCF-HD15-W, IRIS OHYAMA Inc.) while the wind speed (3.0 m/s) was measured by an anemometer (SP-82AT, Mother Tool Co.). The TEGs were irradiated using an artificial solar illuminator (XC-100, SERIC Ltd.) to simulate direct sunlight (approximate light intensity: 1000 W/m2), with the intensity measured by a solar power meter (DT-1307, CEM Instruments). The temperature distribution in the TEGs was measured by a thermographic camera (Type F30W, Japan Avionics). The thermoelectric power was measured using a heat flow logger (LR8432, Hioki Co.).
Thermoelectric properties of SWCNTs and their films
The thermoelectric properties of the SWCNT films are presented in Table 1. The in-plane electrical conductivity, Seebeck coefficient, and power factor near 20°C are 88 S/cm, 55 µV/K, and 26.7 µW/(m·K2), respectively. Additionally, the in- and cross-plane thermal diffusivities, Din and Dcross, are 18.1 and 0.3 mm2/s, respectively. The thermal conductivity can be determined from the thermal diffusivity (D), density (ρ), and specific heat (Cp) based on the equation κ = DρCp38. The density of the SWCNT film was measured as 0.31 g/cm3, while a specific heat of 0.96 J/(g·K) was referred from the literature39. As a result, the in- and cross-plane thermal conductivities were determined to be 5.4 and 9.0 × 10-2 W/(m·K), respectively, which are lower than those of SWCNT films reported previously owing to differences in the electrical conductivity due to different synthesis methods40. However, the lower thermal conductivities of the SWCNT films in this study facilitate effective temperature gradient generation in the films.
Performance of water-floating SWCNT film TEGs
The temperature distribution and performance of the water-floating SWCNT film TEGs in response to various environmental conditions are shown in Fig. 3. The ambient temperature in all experiments was approximately 20°C. A photograph of the experimental setup is provided in Fig. S1 (see Supplementary Material). For a water temperature of approximately 20°C, and in the absence of simulated sunlight and wind, the temperature gradients between the positions with and without substrate holes were invisible in the thermographic image (Fig. 3a). However, a stable thermoelectric power of approximately 120 µV was detected with four SWCNT films (Fig. 3b), indicating that an approximate temperature difference of 0.5 K occurred in each film based on the Seebeck coefficient of the SWCNT films (55 µV/K). As shown in Fig. S2 (see Supplementary Material), we verified that an almost constant value of thermoelectric power was maintained when the TEGs floated on the water for 60 min, thereby demonstrating stable thermoelectric power generation without a heat source. The thermographic image in Fig. 3c shows that for an initial water temperature of 80°C (without simulated sunlight and wind), a clear temperature gradient is established between the positions with and without substrate holes. Furthermore, the ‘hot’ and ‘cold’ areas are located opposite each other on the adjacent films. This implies that the heat flow can be controlled by changing the position of holes in the substrate, and that only one type of (n-type or p-type) film is required to create the TEGs. The approximate temperature difference under these conditions was estimated to be 5 K, and the temperature gradient at edge of the holes is steep because the in-plane thermal conductivity of the SWCNT films (κin = 5.4 W/(m∙K)) is not substantially higher than conventional thermal conductive materials (κ ≈ 400 W/(m∙K) in copper). In addition, owing to the low thermal conductivity of the polyimide substate (κ = 0.16 W/(m∙K)), heat is not transferred between adjacent films. These results demonstrate that the size of the films and the interval between them can be reduced to increase the power density of the TEGs as provided in Fig. S3 (see Supplementary Material). The thermoelectric power at a water temperature of 80°C was 950 µV. The thermoelectric power was observed to decrease with temperature, measuring 400 µV at a water temperature of 50°C (Fig. 3d). A temperature gradient was also observed when exposing the TEGs to simulated sunlight (1000 W/m2) in a wind-free environment at a water temperature of 14°C (Fig. 3e). A stable thermoelectric power of approximately 450 µV was detected, indicating a temperature difference of 2 K in each film (Fig. 3f), representing a 3.7-fold increase in the thermoelectric power compared to the corresponding measurement without simulated sunlight and wind. This is attributed to the temperature of the SWCNT films at hole-free locations increasing owing to the film exhibiting extremely high light absorption, while the temperature of the films at the holes does not increase because of the cooling effect induced by water evaporation. As shown in Fig. S4 (see Supplementary Material), we verified that the humidity above the films at the holes was higher than that above the films at hole-free locations. The thermographic image in Fig. 3g, which corresponds to wind flow (3.0 m/s) without simulated sunlight at a water temperature of 18°C, shows a slight temperature gradient. In addition, a temperature difference between the SWCNT films and the water was also clearly observed. This is because the heat transfer from the water to the SWCNT films was limited by the relatively low cross-plane thermal conductivity of the SWCNT films (κ = 9.0 × 10-2 W/(m·K)). Based on the stable thermoelectric power of approximately 300 µV detected under these conditions, we can infer a temperature difference of 1.4 K in each film (Fig. 3h). Exposing the TEGs to a wind flow of 3 m/s increased the thermoelectric power by a factor of 2.5 relative to the wind and simulated sunlight-free measurement. It is considered that wind causes the concentration of water vapor near the film surface to decrease. As a result, water evaporation is promoted, which lowers the surface temperature. This suggests that the TEGs can generate the thermoelectric power at night. Our experiments demonstrate that water temperature, sunlight, and wind all affect the power generation of the TEGs. Therefore, we determined the combination of water temperature, sunlight exposure, and wind exposure required to optimize the power generation of the TEGs (Fig. 3g). Consequently, we realized a thermoelectric power of 1300 µV at a water temperature of 80°C, simulated sunlight of 1000 W/m2, and wind speed of 3.0 m/s. Decreasing the water temperature while maintaining the other conditions resulted in the thermoelectric power decreasing, with a power of 800 µV recorded at a water temperature of 30°C (Fig. 3h). Finally, we verified that the TEGs can generate the thermoelectric power in actual environmental conditions (natural sunlight and wind) as shown in Fig. S5 (see Supplementary Material).