A high-performance wearable thermoelectric device with epoxy resin/PA/AlN composite heat sink

Abstract

Wearable thermoelectric devices (WTEDs) have garnered significant interest for applications in personal thermal management and energy harvesting from the human body. Nevertheless, a major limitation hindering their performance is inadequate heat dissipation. To overcome this issue, this study introduces an integrated WTED architecture featuring a flexible finned heat sink based on a phase change material (PCM), which is structurally integrated with the thermoelectric unit. The heat sink is composed of epoxy-encapsulated paraffin PCM doped with high-thermal-conductivity aluminum nitride (AlN) additives, significantly enhancing the composite's effective thermal conductivity and overall thermal management capability. Experimental results validate the efficacy of this integrated design: at a temperature difference of 30 K, the device delivers an open-circuit voltage of 239.0 mV and a power output of 9212.85 µW, corresponding to a 1386% enhancement compared to a system without the heat sink. Moreover, the integrated device demonstrates markedly improved conversion efficiency even under low temperature gradients (<10 K), rendering it highly suitable for integration with boosting circuits in wearable electronics. Based on heat transfer optimization, this device achieves portable wearable cooling without fans or liquid cooling assistance. It provides up to a 4.4 °C skin temperature drop and maintains a temperature below body temperature for over 600 s. This work presents an effective, fan- and liquid-free thermal management solution with promising applications in small portable electronics and personalized cooling.

---

1 Introduction

Wearable thermoelectric devices (WTEDs) have garnered significant attention for their potential to bridge energy harvesting and thermal regulation, with broad applications in personal thermal management (PTM) and human energy harvesting.^1–7 Yet, their practical performance is hindered by a critical limitation: inefficient and inflexible heat dissipation. Current heat management systems in WTEDs fail to balance thermal conductivity, flexibility, and heat transfer efficiency, directly restricting both power generation capacity and cooling efficacy—key barriers to their broader adoption.^8–14 Enhancing WTED performance through targeted heat sink optimization thus remains a pivotal research focus.

Existing efforts to address WTED heat dissipation have shown limited success in wearable scenarios.^15,16 Park et al. improved thermal uniformity with topology-optimized air-cooled heat sinks,¹⁷ and Wang et al. boosted efficiency via water-cooled thermoelectric cooler (TEC) integration.^18,19 However, these designs suffer from bulkiness and rigidity, making them incompatible with the conformability demands of wearable devices.

To improve the performance of thermoelectric devices and provide heat dissipation solutions suitable for wearable devices, many scholars have added flexible heat dissipation layers^20–28 to suppress thermal conduction from the hot end to the cold end, thereby achieving good cooling effects. Among the materials explored for flexible heat dissipation, phase change materials (PCMs) are notable due to their ability to absorb or release a large amount of heat during the phase change process, and they are widely used in the field of heat recovery.^29–31 Paraffin (PA)-based PCMs are particularly attractive for wearable applications owing to their high latent heat capacity (typically 180–220 J g⁻¹) and their phase transition temperature being close to that of the human body, making them suitable for thermal regulation. Cheng et al. encapsulated a PCM (paraffin) with epoxy resin to prevent the leakage of the PCM.³² However, although PCMs have significant phase change latent heat, they exhibit poor thermal conductivity,^33–35 which easily leads to heat transfer lag and other issues. Therefore, achieving collaborative optimization of thermal conductivity and latent heat has become a key challenge for improving the heat dissipation layer of wearable thermoelectric devices.

To address these challenges, in this paper, epoxy resin is employed to encapsulate a paraffin-based PCM, while AlN is incorporated to improve the effective thermal conductivity of the entire heat dissipation layer. This is because AlN offers high intrinsic thermal conductivity (180–200 W m⁻¹ K⁻¹) while maintaining electrical insulation. Additionally, to further enhance the heat dissipation performance, the heat dissipation layer is fabricated into a fin-shaped structure to enlarge the surface area and thereby enhance heat dissipation performance. Simulations and experiments demonstrated that this flexible heat dissipation layer can effectively improve the cooling performance and power generation performance of wearable thermoelectric devices. The WTED can achieve an open-circuit voltage of 239.0 mV and a power of 9212.85 µW at a temperature difference (ΔT) of 30 K and can reduce human skin temperature by 4.4 °C. The device can promptly and evenly cool the human body post-exercise, thereby reducing the risk of thermal injury. These advancements not only address the longstanding challenges in wearable thermoelectric heat dissipation but also provide an innovative technical approach with clinical application value for cooling and rehabilitation in the field of sports medicine.

2 Experimental section

All experiments involving human participants were performed in accordance with relevant guidelines and regulations, and informed consent was obtained from all participants.

2.1 Materials

The p-type Bi_0.5Sb_1.5Te₃ and n-type Bi₂Te_2.7Se_0.3 thermoelectric legs, each sized at 1.4 × 1.4 × 2.5 mm³, were obtained from Hubei Segre New Energy Technology Company Limited. Flexible printed circuit boards (FPCBs) were obtained from Jialichuang Company, while the thermally conductive material AlN was obtained from Suzhou Yuante New Materials Company. Epoxy resin (Leini 105) was obtained from Huizhou Tinglan Technology Company, and the phase-change material (PCM-A-30, PA) was obtained from Dongguan Zhangmutou Shengbang Plastic Raw Material Operation Department, which served as the PCM.

2.2 Fabrication of the WTED

The fabrication process of the WTED is shown in the SI (Fig. S1). The preparation process of the WTED consists of three main parts, namely, the preparation of the thermoelectric part, the preparation of the liquid PCM composite heat sink, and the final curing step, corresponding to Fig. S1(a–c), respectively.

2.2.1 Thermoelectric part. The procedure for assembling the thermoelectric part is illustrated in the SI (Fig. S1(a)). The overall FPCB size was 24 mm × 44 mm and the FPCB was placed on an unheated heater plate, and low-temperature solder paste [Sn₄₂Bi₅₈ (melting point is 138 °C)] was loaded onto the mask and swiftly printed onto the substrate electrode using a squeegee. Then, the p-type and n-type thermoelectric legs were deposited onto the bottom electrode of the FPCB. Next, the whole sample was heated up to 200 °C for solder bonding, and the soldering time was 5 min. The top electrodes were also connected to the thermoelectric legs using the same solder for electrical connectivity.

2.2.2 Liquid PCM composite heat sink part. Epoxy resin A and B agents were added sequentially, mixed, and stirred at 50 °C for 10 min, as demonstrated in the SI (Fig. S1(b)). PA and AIN were then added, mixed, and stirred at ratios of 15 wt% and 5 wt% for another 10 min (epoxy resin : PA : AIN = 8 : 1.5 : 0.5). After mixing, the mixture was vacuumed for 5 min to remove any bubbles before being poured into a mold.

2.2.3 Curing part. The preparation process for this part is shown in Fig. S1(c). First, the liquid PCM composite heat sink prepared in Section 2.2.2 was poured into the mold, and vacuuming was performed to ensure that the liquid PCM composite heat sink filled the mold.¹⁹ Second, the thermoelectric part prepared in Section 2.2.1 was placed on a support tool, and the liquid PCM composite heat sink was ensured to saturate the top electrode of the thermoelectric part; subsequently, the assembly was placed in an oven at 50 °C for 8 hours to allow the liquid PCM composite heat sink to cure. It is worth noting that, as the liquid PCM composite heat sink cured, the thermoelectric portion and the PCM composite heat sink had become one and inseparable at this point. Finally, the whole device was molded, and the FPCB at the bottom of the device was cut to release the stress. The resulting device is a flexible WTED. The as-fabricated finned heat dissipation layer consists of 153 fins, each measuring 9 mm in height and 1.5 mm in diameter, with an inter-fin spacing of 1 mm. This structural configuration provides an approximate 443% increase in heat dissipation surface area compared to a flat heat sink of the same footprint, significantly enhancing the overall thermal management performance.

2.3 Testing instruments and methods

Pure epoxy resin and the PCM composite heat sink underwent differential scanning calorimetry (DSC) experiments (NETZSCH, Germany) to assess their heat absorption and release capabilities. The DSC experiments were set to start at 18 °C and end at 55 °C, with the ramp-up and ramp-down rates both set at 3 °C min⁻¹. All samples were tested in a nitrogen (N₂) atmosphere. The optical photos of pure epoxy resin and the PCM composite heat sink were obtained using an OLYMPUS camera. The Fourier transform infrared (FT-IR) spectra of pure PA, pure epoxy resin and the PCM composite heat sink were recorded using a Thermo Fisher Scientific Nicolet iS20 produced by Nicolet in the United States, and the scanning range was 400–4000 cm⁻¹. The bending and tensile properties were evaluated using a flexible electronic tester. The skin temperature before and after cooling was recorded using a thermal infrared imager (FLUKE Ti480 PRO). Additionally, a thermocouple was placed at the interface between the skin and the cold end of the WTED to measure temperature. The internal resistance of the device was measured by using an electrochemical analyzer (CHI650e, CH Instruments, Inc). The reported data are averages of at least three measurements.

2.2.4 Fin geometry optimization. The geometry of the heat dissipation fins was optimized by considering thermal performance, manufacturability constraints, and deformation risks. The selected fin diameter of 1.5 mm and height of 9 mm achieve a balance among multiple critical manufacturing limits, thermoelectric performance requirements, and deformation resistance. The detailed theoretical analysis, simulation process, and experimental validation are provided in Fig. S2.³⁶

3 Results and discussion

3.1 WTED and materials characterization

Fig. 1a illustrates the schematic diagram of human-body energy harvesting, cooling applications, and the composition of the proposed WTED. The performance of our WTED is compared with other state-of-the-art devices in multiple aspects, as summarized in Fig. 1b, demonstrating its competitive advantages.

Fig. 1 Schematic of human application of the WTED and material characterization of the heat sink. (a) Schematic diagram of WTED's power generation, cooling functions and the structure. (b) WTED performance radar chart comparison. (c) Change in thermal conductivity of epoxy resin, epoxy resin/PA composite and epoxy resin/PA/AIN composite. (d) Comparison of DSC images of pure epoxy resin, epoxy resin/PA composite and, epoxy resin/PA/AIN composite. (e) WTED power generation demonstration.

As shown in Fig. 1c, the thermal conductivity of the epoxy resin/PA composite exhibited no significant difference compared to that of pure epoxy resin, which was 0.201 W m⁻¹ K⁻¹ and 0.203 W m⁻¹ K⁻¹, respectively. However, incorporation of AIN particles resulted in an increase in thermal conductivity from 0.203 W m⁻¹ K⁻¹ to 0.309 W m⁻¹ K⁻¹, representing a 52.2% enhancement over the composite without AIN particles. It is noteworthy that a significant discrepancy exists between the intrinsic thermal conductivity of AlN (180–200 W m⁻¹ K⁻¹) and the effective thermal conductivity of the composite (0.309 W m⁻¹ K⁻¹). This gap is fully explainable by the principles of composite physics and aligns with predictions from validated micromechanics models, such as the multi-step Mori-Tanaka (MMT) model.³⁷ Furthermore, Fig. 1d compares the DSC curves of pure epoxy resin, the epoxy resin/PA composite, and the epoxy resin/PA/AIN composite. The DSC analysis reveals that the addition of PA enhances the latent heat capacity of epoxy resin near the phase transition temperature, whereas the inclusion of AIN does not significantly alter these thermal properties. When the WTED is attached to a beaker and hot water at 35 °C is added into the beaker, the LED lights up immediately, as shown in Fig. 1e.

3.2 Performance optimization method and testing of the WTED

This paper proposes a performance optimization scheme for TEDs. As shown in Fig. 2a, this scheme significantly improves power generation performance and cooling effects by reducing interfacial thermal resistance, increasing heat dissipation area and enhancing thermal conductivity. The underlying principles for optimizing both power generation and cooling performance are now analyzed in turn.

Fig. 2 Performance optimization method for thermoelectric devices. (a) Structural schematics of 5 TEDs. (b) Comparison of simulation output voltage performance of 5 TEDs. (c) Comparison of simulated cooling effects of 5 TEDs.

For power generation, based on previous research,³⁸ the ΔT across the TE leg is proportional to the ΔT across the whole device and also proportional to the TE output voltage:

A steady-state heat balance equation for the cold end is established as follows, with the hot end temperature held constant:

which can be transformed into:where T_h and T_c are the temperatures at the hot and cold ends of the TE leg, respectively; R, R_HS and R_PI are the thermal resistances of the TE leg, heat sink, and PI layer, respectively; and are the temperatures at the hot and cold ends of the TED, respectively. K₃ represents the effective thermal conductance of the TED, is the heat transfer coefficient at the cold end of the TED, S is the contact surface area for heat exchange with air, and T_air is the air temperature.

To improve the TED performance, a heat sink was incorporated into TED-1 to improve convective heat dissipation at the cold end. The effectiveness of this PCM-enhanced design is clearly demonstrated by the temperature results: TED-1 achieved a cold surface temperature of 17.90 °C, whereas TED-0 (without the PCM) reached 21.42 °C. This 3.52 °C improvement in cooling performance can be attributed to the increased latent heat capacity provided by the PCM, which enables more efficient absorption and storage of thermal energy during operation. As indicated by eqn (3), an increase in results in a decrease in , which in turn raises , thereby increasing the output voltage. In wearable thermoelectric generators (TEGs), minimizing interfacial thermal resistance is crucial for performance enhancement, a principle well-established in the field. Pioneering work, such as that by Malakooti et al.,³⁹ employed liquid metal supercooling technology to depress the freezing point of the metal, thereby maintaining high flexibility at extremely low temperatures. The conformal contact capability of this material reduces the contact thermal resistance between the TEG and the heat sink, ultimately improving the TEG's output performance in low-temperature environments. This study demonstrates that enhancing interfacial heat transfer is vital for boosting device performance. Inspired by this principle, this work addressed the significant interfacial thermal resistance introduced by the separate design of the thermoelectric module and the heat sink in TED-1, which reduces heat transfer efficiency. Therefore, adopting a similar optimization philosophy, in TED-2 we removed the polyimide (PI) layer between the thermoelectric unit and the heat sink, integrating them into a monolithic structure. This integration largely reduced the interfacial thermal resistance. According to eqn (1), R_PI = 0 results in an increase in ΔT_leg, consequently enhancing the output voltage. Furthermore, in TED-3, the heat sink was fabricated into a finned structure, expanding the heat dissipation area. According to eqn (3), an increase in S leads to a decrease in , thus increasing the output voltage. Finally, by incorporating high thermal conductivity AlN into the heat sink, the output voltage was further increased. According to formula eqn (3), increases further, enhancing the TE output performance of the WTED (output voltage of 1.99 mV when ΔT = 30 K), representing a 69% improvement compared to TED-0. Furthermore, in TED-3, the heat sink was designed with a finned structure to increase the available heat dissipation area. As derived from eqn (3), a larger surface area S contributes to a lower cold-side temperature , thereby raising the output voltage. Additionally, the use of high-thermal-conductivity AlN in the heat sink further improved performance. Eqn (3) indicates that this modification enhances , which in turn boosts the thermoelectric output. As a result, the WTED achieved the best output voltage of 1.99 mV when ΔT = 30 K, marking a 69% enhancement compared to TED-0 (Fig. 2b). Meanwhile, the thermoelectric conversion efficiency of the WTED reached 0.17%, representing a 143% improvement over that of TED-0 (0.07%), as detailed in Fig. S3.

Subsequently, the influence of device optimization on the cooling performance was investigated by establishing and rearranging the heat balance equation at the TED's hot end (position 2 in Fig. S4). The simplified equation is as follows:

where K₂ represents the effective thermal conductance of the PCM and PI, and h₂ denotes the heat transfer coefficient between the hot end and the ambient air. Simulation results indicate that TED-0, which lacks a heat sink, achieved only a temperature drop of 3.58 °C. In contrast, TED-1 incorporated a heat sink, resulting in an increase in h₂ and a consequent reduction in . Additionally, the heat balance equation at the hot end of the thermoelectric leg (position 1 in Fig. S4) was derived and reformulated as follows:where K₁ denotes the effective thermal conductance of the TE leg, α is the Seebeck coefficient, I is the driving current, and R denotes the electrical resistance of the device. From eqn (5), it follows that a decrease in leads to a corresponding reduction in T_h. The heat balance equation at the cold end (position 3 in Fig. S4) was subsequently established and rearranged into the following form:

According to eqn (6), a decrease in T_h leads to a decrease in T_c. Assuming that the cold end heat transfer conditions remain unchanged, this leads to a decrease in . Therefore, the cooling performance of TED-1 is improved compared to TED-0.

TED-2 removed the PI layer compared to TED-1, resulting in an increase in K₂. Transforming eqn (4) to:

According to eqn (7), an increase in K₂ leads to an increase in . Combining eqn (4) and (5) yields:

It can be observed that an increase in results in a decrease in T_h. Following the same logical sequence applied in the enhancement from TED-0 to TED-1, it can be similarly inferred that decreases.

TED-3, building upon the design of TED-2, introduced a finned heat sink structure to increase the heat dissipation surface area. As indicated by eqn (4), an increase in S causes to decrease, which, as previously described, subsequently leads to a reduction in . The final version, WTED, incorporated high-thermal-conductivity AlN into the TED-3 configuration. As in the initial optimization step, this further increase K₂, resulting in a decreased . The simulation results demonstrated a final temperature drop of 10.55 °C, which is 6.97 °C greater than that of TED-0, representing a 195% improvement in cooling performance (Fig. 2c). The complete derivation process is provided in the SI (Fig. S4). Fabrication procedures are detailed in Fig. S1, and boundary conditions are also included in the SI. Finite element simulations theoretically confirm the feasibility and effectiveness of the proposed optimization strategies.

To further validate the optimization design theory, TED-0, TED-1, TED-2, TED-3, and WTED were fabricated, and their performance was experimentally characterized. Fig. 3 summarizes the experimental results of the performance optimization process. As shown in Fig. 3a, the cold-side temperatures of TED-0 through the WTED are plotted over time under a constant current (0.3 A), demonstrating a consistent improvement in cooling performance with each iterative design.

Fig. 3 Experimental validation of the optimization method and performance comparison of the WTED with those reported in the literature. (a) Cold surface temperature profiles of TED-0, TED-1, TED-2, TED-3 and WTED over time at the same current. Comparison of (b) output voltage and (c) power density of TED-0, TED-1, TED-2, TED-3 and WTED at a ΔT of 15 K. (d) Power generation performance of the WTED at different ΔT values. (e) Comparison of the output voltage of our WTED with other thermoelectric devices reported in the literature. (f) Comparison of the output power of our WTED with other thermoelectric devices reported in the literature.

Fig. 3b and c show the open-circuit output voltage and power density, respectively, of the five devices at a ΔT of 15 K. Both the voltage and power density increase progressively from TED-0 to the WTED, confirming the effectiveness of the optimization strategy.

The output performance of the WTED across different temperature differences is shown in Fig. 3d. At ΔT = 30 K, the WTED achieves an open-circuit voltage of 239.0 mV and a power output of 9212.85 µW, representing a 1386% improvement compared to TED-0 (which yielded 31 mV and 619.98 µW under the same conditions).

Furthermore, it is essential to situate our work within the context of recent pioneering advances in wearable TEDs,^{20,23,24,26–28,39,40} which are closely aligned with the scope of the present study (Fig. 3e and f). For example, Han et al.⁵ developed a highly efficient stretchable thermoelectric generator capable of retaining stable electrical performance over 15 000 stretching cycles at 30% strain, achieving leading peak output voltage and power among comparable devices. In subsequent work,⁶ the same group introduced a 3D flexible TED architecture with electrical self-healing capability, which remained fully functional after multiple punctures and 2000 stretching cycles at 50% strain, while also delivering notable thermoelectric output under low temperature gradients. These studies demonstrate outstanding mechanical compliance and performance under active cooling conditions. In contrast, the present work focuses on a complementary aspect of wearable thermoelectric design: optimizing passive heat dissipation for daily-use scenarios. Our approach aims to enhance thermal management under natural convection, thereby improving thermoelectric output in practical, passively cooled wearable applications.

3.3 Reliability testing

A stability test was first conducted to evaluate the heat absorption capacity of the epoxy resin/PA/AlN composite heat sink. Specifically, the performance of the original unit was compared with that of the same heat sink after undergoing 150 cycles of thermal charging and discharging. The results are summarized in Fig. 4a. The DSC curves show minimal variation in the thermal energy storage and release capabilities of the phase change heat sink after 150 thermal cycles, indicating high stability. Additionally, the output voltage after thermal cycles is shown in Fig. S5. The leakage resistance of the phase change heat sink used in the WTED²⁹ was investigated. As shown in Fig. 4b, the WTED was placed on a hot plate at 100 °C for one hour, and no noticeable leakage was observed, confirming the reliability of its encapsulation.

Fig. 4 Stability test of the WTED. (a) Comparison of DSC images of epoxy resin/PA/AIN composites before and after 150 heat-absorbing and exothermic cycles. (b) WTED was placed on a heated bench at 100 °C to demonstrate its phase change heat sink leakage prevention properties. (c) Internal resistance stability at different bending radii when the WTED is bent inward. (d) Internal resistance stability of the WTED with 5000 repetitions of bending with a radius of 20 mm to the inside. (e) Internal resistance stability of the WTED when bending to the outside with different bending radii. (f) Internal resistance stability of the WTED bent to the outside for 5000 repetitions with a radius of 20 mm.

Experimental results demonstrate that the developed WTED exhibits excellent bidirectional flexibility and remarkable stability in internal resistance. As shown in Fig. 4c, when bent inward, the device maintained stable internal resistance even as the bending radius decreased from 45 mm to 20 mm. Furthermore, 5000 repeated bending cycles at a radius of 20 mm were conducted, and the results further confirm the consistency of the internal resistance (Fig. 4d).

Similarly, under outward bending conditions, the internal resistance remained highly stable with the bending radius reduced from 45 mm to 20 mm (Fig. 4e). An additional 5000 cyclic bending tests at 20 mm radius were performed, and the results, presented in Fig. 4f, (detailed experiments on the effect of bending cycles on thermoelectric output are presented in Fig. S6) again demonstrate exceptional resistance stability under outward bending.

3.4 WTED's power generation application

The human body continuously generates heat through metabolism, representing a utilizable source of energy. Collecting and analyzing surface heat from different body regions helps elucidate patterns of human thermal distribution, thereby informing the optimal placement and structural design of TEDs to improve energy harvesting efficiency. Such measurements also offer valuable insights into physiological status, supporting health monitoring and early disease detection. In this study, WTEDs were attached to the wrist, palm, and elbow to determine the temperature across the arm based on output voltage comparisons. As shown in Fig. 5a, after the participant rested quietly for 10 minutes, the wrist produced 84.8 mV, the palm 75.0 mV, and the elbow 65.8 mV at an ambient temperature of 19.7 °C and an arm temperature of 32.3 °C. These results indicate that the wrist exhibits a higher surface temperature compared to the palm and elbow, rendering it a more favorable site for wearable thermoelectric energy harvesting. This finding is critical for guiding device positioning and structural optimization, as selecting regions with higher thermal gradients can substantially enhance power output. Thus, the study supports the strategic deployment of WTEDs on areas with higher heat emission, providing key insights for future designs aimed at maximizing power generation and ensuring wearer comfort.

Fig. 5 WTED for human power generation applications and output stability. (a) The curves of open-circuit output voltage versus time for the WTED attached to the human wrist, palm and elbow, respectively. (b) The WTED input voltage and output boost voltage of the DC–DC converter while the TEG is tested for long-term stability on a heated plate set to 37 °C, with a room temperature of 19.8 °C. (c) The voltage provided by the WTED when attached to the wrist can power an LED lamp after passing through the boosting circuit board. (d) TED-0 placed in the palm showed no LED illumination after 25 s. (e) WTED lit up the LED instantaneously upon palm contact.

Fig. 5b presents the output voltage of the WTED mounted on a 37 °C heating stage along with the corresponding boosted output, confirming that the device can effectively and stably drive the boost circuit. This result underscores the suitability of the optimized WTED for powering wearable electronics with low power consumption, and as shown in Fig. 5c, the WTED can successfully light up an LED when worn on the wrist. To further clarify the operational boundaries of the device, this study also tested the limit conditions for LED lighting, as detailed in Fig. S7. To further demonstrate the enhanced performance in a practical setting, wearable tests were conducted (Fig. 5d and e). When TED-0 was attached to the skin, the LED failed to light up via the boost circuit even after 25 seconds. In contrast, the WTED illuminated the LED almost immediately upon attachment, offering direct visual evidence of its superior electrical output. These observations clearly indicate that TED-0 falls short of meeting the efficiency and stability requirements for wearable applications, whereas the WTED exhibits excellent performance in harvesting body heat and generating usable power, highlighting its strong potential for practical thermal energy harvesting on the human body.

3.5 WTED's cooling application

During exercise, the metabolic rate of muscles increases by three to five times compared to resting conditions, resulting in a substantial elevation in body temperature. Failure to dissipate this excess heat adequately after physical activity may lead to increased secretion of heat stress hormones, a reduced threshold for exercise-induced fatigue, and heightened health risks such as heatstroke. Implementing post-exercise cooling provides multiple physiological advantages: it promotes vasoconstriction, which helps reduce swelling and pain after intense activity; it facilitates a decrease in core body temperature, thereby alleviating muscle fatigue and preventing cramps, particularly following exercise in hot environments; and it accelerates recovery, allowing athletes and regular exercisers to return more rapidly to subsequent training sessions.

To address these requirements, the WTED was employed for efficient localized cooling. As depicted in Fig. 6a the device was designed for wrist wear. Thermal imaging results in Fig. 6b demonstrated its cooling performance: skin temperature decreased from 32.0 °C to 27.6 °C, a reduction of 4.4 °C, and remained below body temperature for over 600 seconds. These results confirm the cooling capability of the WTED (Fig. 6c), which is attributed to its optimized heat transfer structure.

Fig. 6 Human cooling application of the WTED. (a) WTED fitted to human skin for cooling. (b) Infrared photo of human skin before and after the action of the WTED. (c) Temperature change curve of human skin temperature under the action of the WTED. (d) Exercise schematic: 5 min dumbbell lifting repetition. (e) Comparison of post-exercise cooling curves under natural convection and WTED-assisted cooling.

To further evaluate the effectiveness of the device, an exercise experiment was conducted. The participant rested quietly at room temperature for 10 minutes before performing repeated dumbbell lifts for five minutes. The experimental setup is illustrated in Fig. 6d. Fig. 6e compares post-exercise arm temperature changes under two cooling conditions: natural convection and active cooling with the WTED. The pre-exercise skin temperature was 32.7 °C, which increased to 33.5 °C after exercise. Under natural convection, the skin temperature decreased only to 32.1 °C, whereas with the WTED, it dropped rapidly to 29.2 °C at the same time point. The temperature decline was gradual under natural convection, but the WTED enabled a significantly faster cooling rate. After five minutes, the skin temperature with WTED cooling remained approximately 1 °C lower than that under natural convection.

These findings confirm that the WTED effectively reduces skin temperature following exercise and promotes a quicker return to a comfortable thermal state. This technology offers a practical cooling solution for athletes and sports enthusiasts, with important implications for enhancing post-exercise recovery and thermoregulation.

4 Conclusions

This study presents a monolithic WTED with an epoxy resin/PA/AIN composite heat sink, which effectively addresses the critical heat dissipation challenge in traditional compact TEDs. The device employs a unified structural architecture based on an epoxy resin skeleton, eliminating the performance degradation caused by thermal interface resistance between the TED and the heat sink in conventional designs. Through optimized heat transfer strategies, including (1) enhanced latent heat from PA additives; (2) improved thermal conductivity via AIN fillers, and (3) an enlarged heat dissipation area via a fin-shaped design, the WTED achieves significantly improved heat dissipation, leading to superior TE performance. Experimental results demonstrate outstanding performance: at a ΔT of 30 K, the WTED delivers an open-circuit voltage of 239.0 mV and a power output of 9212.85 µW, representing a 1386% improvement compared to a traditional TED-0 without a heat sink. Furthermore, the WTED achieves a skin temperature reduction of 4.4 °C and maintains sub-body temperature for over 600 seconds. This work offers an effective fan-free and liquid-free heat dissipation solution with strong potential for applications in powering small portable devices as well as personalized thermal management.

Author contributions

Z. Z., Z. X., Y. H., C. L, and Z. W. conceived the idea of this work. Z. Z., Z. X., W. C., X. S., C. L., Q. D., W. Z., Z. J., and L. W. conducted the characterization and measurements of this work. Z. Z., Z. X., H. T., Y. H., C. L, and Z. W. wrote and revised the manuscript. All of the authors reviewed the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: detailed formula derivations, as well as the complete procedures and corresponding data for TED design. See DOI: https://doi.org/10.1039/d5ta07253k.

Acknowledgements

The authors sincerely thank Prof. Zhengyou Liu for providing the authorization to use COMSOL software, which has been essential for our research simulations. This work was supported by the National Natural Science Foundation of China (grant no. 12302220 and 12474093), Fundamental Research Funds for the Central Universities (grant no. 2042025kf0037 and 2042025kf0013), and Institute for Goneo New Energy Foundation of Wuhan University.

Notes and references

1. L. Li, W. D. Liu, Q. Liu and Z. G. Chen, Adv. Funct. Mater., 2022, 32, 2200548.
2. J. D. Periard, T. M. H. Eijsvogels and H. A. M. Daanen, Physiol. Rev., 2021, 101, 1873–1979.
3. S. Wiecha, I. Cieslinski, P. Wisniowski, M. Cieslinski, W. Pawliczek, P. Posadzki, R. Prill, J. Zajac and M. Plaszewski, Sports Med., 2025, 55, 1183–1212.
4. M. Al-Rjoub, M. J. Kazmierczak, A. Bhattacharya, S. Rakkimuthu, S. Ramadurai, J. P. Stuckey and R. K. Banerjee, Int. J. Heat Mass Tran., 2021, 173, 121204.
5. Y. Han, L. E. Simonsen and M. H. Malakooti, Adv. Energy Mater., 2022, 12, 2201413.
6. Y. Han, H. Tetik and M. H. Malakooti, Adv. Mater., 2024, 36, 2407073.
7. V. O. Ottone, F. De Paula, P. F. A. Brozinga, M. A. de Matos, T. C. Duarte, K. B. Costa, B. C. C. Garcia, T. J. Silva, F. C. Magalhaes, C. C. Coimbra, E. A. Esteves, K. M. C. Pinto, F. T. Amorim and E. Rocha-Vieira, Front. Physiol., 2022, 13, 867362.
8. X. Zuo, X. Zhang, L. Qu and J. Miao, Adv. Mater. Technol., 2022, 8, 2201137.
9. C. Wang, H. Dong, C. Cheng, K. Sun, T. Jin and Q. Shi, ACS Sustainable Chem. Eng., 2022, 10, 16368–16376.
10. Y. Guo, C. Dun, J. Xu, J. Mu, P. Li, L. Gu, C. Hou, C. A. Hewitt, Q. Zhang, Y. Li, D. L. Carroll and H. Wang, Small, 2017, 13, 1702645.
11. A. Hazarika, B. K. Deka, D. Kim, H. E. Jeong, Y. B. Park and H. W. Park, Nano Lett., 2018, 18, 6731–6739.
12. J. Huang, Y. Li, Z. Xu, W. Li, B. Xu, H. Meng, X. Liu and W. Guo, Nanotechnology, 2019, 30, 325203.
13. P. Zhu, X. Luo, X. Lin, Z. Qiu, R. Chen, X. Wang, Y. Wang, Y. Deng and Y. Mao, Energy Convers. Manag., 2023, 285, 117017.
14. J. Li, S. Ren, T. Zhang, L. Zhao, J. Wang, Y. Du and J. Fang, ACS Appl. Mater. Interfaces, 2024, 16, 64042–64049.
15. M. He, B. Zhao, X. Yue, Y. Chen, F. Qiu and T. Zhang, Nano Energy, 2023, 116, 108821.
16. X. Wang, X. Du, D. Xu, T. Wei, Q. Zhang, Z. Chen, D. Kong, K. Wei, L. Zhou, B. Zhu, W. Xu and J. Zhu, ACS Nano, 2025, 19, 19977–19988.
17. J. Park, H. Yu, K. M. Bang, W. Kim and H. Jin, Energy, 2025, 320, 135127.
18. W. Wang, Z. Zhao, N. Kuang, H. Chen, J. Liu and Z. Zuo, Appl. Therm. Eng., 2020, 181, 115431.
19. L. Fei, W. Yu, J. Tan, Y. Yin and C. Wang, Adv. Fiber Mater., 2023, 5, 955–967.
20. F. Kim, B. Kwon, Y. Eom, J. E. Lee, S. Park, S. Jo, S. H. Park, B.-S. Kim, H. J. Im, M. H. Lee, T. S. Min, K. T. Kim, H. G. Chae, W. P. King and J. S. Son, Nat. Energy, 2018, 3, 301–309.
21. S. J. Kim, J. H. We and B. J. Cho, Energy Environ. Sci., 2014, 7, 1959–1965.
22. K. Nan, S. D. Kang, K. Li, K. J. Yu, F. Zhu, J. Wang, A. C. Dunn, C. Zhou, Z. Xie, M. T. Agne, H. Wang, H. Luan, Y. Zhang, Y. Huang, G. J. Snyder and J. A. Rogers, Sci. Adv., 2018, 4, eaau5849.
23. S. H. Park, S. Jo, B. Kwon, F. Kim, H. W. Ban, J. E. Lee, D. H. Gu, S. H. Lee, Y. Hwang, J. S. Kim, D. B. Hyun, S. Lee, K. J. Choi, W. Jo and J. S. Son, Nat. Commun., 2016, 7, 13403.
24. C. S. Kim, H. M. Yang, J. Lee, G. S. Lee, H. Choi, Y. J. Kim, S. H. Lim, S. H. Cho and B. J. Cho, ACS Energy Lett., 2018, 3, 501–507.
25. M. K. Kim, M. S. Kim, S. Lee, C. Kim and Y. J. Kim, Smart Mater. Struct., 2014, 23, 105002.
26. Y. Liu, Y. Lu, Z. Wang, J. Li, P. Wei, W. Zhao, L. Chen and K. Cai, J. Mater. Chem. A, 2022, 10, 25644–25651.
27. X. Sun, Y. Hou, Z. Zhu, B. Zhu, Q. Ding, W. Zhou, S. Yan, Z. Xia, Y. Liu, Y. Hou, Y. Yu and Z. Wang, Nat. Commun., 2025, 16, 4220.
28. J. Jin, Y. Hou, C. Li, X. Liang, X. Sun, Q. Ding, X. Zhang, Z. Xia, S. Yan, Y. Yang and Z. Wang, Nano Energy, 2024, 132, 110388.
29. Q. Huang, X. Li, G. Zhang, Y. Kan, C. Li, J. Deng and C. Wang, Appl. Energy, 2022, 309, 118434.
30. K. Sun, H. Dong, Y. Kou, H. Yang, H. Liu, Y. Li and Q. Shi, Chem. Eng. J., 2021, 419, 129637.
31. T. Luo, L. Kong, L. Li, J. Lu, Z. Yu, B. Lin, L. Fu and C. Xu, Chem. Eng. J., 2024, 486, 150443.
32. P. Cheng, K. Wei, W. Shi, J. Shi, S. Wang and B. Ma, J. Energy Storage, 2022, 47, 103581.
33. T. Hu, Y. Li and Y. Tian, et al, Rare Met., 2024, 43, 4412–4424.
34. H. Su, P. Lin, D. Li and Y. Chen, ACS Appl. Mater. Interfaces, 2024, 16, 3334–3347.
35. D. Ao, B. Wu and W. Liu, et al, Rare Met., 2025, 44, 1222–1230.
36. C. Borgnakke, R. E. Sonntag, Fundamentals of Thermodynamics, John Wiley & Sons, 2020.
37. L. Zhou, S. Bustos-Nuño and K. Manohar, et al., Compos. Sci. Technol., 2025, 111258.
38. C. Li, J. Jin and W. Cao, et al, ACS Appl. Mater. Interfaces, 2024, 16, 23406–23415.
39. M. H. Malakooti, N. Kazem and J. Yan, et al., Adv. Funct. Mater., 2019, 29(45), 1906098.
40. S. Lin, L. Zhang and W. Zeng, et al, Commun. Mater., 2022, 3, 44.

---

Footnote

† Zheng Zhu and Zhanglong Xia contributed equally.

---