Scalable-produced 3D elastic thermoelectric network for body heat harvesting

The study addresses a central challenge in wearable thermoelectrics: achieving materials that are simultaneously highly flexible/elastic and deliver strong thermoelectric performance for efficient body-heat harvesting. While flexible electronics are rapidly growing, their endurance is limited by power sources. Thermoelectric generators (FTEGs) are attractive because they directly convert body heat to electricity without relying on light or motion. Conventional bulk inorganic thermoelectric materials and conducting polymers are intrinsically inflexible, and thin films typically operate with in-plane temperature gradients that are less ideal for body-heat harvesting. Device-level 3D approaches combining rigid legs with flexible matrices can boost performance but compromise comfort and reliability due to rigidity and weight. Polymer- or CNT-elastomer composites provide elasticity but have low Seebeck coefficients and poor zT. Therefore, there is a need for an elastic, lightweight, low-thermal-conductivity thermoelectric material with high performance suitable for scalable fabrication and wearable comfort. The authors propose a 3D Ag2Se network fabricated via a simple two-step impregnation method to meet these needs.

Prior work on flexible energy harvesting spans triboelectric and piezoelectric nanogenerators and flexible photovoltaics. For thermoelectrics, strategies include: (1) thin-film/sheet thermoelectrics (e.g., Bi2Te3 films, Ag2Se films) that can bend or stretch but mainly exploit in-plane gradients; (2) 3D bulk-leg devices embedded in elastomers (e.g., Bi2Te3 legs in PDMS) offering high performance but poor wearability due to rigid, heavy legs; and (3) stretchable thermoelectric elastomers formed by mixing conducting polymers or CNTs with elastomers, which provide elasticity yet suffer from low Seebeck coefficients and very low zT (<0.03). Structuring inorganic materials into 3D architectures has emerged as an alternative route. The literature highlights the trade-off between mechanical compliance and thermoelectric efficiency and calls for materials and architectures that combine ultralow thermal conductivity for large temperature differences with mechanical softness and durability for skin conformity.

The authors developed a scalable two-step impregnation process to fabricate a 3D Ag2Se thermoelectric network using a melamine foam template. Step 1 (silvering, inspired by Tollens' reaction): The melamine template is pretreated in NaOH, then sensitized by immersion in SnCl2 solution acidified with HCl to deposit Sn2+. A reductant solution (glucose monohydrate 40 g/L, potassium sodium tartrate tetrahydrate 14 g/L, PEG1000 0.1 g/L in diluted ethanol 100 g/L) is mixed with Tollens' reagent (AgNO3 with NaOH in NH4OH). The Sn2+ reduces Ag+ to nucleate Ag, and the reductant reduces Ag(NH3)2+ to form a continuous silver network within the template. Reaction time is ~5 h; silver loading is tuned by ammonia concentration (~4.5 mol/L optimal) and number of repetitions at fixed AgNO3 (50 g/L) and NaOH (25 g/L). Step 2 (selenization): The silver network is impregnated in an aqueous selenium solution prepared by dissolving Na2S·9H2O (60 g/L) and Se powder (20 g/L) for 10 h to convert the silver network into Ag2Se via in-situ selenization at ambient conditions. Large-scale fabrication was demonstrated in a 2000 × 1000 × 100 mm3 polypropylene tank, enabling samples up to ~1.8 × 0.9 m2. Porosity is tunable from ~99% (pristine) to 95–98% via reaction parameters. Device fabrication: 10 µm-thick Cu foils serve as electrodes attached to Ag2Se networks with silver paste. A module with 40 Ag2Se network legs in series was integrated into a commercial jacket by replacing a 20 × 20 cm2 section of filler. A separate small n–p device combining two p-type Bi2Te3 legs and two n-type Ag2Se networks (4.1 × 2.6 cm2) was also assembled for comparison. Characterization: Porosity (Archimedes method), morphology/EDS (FE-SEM), crystal structure (XRD, Cu Kα), mechanical properties (stress–strain apparatus), Seebeck coefficient and resistivity (ZEM-3, under tensile/compressive strains), thermal conductivity (hot wire method), optical bandgap (FTIR reflectance). Durability tests included cyclic stretching/compression and bending (200 cycles at 5 mm radius). Device output was measured on human skin without heat sinks using Keithley 2400/2182; thermal images via FLIR T620; ambient conditions logged with a portable weather station. Simulation: COMSOL thermoelectric module simulations under natural convection compared temperature fields and outputs for (i) full-filled bulk FTEG (Bi2Te3 legs in PDMS), (ii) 10% filled bulk FTEG, and (iii) Ag2Se network-based FTEG, all at 3 mm thickness. Single-leg open-circuit voltage and power density were computed versus ambient temperature (273–305 K). Material parameters are in Table S2; thickness optimization indicated 3 mm as suitable.

• Scalable fabrication: A simple, ambient-condition two-step impregnation yields large-area (up to ~1.8 × 0.9 m2) 3D Ag2Se networks with tunable porosity (95–99%), easily shaped to match complex surfaces.
• Mechanical properties: The network shows high elasticity with tensile strain >100% and compressive strain >80%. Young’s modulus ~0.03 MPa (tissue-like), enabling intimate skin conformity. Ultralight density ~0.28 g cm−3. Durable under cyclic tensile/compressive loading with minimal resistance drift and survives 200 bending cycles (5 mm radius) without significant deterioration.
• Thermoelectric properties: At room temperature, Seebeck coefficient ~−130 µV K−1; ultralow thermal conductivity ~0.04 W m−1 K−1; room-temperature zT ≈ 0.11, one to two orders higher than previously reported 3D flexible polymer-composite thermoelectrics.
• Strain effect: Seebeck coefficient remains nearly unchanged under strain; power factor changes are mainly due to resistivity variations from geometric deformation.
• Thermal management advantage: Simulations at 3 mm thickness show the network-based FTEG achieves a large internal temperature difference (~6.5 K), about one order higher than a fully filled bulk FTEG and ~3× higher than a 10% filled bulk FTEG, while maintaining the hot-side temperature close to skin temperature (~304 K vs. skin ~306 K), minimizing skin cooling.
• Device performance on-body: Measured open-circuit voltages and power densities on human skin (no heat sinks) are slightly below simulations due to contact resistance but exceed PEDOT/CNT-based FTEGs and are comparable to high-performance bulk-leg FTEGs. The network-based flexible thermoelectric generator achieves output power densities up to ~4 µW cm−2 (reported) and operates continuously for at least 50 h without degradation.
• Wearable integration: A thermoelectric jacket (20 × 20 cm2 module with 40 legs in series) produces ~0.6 mW at 290 K ambient while seated and ~1 mW while walking at 1 m s−1, demonstrating practical body-heat harvesting in daily wear.
• n–p proof-of-concept: A hybrid n–p device (n-type Ag2Se network, p-type Bi2Te3) delivers ~15 mV and ~1.7 µW cm−2 at 297 K (vs. ~0.44 mV and ~1.2 µW cm−2 for a single-leg device), indicating further gains with matched p-type flexible materials.
• Textile compatibility: The impregnation method applies to common fabrics (cotton, linen, silk), enabling thermoelectric textiles with diverse form factors.

The findings demonstrate that structuring an inorganic thermoelectric (Ag2Se) into a 3D elastic, ultralow-thermal-conductivity network can reconcile the typical trade-off between flexibility and performance in wearable thermoelectrics. The network architecture enables large internal temperature differences at small thicknesses and near-skin hot-side temperatures, improving both output and comfort versus bulk-leg FTEGs that require tall legs, reduced fill factors, or heat sinks. The tissue-like modulus and low density enhance wearability and reliability under deformation, while the network shows stable electrical performance under repeated strains and bending. On-body tests and jacket integration validate practical energy harvesting from body heat at milliwatt levels, sufficient for many low-power sensors and ICs. Simulations and hybrid n–p tests suggest that pairing the Ag2Se network with suitable p-type flexible materials and low-resistance, compliant electrodes could further improve voltage and power density. The method’s scalability and compatibility with fabrics indicate a pathway toward mass-producible thermoelectric wearables and personalized thermoregulation.

This work introduces a facile, scalable two-step impregnation method to fabricate a 3D Ag2Se thermoelectric network combining elasticity, ultralight weight, ultralow thermal conductivity, and competitive thermoelectric performance (zT ~0.11). The network-based FTEG achieves high on-body power density (up to ~4 µW cm−2), robust durability, and practical milliwatt-level output when integrated into a jacket, rivaling bulk-based flexible thermoelectrics while significantly improving comfort and conformability. The approach extends to textiles, enabling application-specific thermoelectric fabrics. Future directions include: developing matched high-performance flexible p-type materials to build fully flexible n–p modules; reducing contact resistance and thermal bypass with compliant, low-resistance electrodes (e.g., liquid metals) and optimized interconnects; geometric optimization of leg density, thickness, and module architecture; and long-term wear and laundering studies for textile-integrated devices.

• The measured device outputs are slightly below simulations due to contact resistance and non-ideal interfaces (e.g., untight hot-side contact in the jacket, hand-assembled gaps).
• Current modules use only n-type legs; single-polarity designs increase internal resistance and can introduce thermal bypass when thicker electrodes are used.
• A demonstrated n–p device used rigid p-type Bi2Te3, which compromises portability and flexibility; matched flexible p-type materials are still needed.
• Initial minor mechanical degradation under the first tensile cycles was observed, attributable to fibril alignment and template viscoelasticity.
• Reported performance metrics are at room temperature on human skin without heat sinks; broader environmental testing and long-term wearable conditions (e.g., sweat, washing) are not detailed.