Fully printed origami thermoelectric generators for energy-harvesting

With the ongoing digitization of manufacturing (Industry 4.0) and the growth of the Internet of Things (IoT), billions of autonomous sensors require sustainable power sources. Thermoelectric generators (TEGs) provide maintenance-free energy harvesting from small temperature differences. Conventional bulk Bi₂Te₃ TEGs are limited by high costs and Te availability. Printed TEGs using earth-abundant elements and scalable printing (screen, inkjet, 3D printing) offer customizable geometries enabling thermal impedance matching and reduced manufacturing complexity and cost. Thermoelectric performance is characterized by Seebeck coefficient S, electrical conductivity σ, and thermal conductivity κ, with figure of merit ZT = σS²T/κ influencing maximum efficiency. Practical TEGs connect many n- and p-type elements thermally in parallel and electrically in series; the device-level effective ZT depends on the number of thermocouples, Seebeck coefficients, total thermal conductance K, and total electrical resistance R. Optimal power extraction requires both thermal and electrical impedance matching (K_contact and R_load conditions). Thin printed films typically have too high cross-plane thermal conductance because of micrometer thicknesses. Applying ΔT in-plane to printed legs decouples device thickness from film thickness, enabling impedance matching via geometry. Corrugated architectures fold printed rows into meanders but risk electrical shorts due to flexible substrates and require insulation between adjacent thermoelectric legs. The present work introduces a two-step origami folding technique that uses the substrate itself as insulation without extra non-TE materials, reducing parasitic heat paths. A fully printed robust origami TEG achieves a high thermocouple density (~190 cm⁻²) and high power density (up to 47.8 µW cm⁻² at ΔT = 30 K), sufficient to power a Bluetooth-based weather sensor for Industry 4.0 and IoT applications.

The paper reviews thermoelectric energy harvesting for wearables, sensor networks, and waste heat recovery, highlighting limitations of bulk Bi₂Te₃ devices (cost and tellurium scarcity) and the promise of printed TEGs using earth-abundant materials. Large-scale printing (screen, inkjet, additive) enables complex, customizable geometries, reduced processing steps, and potential cost reductions. Prior flexible and corrugated TEG architectures have been reported, but often fail to achieve simultaneous thermal and electrical impedance matching, limiting converted energy. Corrugated thin-film approaches enable in-plane heat flow but require reliable electrical insulation between legs and can suffer from mechanical instability. The authors build on these insights to propose an origami folding approach that integrates insulation via the substrate and allows layout-based impedance tuning.

Device architecture and folding: A checkerboard layout of p-type (PEDOT nanowires) and n-type (TiS₂:hexylamine complex) legs is screen printed on a 6 µm PEN substrate. Legs in each column are electrically connected via small overlap regions; columns are interconnected by a higher-conductivity p-type strip (or optional metal). The folding is performed in two series: (1) columns are stacked with an added strip of unprinted substrate (width equal to one column, 10.62 mm) folded over the first column to serve as insulation, yielding a multilayer ribbon with 13 layers of TE legs separated by 14 substrate layers; (2) the ribbon is creased with a hot blade along fold lines and folded into a corrugated pattern such that hot and cold sides alternate, forming a compact cuboid. Kapton tape fixes the final shape. This origami aligns TE legs vertically for in-plane heat flow and uses the substrate as inter-leg insulation to avoid shorts without introducing extra non-TE materials.

Layout and geometry: The printable area (154 mm × 156 mm) contains 254 p-legs and 253 n-legs arranged in 13 columns × 39 rows. Each element is 10.62 mm wide and 4.17 mm tall (including a 0.3 mm overlap). Columns are spaced by 1.5 mm and connected by 2 mm wide p-type strips. After folding, the cuboid measures approximately 12.5 × 10.6 × 4.1 mm. The resulting thermocouple density is ~190 cm⁻². A quick estimation using material thermal conductivities (PEN ≈ 0.22 W m⁻¹ K⁻¹; air 0.0264 W m⁻¹ K⁻¹) indicates the origami’s inlaid substrate reduces thermal resistance by 24.53% compared to a 6 µm air gap.

Materials and printing: PEDOT nanowires (p-type) were synthesized following Han and Foulger, and formulated into a screen-printable ink (ethylene glycol solvent). TiS₂:hexylamine complex (n-type) was prepared per Wan et al. and formulated in N-methylformamide. Screen printing used a Kammann K15 Q-SL semi-automatic machine onto 6 µm PEN. Two layers of each material were printed with drying at 140 °C for 2 min after each pass, yielding ~7 µm thickness per material. The first and last series elements were extended with p-type ink (via calligraphy) and heat-gun dried to create accessible contact pads.

Measurements: Material properties were characterized on doctor-bladed films: Seebeck coefficient via an in-house setup; electrical conductivity via four-point probe on patterned strips; thickness measured with a Bruker Contour GT-K 3D Optical Microscope; thermal conductivities taken from literature. Devices (N=24) were electrically characterized before and after folding to assess internal resistance. Thermoelectric performance was measured using two temperature-controlled copper blocks (with thermal paste at interfaces) to apply precise ΔT (uncertainty ±0.5 K). Electrical measurements had 0.33% uncertainty. I–V curves, open-circuit voltage (VOC) versus ΔT, and power versus load were recorded. Impedance matching behavior and maximum power point (MPP) were analyzed.

Application demonstration: A customized wireless sensor node was powered by the TEG via a power management IC (Texas Instruments BQ25570) with MPP tracking and a 57.5 µF capacitor (6.3 V) for energy storage. The node included a Bosch BME280 sensor and a BLE module transmitting every 4 s at 1.8 V (provided by an on-chip step-up). Average current draw of the sensor node was 8.3 µA (14.9 µW). The power management drew an average of 194.3 µA at ~0.316 V from the TEG, with initial charging and MPP stabilization at 0.267 V and 238 µA. Total average system power was 61.3 µW.

Scalability and tunability: Screen printing affords high layer thickness per pass and scalability (including roll-to-roll). Device thermal conductance can be tuned over orders of magnitude by adjusting element length (and hence generator height), enabling thermal impedance matching. Electrical resistance is tuned via element width and column count, balancing required voltage and avoiding shorts.

• Fully screen printed, flexible origami TEGs using PEDOT nanowires (p-type) and TiS₂:hexylamine (n-type) were realized and folded into compact cuboids with vertically aligned TE legs and substrate-based insulation.
• High thermocouple density: ~190 cm⁻² due to 6 µm substrate spacing and compact stacking.
• Power and voltage performance (best device, TEG #6): • VOC scales linearly with ΔT with slope 19.7 mV K⁻¹. • At ΔT = 60 K: P_MPP = 243 µW (several 100 mV output). • At ΔT = 30 K: P_MPP = 63.4 µW, VOC = 534 mV, internal resistance R ≈ 1124 Ω, power density = 47.8 µW cm⁻². • Power vs. load peaks near R_load ≈ R across ΔT, consistent with negligible Peltier effects due to low effective ZT.
• Across 24 devices at ΔT = 30 K (25 °C cold, 55 °C hot): average P_MPP = 40.5 µW ± 28.9%, average VOC = 394 mV ± 16.5%.
• Internal resistance changed from 8400 Ω ± 7.46% (before folding) to 10070 Ω ± 27.5% (after folding), attributed to microcrack formation during folding; some devices showed reduced resistance due to occasional shorts from large TiS₂ particles piercing the substrate.
• Parasitic thermal pathway: using PEN substrate as inlaid insulation reduces the thermal resistance by 24.53% compared to a 6 µm air gap, impacting efficiency but enabling robust insulation without added bulk.
• Wireless sensor demonstration: The TEG powered a BLE weather sensor via a BQ25570-based power management system. The sensor averaged 14.9 µW at 1.8 V; the full system consumed 61.3 µW on average, below TEG #6’s available power at ΔT = 30 K.
• Tunability: Device thermal conductance and electrical resistance are adjustable via layout (element length/width, column count), enabling thermal and electrical impedance matching for diverse heat sources.

The study addresses the challenge of creating low-cost, scalable thermoelectric generators capable of harvesting low-grade heat for IoT and Industry 4.0 applications. By printing TE materials and employing an origami folding scheme that uses the substrate as insulation, the authors overcome limitations of thin-film cross-plane conductance and mechanical instability in prior corrugated designs. The vertical alignment of TE legs supports in-plane heat flow, allowing device thickness (and thus thermal conductance) to be set by printed leg length rather than film thickness, enabling thermal impedance matching via layout adjustments. The high thermocouple density yields macroscopic voltages at moderate ΔT, and measured power outputs (e.g., 63.4 µW at 30 K with 534 mV VOC) are sufficient for low-power electronics. Although the inlaid substrate introduces a parasitic thermal path (reducing thermal resistance by ~24.5% vs. an air gap) and the effective ZT remains modest, the demonstrated performance, reproducibility across 24 devices, and successful powering of a BLE weather sensor validate the approach. The platform’s print-layout tunability provides a practical route to match both thermal and electrical impedances to specific heat sources/sinks and loads, advancing printed TEGs toward real-world deployment.

The paper introduces a fully screen printed, origami-folded TEG architecture that combines mass-producible fabrication with a compact, robust 3D form factor using the substrate itself for electrical insulation. The design achieves high thermocouple density, tunable thermal/electrical impedances via print layout, and delivers sufficient power from modest temperature differences to operate a wireless BLE weather sensor. The approach is compatible with various screen-printable TE materials, offering a path toward low-cost, customizable, and scalable energy-harvesting solutions for autonomous sensors and waste-heat recovery. Potential future work includes improving material stability (e.g., encapsulation for TiS₂ to prevent oxidation), further reducing parasitic thermal conduction (thinner/low-κ substrates), enhancing print uniformity and mechanical robustness to mitigate microcracks, optimizing power management efficiency at low input voltages, and scaling to roll-to-roll manufacturing for larger-area modules.

• Parasitic heat flow through the inlaid PEN substrate reduces thermal resistance by ~24.53% compared to an equivalent air gap, lowering efficiency.
• TiS₂ (n-type) oxidizes within hours under ambient conditions, necessitating encapsulation (e.g., pouch sealing) which can further reduce effective ΔT.
• Manual folding introduced variability; post-folding internal resistance increased (microcracks at edges), with higher variance across samples.
• Occasional shorts due to large TiS₂ particles piercing the 6 µm substrate.
• Low effective ZT leads to negligible Peltier effects and constrains conversion efficiency; overall power remains modest for higher-power loads.
• Power management efficiency at sub-0.5 V inputs is limited, reducing end-to-end system efficiency.
• The use of substrate insulation, while robust, inherently adds a parallel thermal path; optimization requires careful trade-offs between mechanical stability and thermal performance.