Sustainable poly(butylene furanoate)/carbon nanotube composite fibers for wearable thermoelectric generators

Abstract

Due to the growing interest in renewable energy, thermoelectric materials that are capable of directly converting heat into electricity have attracted significant attention, particularly for wearable thermoelectric generators (WTEGs) that can harvest body heat. Herein, a sustainable composite fiber is developed by combining the bio-based polymer poly(butylene furanoate) (PBF) with single-walled carbon nanotubes (SWCNTs) via wet spinning. Post-annealing of the composite fibers at 150 °C enhances the crystallinity of the PBF and promotes strong π–π interactions with the SWCNTs, thereby resulting in a denser nanotube network along with a three-fold increase in electrical conductivity (113.0 ± 11.6 S cm⁻¹) and a maximum power factor of 7.60 ± 0.79 µW m⁻¹ K⁻². The composite fiber also exhibits excellent flexibility, air stability, and thermal durability. For device integration, n-type fibers doped with N-DMBI are woven with the as-fabricated p-type fibers to construct a WTEG. The device with 30 p–n junctions delivers an output voltage and power of 29.4 mV and 15.72 nW, respectively, under a temperature gradient of 30 K. When worn on the human wrist and forehead, the WTEG generates voltages of 1.5 and 3.1 mV, respectively. These results highlight the potential of PBF/SWCNT composites for sustainable, flexible thermoelectric energy harvesting in wearable electronics.

---

Introduction

Amid growing concerns over global energy shortages and environmental sustainability, thermoelectric materials have attracted considerable interest due to their ability to directly convert waste heat into electricity via the Seebeck effect.^1–3 With the rapid proliferation of the Internet of Things (IoT), wearable thermoelectric generators (WTEGs) have emerged as promising self-powered solutions for harvesting body heat in order to energize low-power wearable and portable devices. The intrinsic self-powered capabilities of the WTEGs, along with their flexibility, lightweight form factor, and seamless integration into textiles, make them particularly strong candidates for use as next-generation smart wearable devices.^4–7

Among various TE material geometries, one-dimensional (1D) fibers offer unique advantages over their bulk or thin-film counterparts for wearable applications. Their mechanical flexibility, ease of integration into fabrics, and compatibility with curved surfaces such as human skin position them as ideal candidates for wearable thermoelectrics.^8–10 For example, Wen et al. fabricated a wet-spun thermoelectric fiber from poly(3,4-ethylenedioxythiophene):poly(styrenesulfonic acid) (PEDOT:PSS) blended with waterborne polyurethane (WPU) to achieve an output voltage of 0.44 mV on the human wrist when implemented in a WTEG.¹¹ Moreover, a wet-spun PEDOT:PSS/silver telluride (Ag₂Te) nanowire composite fiber produced an output of 4.2 mV under a temperature gradient of 25 K when integrated into non-woven fabric.¹² Furthermore, other studies have shown that 1D fibers can be woven into three-dimensional (3D) flexible fabrics in order to generate a vertical temperature gradient across the human skin, thereby enhancing the efficiency of body heat harvesting.^10,13,14 For instance, Li et al. immersed carbon nanotube (CNT) yarns in either polyaniline (PANI) or polyethyleneimine (PEI) to obtain p- and n-type thermoelectric fibers, respectively. When assembled into a flexible silicone strip, the resulting WTEG achieved an output power of 377 nW, thereby demonstrating the potential of fiber-based architectures.¹⁵

Among various thermoelectric fillers, single-walled carbon nanotubes (SWCNTs) are regarded as compelling alternatives to conventional inorganic materials due to their high electrical conductivity, low toxicity, and cost-effectiveness. In addition, the conduction mechanism of SWCNTs can be readily converted from p- to n-type by doping with molecules such as oleylamine or tetrabutylammonium salts.^16–18 However, their strong intrinsic π–π interactions often result in bundled agglomerates, which disrupt the percolation network and reduce the charge carrier transport efficiency, thereby highlighting the importance of strategies for reducing the aggregation of nanotubes in order to enable the development of thermoelectric composites. Therefore, polymer matrices are commonly used to improve nanotube dispersion and promote the formation of a conductive network.^19–23 For example, Liu et al. used poly(ethylene terephthalate) (PET) to enhance the conductivity of SWCNT-based thin films.²⁴ Nevertheless, given the increasing demand for sustainability, it is desirable to replace petroleum-derived polymers with bio-based alternatives. Polymers such as polyesters and polylactic acid (PLA) have shown potential in energy and electronic devices.^25–27 For example, in previous work, polyethylene furanoate (PEF) from renewable feedstocks was used as a PET substitute to provide a flexible substrate for TEGs with superior flexibility and good thermal stability.²⁸ Besides, Lam et al. fabricated an organic photovoltaic device and an environmentally friendly memory device on PEF substrates, thereby proving the potential of PEF in various applications.^29,30 Among various kinds of bio-based polymers, furan-based polyesters possess a furan ring in their backbone, which offers the potential for stronger π–π interactions between the polymer and SWCNTs and further achieves a more uniform dispersion of the SWCNTs, which is important for improving the overall thermoelectric performance. Among the polyesters, poly(butylene furanoate) (PBF) synthesized from the sugar-derived monomers, 2,5-furandicarboxylic acid (FDCA) and 1,4-butanediol (BDO), has emerged as a particularly promising bio-based alternative to PBT. Distinguished by a relatively high glass transition temperature of 40 °C, PBF offers superior ambient stability, excellent gas barrier properties, and good thermal processability owing to its lower melting temperature of 170 °C, enabling tuning of the crystallinity of PBF, thereby optimizing the thermoelectric performance.^31–35 Moreover, the presence of butylene segments imparts a lower Young's modulus and higher elongation compared to PEF, thus rendering PBF softer and more compliant. This softer nature allows the incorporation of PBF to enhance the overall flexibility of the composite, which is advantageous for wearable electronics.^36,37

The present study examines the fabrication and characterization of a flexible, high-performance WTEG based on the integration of PBF/SWCNT composite fibers into a non-woven textile substrate. The composite fibers are prepared by wet spinning and subjected to thermal annealing in order to optimize their microstructure and thermoelectric properties. The results indicate that thermal annealing is particularly effective in promoting polymer chain reorganization and crystallization, as well as facilitating the formation of conductive pathways by enhancing the filler dispersion and interfacial interactions. Microbeam X-ray diffraction (XRD) and Raman spectroscopy confirm that annealing promotes PBF crystallinity and strengthens π–π stacking interactions between PBF and SWCNTs, thus leading to tighter polymer wrapping and more compact nanotube bundles. These morphological improvements significantly enhance the charge carrier transport and electrical conductivity, thereby resulting in an optimized power factor of 7.60 ± 0.79 µW m⁻¹ K for the 1 : 1 composite fiber after annealing at 150 °C. Additionally, the optimized composite fibers demonstrate excellent mechanical flexibility, ambient air stability, and thermal stability. Moreover, to demonstrate their practical applications, n-type fibers doped with N-DMBI are combined with p-type fibers to construct a fiber-based WTEG. Compared to analogous systems reported in the literature, the as-fabricated device composed of 30 p–n junctions generates superior output voltages of 1.5 and 3.1 mV when worn on the wrist and forehead, respectively (Table S1).^8,11,38,39 These findings validate the performance and durability of the PBF/SWCNT-based WTEG and highlight its promising potential for self-powered wearable electronics and sustainable energy harvesting platforms.

Results and discussion

Fabrication of composite fibers

The molecular structures of the components used in the present study are displayed in Fig. 1(a), including PBF and SWCNTs. As shown schematically in Fig. 1(b), the composite fibers consisting of PBF and SWCNTs were fabricated via a wet-spinning method. The spinning solution was prepared by ball-milling to ensure a homogeneous dispersion of SWCNTs in the polymer matrix, followed by injection into an ethanol coagulation bath. The as-fabricated fibers were then prepared with different post-treatments, including no thermal annealing (denoted as w/o), annealing at 100 °C for 1 h (denoted as 100) and annealing at 150 °C (denoted as 150) for 1 h to enhance their thermoelectric properties. These composite fibers were labeled based on the PBF : SWCNT weight ratio (1 : 1, 2 : 1, and 3 : 1) and the specific post-treatment condition (e.g., 1 : 1 100 for 100 °C annealing of the 1 : 1 composite).

Fig. 1 (a) Chemical structures and key properties of PBF and SWCNTs. (b) Schematic illustration of the fabrication procedure for the PBF/SWCNT composite fibers.

The successful fabrication and morphological integrity of the fibers are confirmed by the scanning electron microscopy (SEM) results shown in Fig. 2(a–c) and S1. The composite fibers remain linear and uncurved, regardless of the post-treatment condition. However, the average fiber diameter is seen to decrease slightly from approximately 70 µm for the 1 : 1 w/o sample to ∼67 µm for both the 1 : 1 100 and 1 : 1 150, thereby indicating minor densification due to the thermal treatment. Furthermore, cross-sectional SEM images in Fig. 2(d–f) reveal that the fibers exhibit an elliptical geometry with an average cross-sectional area of approximately 4000 µm². Further examination at higher magnification (Fig. S2) confirms the absence of any visible aggregation or clustering of the SWCNTs, thereby demonstrating that the ball-milling process enabled uniform dispersion of the carbon nanotubes within the PBF matrix. This homogeneity is critical for effective charge carrier transport and consistent thermoelectric performance throughout the fiber.

Fig. 2 SEM images of composite fibers: (a) 1 : 1 w/o, (b) 1 : 1 100 and (c) 1 : 1 150. Cross-section SEM images of (d) 1 : 1 w/o, (e) 1 : 1 100 and (f) 1 : 1 150.

Interactions between PBF and SWCNTs

The effect of annealing on PBF crystallization is revealed by the differential scanning calorimetry (DSC) results in Fig. S3. For the pristine PBF, the glass transition temperature (T_g) is found to be around 40 °C, and the melting temperature (T_m) is 170 °C. Therefore, to simultaneously ensure structural integrity and enhance the thermoelectric performance of the composite fibers, the investigated annealing temperatures were selected as 100 °C and 150 °C, i.e., between the T_g and the T_m of the pristine PBF, to prevent PBF from melting and subsequently disrupting the integrity of the composite fibers.

The molecular-level interactions and packing behavior of the PBF and SWCNTs depending on the post-treatment conditions are elucidated by the microbeam X-ray diffraction (XRD) results in Fig. 3(a) for the composite fibers and in Fig. S4 for the pristine PBF. In the case of pristine PBF, distinct reflections are observed at scattering vectors (q) of 0.76, 1.28, 1.56 and 1.76 Å⁻¹ (corresponding to interplanar spacings (d) of 8.26, 4.91, 4.02 and 3.57 Å, respectively) due to the (001), (010), (002) and (100) crystal planes of PBF. After annealing at 100 or 150 °C, sharper peaks are observed around these four crystalline peaks, thereby indicating enhanced molecular ordering. Additionally, a slight shift of the (100) peak toward higher q values is observed, thereby suggesting a reduced interplanar spacing, which implies tighter molecular packing after thermal treatment. For the composite fibers, the XRD patterns in Fig. 3(a) reveal significant changes upon thermal annealing.^40–42 Compared to the 1 : 1 w/o and 1 : 1 100 samples, the 1 : 1 150 fibers exhibit a more intense diffraction peak at q = 1.61 Å⁻¹ (d = 3.90 Å), which relates to the (002) reflection of PBF. This suggests that thermal annealing at 150 °C further promotes the crystallization of the PBF phase within the composite. Notably, an additional diffraction peak emerges at q = 2.13 Å⁻¹ in the 1 : 1 150 sample, which is absent from both the pristine PBF and the pure SWCNTs. This peak can be attributed to an SWCNT-induced transformation in the crystalline structure of PBF. The presence of this new reflection, as well as the intensified crystalline peaks, supports the formation of a denser and more ordered molecular arrangement in the composite fiber, enabled by annealing-driven interfacial reorganization.⁴³

Fig. 3 (a) Diffraction profiles of 1 : 1 composite fibers subjected to different annealing processes. (b) Raman spectra of the composite fibers. (c) A magnified view of the Raman spectra in a range of 1570 to 1610 cm⁻¹. (d) Tensile stress–strain curve of 1 : 1 w/o, 1 : 1 100 and 1 : 1 150 composite fibers.

The non-covalent intermolecular interactions between PBF and SWCNTs are further confirmed by the Raman spectroscopy results, as shown in Fig. 3(b). Here, pure SWCNTs exhibit characteristic G and D bands at ∼1592 and ∼1340 cm⁻¹ due to C–C bond stretching vibrations and the presence of structural disorder, respectively. Moreover, the absence of new Raman peaks upon addition of PBF suggests that no covalent bonding occurs, and the interaction is primarily governed by non-covalent π–π stacking. The intensity ratio of the D to G bands (I_D/I_G) is used to assess the level of structural defects in the SWCNTs. Without annealing, the I_D/I_G value increases slightly with the increase in PBF content, from 0.041 at 1 : 1 PBF : SWCNT, through 0.048 at 2 : 1, to 0.053 for 3 : 1. This indicates that incorporation of the insulating polymer introduces structural defects or disrupts the sp² carbon framework in the SWCNTs. After annealing, however, the I_D/I_G ratio is significantly decreased to 0.023 and 0.018 for the 1 : 1 100 and 1 : 1 150, respectively. This suggests that thermal annealing helps to restore the nanotube integrity by promoting more intimate and uniform wrapping of PBF around the SWCNTs, thereby minimizing defect-induced scattering and enhancing the structural order of the nanotubes.⁴⁴ Furthermore, as shown in Fig. 3(c), a blueshift is observed in the G band from 1591.1 before annealing to 1592.5 cm⁻¹ afterward regardless of the temperature, which is indicative of stronger C–C bond vibrations. This spectral shift provides additional evidence of denser packing and increased intermolecular interactions between the PBF and SWCNTs upon thermal treatment and these effects are expected to improve the charge carrier transport and overall electrical conductivity of the composite.

To assess the suitability of the composite fibers for wearable applications, their mechanical properties are evaluated in Fig. 3(d) and summarized in Table S2. Thus, Young's modulus is seen to increase from 2417 MPa for the 1 : 1 w/o sample to 2942 and 3410 MPa for the 1 : 1 100 and 1 : 1 150 samples, respectively, thereby indicating enhanced mechanical stiffness due to increased crystallinity and stronger polymer/nanotube interactions. These results are consistent with the above-mentioned XRD and Raman results. In addition, compared to the pristine SWCNTs with a Young's modulus of 16.4 GPa, the composite fibers exhibit significantly improved flexibility and elongation at breaking point, which are critical properties for integration into wearable electronics.

Morphologies of the composite fibers

The surface morphologies of the composite fibers are revealed by SEM images in Fig. 4(a–f) and S4. These images elucidate the structural distribution and interfacial arrangement of the SWCNTs and PBF within the wet-spun fibers. As shown in Fig. 4(a–c), the SWCNTs are homogeneously dispersed throughout the 1 : 1 w/o composite fiber to form an interconnected network structure that remains largely intact even after thermal annealing. This preserved network provides continuous and efficient pathways for charge carrier transport, which is critical for maintaining high electrical conductivity. At higher magnification, the SWCNT bundles exhibit a rough and irregular surface morphology (Fig. 4(d)), which is distinct from the smooth morphology of the pristine SWCNTs. This roughness suggests that the PBF chains wrap around the SWCNT surfaces in a non-uniform manner. Such uneven polymer encapsulation can locally interrupt the conductive pathways and hinder charge carrier transport. Moreover, as shown in Fig. S5, larger polymer clusters are observed with increasing PBF loading, thereby indicating that excessive polymer content results in a pronounced deterioration of the conductive SWCNT network. As shown in Fig. 4(e and f), thermal annealing leads to a notable refinement in morphology, particularly for the 1 : 1 150 sample, where the surfaces of the SWCNT bundles become significantly smoother and more continuous. This morphological transition is ascribed to the enhanced mobility of the PBF chains during annealing, which leads to the flattening of the PBF-air surface and, hence, a reduction in the overall surface area. This, in turn, promotes closer contact between SWCNT bundles. At the same time, annealing facilitates the crystallization of PBF, as discussed before. These conditions promote stronger π–π interactions at the interface, which result in a denser and more ordered composite structure. This morphological evolution is summarized schematically in Fig. 4(g). Thus, upon annealing, the PBF chains rearrange more uniformly around the SWCNT bundles, thereby enhancing the interfacial adhesion and minimizing any structural defects. These improvements in interfacial quality, network uniformity, and crystalline order serve to lower the contact resistance, which is critical for facilitating efficient charge carrier transport within the composite fiber, thereby contributing to enhanced thermoelectric performance.⁴⁵

Fig. 4 (a–f) SEM images of 1 : 1 w/o, 1 : 1 100 and 1 : 1 150 composite fibers (a–c) at 30k× and (d–f) 100k× magnification. (g) Schematic illustrations of fiber morphologies before and after annealing.

Thermoelectric properties of the composite fibers

The thermoelectric properties of the PBF/SWCNT composite fibers are systematically evaluated in Fig. 5 and summarized in Table S3. Here, both the effects of varying the PBF-to-SWCNT weight ratio and the post-treatments are examined. Thus, as the PBF content increases, the electrical conductivity of the composite fibers decreases significantly, from 42.9 ± 2.7 S cm⁻¹ for the 1 : 1 w/o sample to 8.35 ± 0.01 S cm⁻¹ for the 3 : 1 w/o sample (Fig. 5(a)). This reduction is attributed to the increasing proportion of electrically insulating polymer, which disrupts the percolation network of conductive SWCNTs. Meanwhile, the Seebeck coefficient remains relatively constant across the various PBF/SWCNT ratios (Fig. 5(b)). Consequently, as the PBF content increases, the power factor decreases from 2.13 ± 0.10 µW m⁻¹ K⁻² for the 1 : 1 w/o sample to 0.40 ± 0.04 µW m⁻¹ K⁻² for the 3 : 1 w/o sample (Fig. 5(c)). Consequently, thermal annealing significantly enhances the electrical conductivity of the composite fibers, giving values of 98.2 ± 27.7 S cm⁻¹ and 113.0 ± 11.6 S cm⁻¹ for the 1 : 1 100 and 1 : 1 150 samples, respectively (Fig. 5(d)). Notably, these values represent approximately 2- and 3-fold improvements over the non-annealed counterpart. Notably, this trend contrasts with the annealing-induced degradation typically observed in pure SWCNTs, thereby implying that the improvement arises from beneficial synergistic interactions between the PBF and SWCNTs. This improvement is attributed to increased crystallinity of the PBF matrix and pronounced π–π interactions at the PBF/SWCNT interface. Upon annealing, the PBF chains gain sufficient mobility to wrap more uniformly around the SWCNTs, thereby facilitating tighter bundle formation and a more continuous conductive network. These structural refinements contribute to the observed increase in charge transport ability. The Seebeck coefficient also shows a positive response to annealing, increasing to 26.0 ± 1.2 µV K⁻¹ for the 1 : 1 150 sample (Fig. 5(e)). A similar trend is observed in the pristine SWCNTs when subjected to annealing (Fig. S6), which is likely due to n-type doping induced by the introduction of nitrogen-containing functional groups during thermal processing. Due to the concurrent enhancements in both electrical conductivity and the Seebeck coefficient, the power factor of the annealed composite fibers increases markedly from 5.44 ± 1.51 µW m⁻¹ K⁻² for the 1 : 1 100 sample to 7.60 ± 0.79 µW m⁻¹ K⁻² for the 1 : 1 150 sample (Fig. 5(f)). These findings clearly confirm that thermal annealing is an effective strategy for improving the thermoelectric performance of PBF/SWCNT composite fibers, driven by interfacial ordering and polymer/nanotube synergy. Compared to previous studies (Table S4), despite a modest thermoelectric performance, the composite fiber 1 : 1 150 successfully demonstrates a crucial balance between thermoelectric performance and mechanical robustness, making it a candidate for wearable electronics.

Fig. 5 The electrical conductivity, Seebeck coefficient and power factor of composite fibers (a–c) with different ratios of PBF to SWCNTs and (d–f) subjected to different post-treatment.

Stability test for composite fibers

To assess the practical applicability of the PBF/SWCNT composite fibers, the mechanical flexibility, ambient air stability, and thermal stability of the optimized 1 : 1 150 sample were systematically investigated. First, the robustness of the composite fiber under mechanical deformation was evaluated during cyclic bending to a radius of 0.4 cm (corresponding to 80% strain), as shown in Fig. S7, and the thermoelectric performance was measured both before and after bending for various numbers of cycles. As shown in Fig. 6(a), both the electrical conductivity and Seebeck coefficient remained nearly unchanged after bending, thus resulting in a negligible variation in the power factor. Specifically, the power factor retention is approximately 94% after 1000 bending cycles, thereby confirming the excellent mechanical flexibility of the fiber.

Fig. 6 Thermoelectric properties of composite fibers after (a) cyclic bending test, (b) storage for different durations, and (c) at different base temperatures.

In addition, the environmental stability of the composite fiber was examined by storing the 1 : 1 150 sample under ambient conditions (relative humidity (RH) = 50%, temperature = 25 °C) for 30 days. As shown in Fig. 6(b), the electrical conductivity, Seebeck coefficient, and power factor exhibit minimal degradation over this period, thereby demonstrating outstanding air stability, which is an essential attribute for wearable and outdoor thermoelectric applications. Furthermore, the results in Fig. 6(c) clearly demonstrate the operational robustness of the 1 : 1 150 composite fiber at temperatures ranging from 30 to 80 °C. Here, both the electrical conductivity and Seebeck coefficient show consistent values throughout the entire temperature range. This stability ensures a reliable performance for applications involving fluctuating thermal environments, such as body heat harvesting or industrial waste heat recovery. These results collectively demonstrate that the PBF/SWCNT composite fibers exhibit a unique combination of flexibility, durability, and environmental tolerance, which makes them highly promising candidates for integration into next-generation wearable thermoelectric devices.

Wearable thermoelectric generators (WTEGs)

To realize practical TEG devices, the integration of both p-type and n-type thermoelectric materials is essential. Hence, the as-prepared PBF/SWCNT fibers (1 : 1 w/o) were immersed in an ethanol solution of N-DMBI for 15 min to confer n-type behavior, with a Seebeck coefficient of −18.6 ± 1.8 µV K⁻¹ and an electrical conductivity of 121.3 ± 8.2 S cm⁻¹. Notably, these values are comparable to those of the optimized p-type fiber (i.e., the 1 : 1 150). Moreover, the n-type fibre exhibits a relatively high power factor of 4.20 ± 0.54 µW m⁻¹ K⁻², and is therefore suitable for TEG device fabrication.

Next, a nonwoven fabric substrate with a thickness of 3 mm was used to assemble a flexible TEG. The as-fabricated p- and n-type fibers were alternately woven into the substrate and electrically connected by using carbon tape and silver paste, as shown in Fig. 7(a). The reliability and responsiveness of the TEG were then evaluated by using a configuration containing five p–n junction pairs. As shown in Fig. 7(b), the output voltage (ΔV) increases proportionally with the applied temperature gradients (ΔT), reaching values of 0.19, 0.38, 0.57, and 0.78 mV at ΔT values of 4.2, 8.4, 12.9, and 17.2 K, respectively. Moreover, the rapid response and durability of the device are further demonstrated by a stable voltage output over time.

Fig. 7 (a) Schematic illustration of the WTEGs. (b) Voltage response of a TEG with 5 p–n pairs. Output voltage and power for the TEG (c) with different numbers of p–n junction pairs and (d) varying ΔT. Photographs showing the TEG in operation on the (e) human wrist and (f) human forehead, as well as the corresponding output voltages.

To enhance the output performance of the device, additional fiber pairs were incorporated. As shown in Fig. 7(c) and S8, the as-fabricated TEGs with 10, 20, and 30 p–n pairs generate output voltages of 11.3, 21.6, and 29.4 mV, respectively, under a constant temperature gradient of 30 K. Although a slight voltage drop is observed due to increased internal resistance, the linear increase in ΔV with the number of junctions confirms the scalable device architecture. Moreover, the corresponding output power reaches 7.13, 13.15, and 15.7 nW, respectively, thereby demonstrating that an increase in the number of p–n pairs effectively boosts the device performance, with greater output voltage and power under the same ΔT.

Next, the as-fabricated WTEG with 30 p–n pairs was further characterized under various temperature gradients, with a measured internal resistance of 14.13 mΩ (Fig. 7(d) and S9). Consistent with the earlier measurements in Fig. 7(b), ΔV exhibits a linear dependence on ΔT, thereby demonstrating that the increased number of series-connected p–n junctions does not compromise the applicability of the TEG. The near-identical slopes for the voltage–current (V–I) curves across different temperatures indicate the stable resistance of the device. Moreover, the maximum output power is 1.41, 6.26, and 15.72 nW under a ΔT values of 10, 20, and 30 K, respectively.

Finally, to demonstrate the practical applicability of the device, the TEG was worn on a human wrist and forehead, as shown in Fig. 7(e and f). When worn on the wrist, a ΔT of 1.6 K between the skin and ambient side results in an output voltage of 1.5 mV, which is relatively higher than previously reported values under similar conditions (Table S1). Moreover, when worn on the forehead, a ΔT of 3.3 K produces a voltage output of 3.1 mV. These results highlight the potential of the as-developed TEG for wearable self-powered electronics, confirming its capability for low-grade heat harvesting and its promise in next-generation smart wearable systems.

Experimental section

Materials

The chemical structures of polybutylene furanoate (PBF) and single-walled carbon nanotubes (SWCNTs) used in this study are illustrated in Fig. 1(a). PBF powder was provided by the Industrial Technology Research Institute (ITRI) in Hsinchu, Taiwan. SWCNTs with a purity >95% and a diameter of 1 to 2 nm were obtained from XFNANO. The solvent 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) was purchased from Seedchem, while ethanol (>99.5%) was purchased from ECHO Chemical. The molecular dopant 4-(2,3-dihydro-1,3-dimethyl-1H-benzimidazol-2-yl)-N,N-dimethylbenzenamine (N-DMBI) was supplied by Sigma-Aldrich. All chemicals and reagents were used as received without further purification.

Fabrication of composite fibers

A wet-spinning technique was employed to fabricate PBF/SWCNT composite fibers for thermoelectric applications. Initially, various amounts of PBF powder (4.5 mg, 9 mg, and 13.5 mg) were dissolved in 3 mL of HFIP at 30 °C and stirred for 24 h to ensure complete dissolution. Subsequently, 4.5 mg of SWCNTs was added to each polymer solution, followed by homogenization via ball milling (Restch MM440) at 30 Hz for 20 min to form a uniform spinning solution. The resulting dispersion was loaded into a 5 mL plastic syringe fitted with an 18 G stainless steel needle (inner diameter: 0.96 mm) via a polytetrafluoroethylene (PTFE) tube. The solution was extruded at a rate of 10 mL h⁻¹ into an ethanol coagulation bath, facilitating fiber formation. The as-spun fibers were collected and dried under vacuum at room temperature overnight to remove residual solvent. To optimize the thermoelectric performance, the composite fibers underwent post-treatment under an inert atmosphere (in a glovebox), including no thermal annealing, thermal annealing at 100 °C for 1 h, and annealing at 150 °C for 1 h.

To prepare n-type composite fibers, the electron-donating molecule N-DMBI was employed as a molecular dopant. The 1 : 1 w/o composite fiber sample was immersed in an ethanol solution of N-DMBI (5 mg mL⁻¹) for 15 min. After doping, the fibers were thermally treated at 150 °C for 1 h to complete the doping process and induce n-type behavior.

Thermoelectric measurements of composite fibers

The thermoelectric properties of the composite fibers were evaluated using a commercial ZEM-3 system (ADVANCE RIKO Inc., Japan) under a helium atmosphere at a base temperature of 303 K. Measurements were conducted under applied temperature gradients (ΔT) of 20 K, 30 K, and 40 K. The electrical conductivity (σ) was calculated according to eqn (1):where R is the electrical resistance of the fiber, and l and A are the length and cross section area of the fiber.

Fabrication and measurements of wearable thermoelectric generators (WTEGs)

To demonstrate the practical application of the composite fibers, wearable thermoelectric generators (WTEGs) were fabricated. A flexible nonwoven fabric was used as the substrate. p-Type and n-type composite fibers were alternately woven into the fabric and electrically connected in series using carbon tape and silver paste. For performance evaluation, the cold-side temperature was maintained at approximately 20 °C, while the hot-side temperature was regulated using an electrical heating system. The output voltage and power generation were measured using a Keithley 2182A nanovoltmeter. The WTEGs were applied to the human wrist and forehead for testing. The participant voluntarily participated in this study and signed informed consent before experiments.

Conclusions

Herein, a flexible and efficient wearable thermoelectric generator (WTEG) was successfully developed by using composite fibers of the biomass-derived poly(butylene furanoate) (PBF) and conductive single-walled carbon nanotubes (SWCNTs). These fibers were fabricated via a wet-spinning process with or without thermal post-treatment, and the optimized fibers were subsequently integrated into a nonwoven fabric substrate. Before annealing, the resulting fibers exhibited diameters of 67–70 µm and elliptical cross-sectional areas of approximately 4000 µm². Upon annealing at 100 or 150 °C, the enhanced crystallinity of the PBF and stronger π–π interactions between PBF and the SWCNTs promoted uniform polymer wrapping and a denser arrangement of nanotube bundles. These morphological improvements facilitated more efficient charge carrier transport, thus leading to a substantial increase in the electrical conductivity of the sample containing a 1 : 1 ratio of PBF : SWCNT from 42.9 ± 2.7 S cm⁻¹ before annealing to 113.0 ± 11.6 S cm⁻¹ after annealing at 150 °C. The same annealed sample also achieved the highest power factor of 7.60 ± 0.79 µW m⁻¹ K⁻², along with excellent mechanical flexibility, air stability, and thermal stability, thereby demonstrating its suitability for practical applications. For device integration, n-type fibers were prepared by doping with N-DMBI and then combined with the p-type fibers to construct a WTEG. The 30-junction TEG exhibited a maximum output voltage and power of 29.4 mV and 15.72 nW under a temperature gradient of 30 K. When worn on the human wrist or forehead, the device generated voltages of 1.5 and 3.1 mV, respectively. These results highlight the strong potential of biomass-derived PBF/SWCNT composites for use in flexible thermoelectric devices and self-powered wearable electronics, thereby offering a promising pathway toward sustainable energy harvesting systems.

Author contributions

Yi-Hsuan Huang: conceptualization, methodology, investigation, writing original draft, writing – review and editing. Chih-Wei Hsu: conceptualization, investigation. Cheng-Yuan Lin: conceptualization, investigation. Jhih-Min Lin: resources. Guang-Way Jang: resources. Yung-Pin Huang: project administration, writing – review and editing. Shih-Huang Tung: conceptualization, writing – review and editing. Cheng-Liang Liu: supervision, project administration, writing – review and editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this study's findings are available within the Supplementary information (SI). Supplementary information: characterizations, thermoelectric properties of composite fibers and pristine SWCNTs, and extended results and discussion (DSC thermogram and WAXS pattern of pristine PBF). See DOI: https://doi.org/10.1039/d5ta09343k.

Acknowledgements

C.-L. L. acknowledges financial funding from the NTU-ITRI Nano Center in Taiwan under grant number 114HZE1100M and from the 2030 Cross-Generation Young Scholars Program of the National Science and Technology Council (NSTC) in Taiwan under grant number 114-2628-E-002-005. Besides, this work is supported by the Advanced Research Center for Green Materials Science and Technology from the Featured Area Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (114L9006). The authors also thank TPS 25A at the National Synchrotron Radiation Research Center (NSRRC) in Hsinchu for providing beamtime. The authors gratefully thank Ms Ya-Yun Yang for assistance with SEM experiments at the Instrumentation Center of National Taiwan University, supported by the NSTC.

References

1. Y. Zhang, W. Wang, F. Zhang, K. Dai, C. Li, Y. Fan, G. Chen and Q. Zheng, Small, 2022, 18, 2104922.
2. X.-L. Shi, N.-H. Li, M. Li and Z.-G. Chen, Chem. Rev., 2025, 125, 7525–7724.
3. M. Massetti, F. Jiao, A. J. Ferguson, D. Zhao, K. Wijeratne, A. Würger, J. L. Blackburn, X. Crispin and S. Fabiano, Chem. Rev., 2021, 121, 12465–12547.
4. P. Yadav, N. Dhariwal, A. Sanger, S. B. Kang and V. Kumar, Nano Energy, 2025, 135, 110696.
5. A. Nozariasbmarz, H. Collins, K. Dsouza, M. H. Polash, M. Hosseini, M. Hyland, J. Liu, A. Malhotra, F. M. Ortiz, F. Mohaddes, V. P. Ramesh, Y. Sargolzaeiaval, N. Snouwaert, M. C. Özturk and D. Vashaee, Appl. Energy, 2020, 258, 114069.
6. X. He, X.-L. Shi, X. Wu, C. Li, W.-D. Liu, H. Zhang, X. Yu, L. Wang, X. Qin and Z.-G. Chen, Nat. Commun., 2025, 16, 2523.
7. Q. Zhou, H. Li, C. Du, Z. Ye, L. Liang and G. Chen, Nano Energy, 2023, 118, 109007.
8. T. Wu, X.-L. Shi, W.-D. Liu, M. Li, F. Yue, P. Huang, Q. Liu and Z.-G. Chen, Adv. Fiber Mater., 2024, 6, 607–618.
9. Y.-Y. Deng, X.-L. Shi, T. Wu, H. Wu, Y.-M. Liu, M. Zhu, W.-D. Liu, M. Li, P. Huang, Q. Liu and Z.-G. Chen, Chem. Eng. J., 2024, 502, 158104.
10. T. Sun, B. Zhou, Q. Zheng, L. Wang, W. Jiang and G. J. Snyder, Nat. Commun., 2020, 11, 572.
11. N. Wen, Z. Fan, S. Yang, Y. Zhao, C. Li, T. Cong, H. Huang, J. Zhang, X. Guan and L. Pan, Chem. Eng. J., 2021, 426, 130816.
12. J. Li, X. He, J. Wang, S. Zhu, M. Zhang, C. Wu, G. Dong, R. Liu, L. Wang, L. Chen and K. Cai, ACS Nano, 2025, 19, 11440–11449.
13. L. Zhang, S. Lin, T. Hua, B. Huang, S. Liu and X. Tao, Adv. Energy Mater., 2018, 8, 1700524.
14. M. Craighero, Q. Li, Z. Zeng, C. Choi, Y. Kim, H. Yoon, T. Liu, P. Sowinski, S. Haraguchi, B. Hwang, B. Mihiretie, S. Fabiano and C. Müller, Adv. Sci., 2024, 11, 2406770.
15. H. Li, Y. Liu, S. Liu, P. Li, H. Zhang, C. Zhang and C. He, ACS Appl. Mater. Interfaces, 2024, 16, 17598–17606.
16. F. Yin, X. Luo, X. Wang, Y. Liang, T. Wu, Y. Li and K.-Q. Zhang, Chem. Eng. J., 2024, 498, 155233.
17. A. D. M. Heriyanto, Y. Cho, N. Okamoto, R. Abe, M. Pandey, H. Benten and M. Nakamura, RSC Adv., 2023, 13, 22226–22233.
18. P.-S. Lin, J.-M. Lin, S.-H. Tung, T. Higashihara and C.-L. Liu, Small, 2024, 20, 2306166.
19. Q.-B. Zheng, C.-C. Tseng, M.-H. Lin, J.-M. Lin, S.-H. Tung, Y.-J. Cheng and C.-L. Liu, J. Mater. Chem. C, 2024, 12, 7446–7455.
20. W.-C. Shih, M. Matsuda, K. Konno, P.-S. Lin, T. Higashihara and C.-L. Liu, Composites, Part B, 2024, 286, 111779.
21. K. Nishiyama, Y.-T. Hsiao, W.-N. Wu, J.-M. Lin, S.-H. Tung, C.-L. Liu and T. Higashihara, J. Mater. Chem. A, 2025, 13, 3551–3561.
22. W. Deng, L. Deng, Z. Li, Y. Zhang and G. Chen, ACS Appl. Mater. Interfaces, 2021, 13, 12131–12140.
23. G. Li, C. Chen, Z. Liu, Q. Sun, L. Liang, C. Du and G. Chen, Mater. Horiz., 2024, 11, 1679–1688.
24. W.-T. Ding, X.-Y. Jiao, Y.-M. Zhao, X.-Y. Sun, C. Chen, A.-P. Wu, Y.-T. Ding, P.-X. Hou and C. Liu, ACS Appl. Mater. Interfaces, 2023, 15, 37802–37809.
25. N. M. Stark and L. M. Matuana, Mater. Today Sustain., 2021, 15, 100084.
26. J. H. Sanders, J. Cunniffe, E. Carrejo, C. Burke, A. M. Reynolds, S. C. Dey, M. N. Islam, O. Wagner and D. Argyropoulos, Mater. Today Sustain., 2024, 8, 2400074.
27. M. R. Miah, Y. Dong, J. Wang and J. Zhu, ACS Sustainable Chem. Eng., 2024, 12, 2927–2961.
28. Y.-H. Huang, Y. Chang, C.-J. Huang, J.-M. Lin, S.-H. Tung, G.-W. Jang and C.-L. Liu, Polymer, 2024, 312, 127619.
29. J.-Y. Lam, C.-C. Shih, W.-Y. Lee, C.-C. Chueh, G.-W. Jang, C.-J. Huang, S.-H. Tung and W.-C. Chen, Macromol. Rapid Commun., 2018, 39, 1800271.
30. J.-Y. Lam, G.-W. Jang, C.-J. Huang, S.-H. Tung and W.-C. Chen, ACS Sustainable Chem. Eng., 2020, 8, 5100–5106.
31. Y. Long, R. Zhang, J. Huang, J. Wang, Y. Jiang, G.-h. Hu, J. Yang and J. Zhu, ACS Sustainable Chem. Eng., 2017, 5, 9244–9253.
32. H. He, R. Zhang, P. Zhang, P. Wang, N. Chen, B. Qian, L. Zhang, J. Yu and B. Dai, Adv. Sci., 2023, 10, 2205557.
33. X.-L. Li, R. Zhu and H.-J. Xu, ACS Sustainable Chem. Eng., 2025, 13, 2223–2259.
34. P. Sahu, A. Thorbole and R. K. Gupta, ACS Sustainable Chem. Eng., 2024, 12, 6811–6826.
35. G. Zhou, L. Li, M. Jiang, G. Wang, R. Wang, G. Wu and G. Zhou, Polym. Degrad. Stab., 2021, 189, 109591.
36. D. Jain, F. Cramer, P. Shamraienko, H.-J. Drexler, B. Voit and T. Beweries, RSC Sustainability, 2025, 3, 3513–3519.
37. G. Guidotti, M. Soccio, M. C. García-Gutiérrez, T. Ezquerra, V. Siracusa, E. Gutiérrez-Fernández, A. Munari and N. Lotti, ACS Sustainable Chem. Eng., 2020, 8, 9558–9568.
38. H. Li, Y. Liu, S. Liu, P. Li, C. Zhang and C. He, Composites, Part A, 2023, 166, 107386.
39. C. Xu, S. Yang, P. Li, H. Wang, H. Li and Z. Liu, Compos. Commun., 2022, 32, 101179.
40. J. Zhu, J. Cai, W. Xie, P.-H. Chen, M. Gazzano, M. Scandola and R. A. Gross, Macromolecules, 2013, 46, 796–804.
41. M. C. Righetti, E. Matxinandiarena, A. F. Sousa and A. J. Müller, Polymer, 2025, 335, 128842.
42. N. Sbirrazzuoli, F.-E. Nadir Lahfaidh, A. F. Sousa, G. Molinari and M. C. Righetti, Polymer, 2025, 335, 128845.
43. M. Li, K. Chen, D. Zhang, Z. Ye, Z. Yang, Q. Wang, Z. Jiang, Y. Zhang, Y. Shang and A. Cao, Adv. Sci., 2024, 11, 2404913.
44. H. Altuntas, F. Oke-Altuntas and S. R. P. Silva, J. Electron. Mater., 2024, 53, 2104–2114.
45. E. M. Sullivan, Y. J. Oh, R. A. Gerhardt, B. Wang and K. Kalaitzidou, J. Polym. Res., 2014, 21, 563.

---