Introduction

The rapid advancement of technologies such as artificial intelligence and 5 G has increased the interest in smart wearable devices1,2. However, traditional power supplies, such as lithium cells and nickel-zinc batteries, demand frequent recharging and/or replacement, which limits the long-term service of wearable devices. Flexible thermoelectric generators (f-TEGs) are gaining prominence because of their portability, light weight, eco-friendliness, rapid responsiveness, and high reliability3. f-TEGs can directly convert low-grade waste heat from non-planar sources (e.g., human skin and cylindrical pipes), as well as sunlight, into electrical power4,5. This capability enables them to demonstrate promising applications in energy harvesting, wearable health monitoring, and smart sensors. The efficiency of f-TEGs predominantly depends on the dimensionless figure of merit of TE materials (ZT = S2σT/κ, where S, σ, κ, and T are the Seebeck coefficient, electrical conductivity, thermal conductivity, and absolute temperature, respectively)6,7.

However, a major challenge in f-TEGs for wearable devices remains. Although inorganic TE materials always exhibit excellent TE properties, the lack of flexibility limits their practical application in f-TEGs. An ideal solution involves combining rigid inorganic materials with polymers to boost the flexibility without compromising their TE performance. Conducting polymers, such as poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS)8, have been recognized as promising candidates for flexible TE materials, due to their inherently low κ and tunable σ; while insulating polymers, like polyvinylpyrrolidone (PVP)9, primarily function as flexible low-κ matrices that facilitate the dispersion of inorganic TE materials, thereby showing certain potential as matrices for flexible TE composites. Thus, inorganic/polymer composites for f-TEGs have gained increasing attention, for example, Bi0.5Sb1.5Te3/PEDOT:PSS10, Te/PEDOT:PSS11, and SWCNT/PVP12 composite films.

Ag2Se, as a typical n-type TE semiconductor, has attracted considerable attention due to its environmental friendliness and outstanding performance near room temperature (RT)13,14. For example, Ding et al.15 reported that through a sequential vacuum filtration and hot pressing process, the fabricated flexible Ag2Se/nylon composite film achieved a power factor (PF = S2σ) of 987.4 μWm−1 K−2 at 300 K. Jiang et al.16 and Cai et al.9 subsequently employed a similar approach to fabricate the flexible Ag2Se film and PVP@Ag2Se composite film, yielding a PF of 1882 μWm−1 K−2 and 1910 μWm−1 K−2 at RT, respectively. However, the complex and high-cost equipment needed for the vacuum hot-pressing technology, and the low design flexibility of the vacuum filtration process, pose significant challenges for the large-scale application of flexible TE composite films.

Direct-ink printing can print different types of slurries, which are composed of pure polymers, organic/inorganic composites, or pure inorganic materials, for the construction of the desired shapes layer by layer. The advantages of the direct-ink printing include simple operation, mass production, high design flexibility, and broad applicability17. More importantly, direct-ink printing through layer-by-layer assembly can control the thickness and configuration of printed materials, which is beneficial for enhancing the flexibility and TE performance, and therefore offers a more sustainable and efficient option for developing advanced flexible TE materials. Direct-ink printing can even fabricate f-TEGs directly, when compared with traditional material preparation processes, such as thermal co-evaporation13, vacuum filtration18, screen printing19, and hot pressing process15. However, few studies on flexible Ag2Se/polymer TE composite films have been reported via the direct-ink printing, and the corresponding composite films also exhibited relatively low TE performance, not to mention the f-TEGs directly printed via this technology.

In this study, we developed flexible polymer-coated Ag2Se TE composite films and f-TEGs using direct-ink printing. Figure 1 shows a comprehensive flowchart of the PVP@Ag2Se/MC composite films and prototypical f-TEGs. A critical aspect of this technology is the design and preparation of a slurry with a suitable viscosity. Methylcellulose (MC), a cellulose derivative, has many advantages, e.g., tunable viscosity, good film-forming capability, flexibility, non-toxicity, and affordability20; herein, it was used as a binder to prepare the slurry. A wet chemical method was applied to coat PEDOT:PSS or PVP onto the surfaces of Ag2Se nanowires (NWs) to improve the dispersion ability of Ag2Se NWs in the MC matrix. Subsequently, the direct-ink printing was employed to fabricate flexible TE films, and their TE properties were then optimized through a sequential cold pressing and annealing process. Furthermore, 3-leg f-TEGs were directly printed from the PVP@Ag2Se/MC composite slurry, and their output performance was determined. The sensing and light-heat-electricity conversion capabilities of the f-TEGs were also systematically evaluated, confirming their significant potential for applications in low-grade heat recovery, position identification, light-heat-electricity conversion, respiratory monitoring, etc.

Fig. 1: Conceptual design of TE films and f-TEGs.
figure 1

Schematic diagrams showing the fabrication of a PVP@Ag2Se/MC films with intriguing phase interfaces, lattice defects, etc., and b f-TEGs for multi-scenario applications.

Full size image

Results

Following the wet chemical synthesis and characterizations described in the Methods and Supplementary Information Note S1 and S2, Ag2Se NWs, PEDOT:PSS-coated Ag2Se NWs, and PVP-coated Ag2Se NWs were synthesized. The improved dispersibility of PEDOT:PSS-coated Ag2Se NWs and PVP-coated Ag2Se NWs was shown in Fig. S1. With MC as a binder, slurries of Ag2Se NWs/MC, PEDOT:PSS@Ag2Se NWs/MC, and PVP@Ag2Se NWs/MC were prepared for direct-ink printing of the corresponding films. Photographs in Fig. S2a–c illustrate the typical PVP@Ag2Se/MC slurry (corresponding viscosity versus shear rate was shown in Fig. S3), and printed film before and after the annealing process, respectively. The printed pattern highlights the significant design freedom of the direct-ink printing (Fig. S2d).

Thermoelectric properties and analysis

Figure 2a–c depicted the temperature-dependent behavior of σ, S, and PF for Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films (300−420 K). At 300 K, the σ of Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films was 660.19 S/cm, 668.62 S/cm, and 753.92 S/cm, respectively (Fig. 2a). The σ gradually increased with rising temperature for all the films, and a σ of 1197.98 S/cm at 420 K was achieved for the PVP@Ag2Se/MC film. According to Fig. 2b, all the films exhibited a negative S, confirming their n-type TE characteristics with electrons as the major charge carriers. Notably, the S of Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films was −129.00 μV/K, −141.95 μV/K, and −147.24 μV/K at 300 K, revealing that the introduction of PEDOT:PSS or PVP could improve the absolute value of S. The |S| of all the films remained almost unchanged in the 300–400 K range, whereas a sharp drop was observed above 400 K, mainly ascribed to the β to α phase transition of Ag2Se occurring at approximately 407 K15.

Fig. 2: Evaluation of TE properties and modeling analysis.
figure 2

a σ, b S, and c PF versus temperatures (300–420 K) for the Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films. d nH & μ for the composite films at 300 K. e Pisarenko plots of |S| versus nH for the composite films at 300 K. f Temperature-dependent nH & μ (300–420 K), and g ln(μ(m*T)1/2) versus (kBT)−1 for the PVP@Ag2Se/MC film. h PF versus nH of PVP@Ag2Se/MC film compared with reported Ag2Se based films9,15,16,36,37, and i ZT value of PVP@Ag2Se/MC film compared with reported flexible organic/inorganic composite films prepared via 3D printing43,44,45,46,47 (TPU thermoplastic polyurethane, L-MWCNT long multiwalled carbon nanotube, CMC carboxymethyl cellulose, CB carbon black, BTBA Bi2Te3 based alloy, PLA polylactic acid).

Full size image

In Fig. 2c, the TE performance for all the films showed significant enhancement compared to their corresponding samples before annealing (Fig. S4), where both σ and |S| were substantially improved, thus yielding a significant enhancement in PF. Notably, the PF of PEDOT:PSS@Ag2Se/MC and PVP@Ag2Se/MC films was significantly enhanced compared to that of the Ag2Se/MC film. A PF of 1634.37 μWm−1 K−2 at 300 K was achieved for the PVP@Ag2Se/MC film, and this value increased steadily up to 400 K but sharply declined beyond 400 K due to the abrupt drop in |S | . Consequently, the PF of the PVP@Ag2Se/MC film peaked at 2191.5 μWm−1K−2 with the corresponding S of −140.4 μV/K and σ of 1112.4 S/cm at 400 K, positioning it at the forefront of reported flexible Ag-Se-based TE films9,15,19,21,22,23,24,25,26,27,28,29,30,31,32 (Fig. S5), and marking the highest value so far among flexible organic/inorganic composite TE films fabricated via the direct-ink printing.

To gain a deeper understanding of the obtained TE properties, an extensive investigation of the underlying carrier transport mechanisms was undertaken. Figure 2d showed the carrier concentration (nH) and mobility (μ) based on the Hall effect measurement. The PVP@Ag2Se/MC film exhibited the highest nH when compared with Ag2Se/MC and PEDOT:PSS@Ag2Se/MC films. This might be attributed to the existence of more interstitial Ag27, which ultimately contributed to the highest σ (σ = nHμe, where e is the electron charge) among all the composite films, although its μ was moderately reduced.

Interestingly, the |S| of the PVP@Ag2Se/MC film was higher than those of the other films at 300 K. For semiconductors, the S was typically proportional to the effective mass (m*) of the charge carriers and inversely proportional to (nH)2/3, as described by the Eq. (1):

$$S=\frac{8{\pi }^{2}{{k}_{B}}^{2}}{3e{h}^{2}}{m}^{*}T{\left(\frac{\pi }{3{n}_{H}}\right)}^{2/3}$$
(1)

Here, h, kB, and e denote the Planck’s constant, Boltzmann constant, and electron charge33,34. Figure 2e showed the corresponding Pisarenko plots of |S| versus nH. The simulation based on the single parabolic band (SPB) model was detailed in the Supplementary Information Note S3. When compared with the other films, the higher nH in the PVP@Ag2Se/MC film at 300 K commonly led to a lower |S | , while this effect was offset by the substantial enhancement in m* from 0.17 me to 0.24 me (me is the electron mass), ultimately resulting in a slight increase in |S| at 300 K.

Figure 2f showed the temperature-dependent nH and μ of the PVP@Ag2Se/MC film. The nH and μ were 4.55 × 1018 cm−3 and 1032.40 cm2V−1s−1 at 300 K, respectively. The nH decreased moderately before 340 K and then increased rapidly from 340 K to 420 K. The μ increased quickly before 340 K and then remained stable until 400 K, followed by a slight decline between 400 K and 420 K. Thus, the nH and μ jointly determined the gradually increasing trend of its σ with temperature rising.

Under the assumption of homogeneous grain size and distribution in a crystal, the μ can be given as follows35:

$$\mu={Le}{\left(\frac{1}{2\pi {m}^{*}{k}_{B}T}\right)}^{1/2}exp \left(-\frac{{E}_{B}}{{k}_{B}T}\right)$$
(2)

Here, L denotes the average crystallite size, and the temperature-dependent m* of the PVP@Ag2Se/MC film calculated based on the SPB model was illustrated in Fig. S6. The relationship between ln(μ(m*T)1/2) and (kBT)-1 was shown in Fig. 2g, and notably, the ln(μ(m*T)1/2) showed an obvious turning point above 400 K, which might be ascribed to the β to α phase transition of Ag2Se occurring at approximately 407 K. Compared with other reported Ag2Se-based composite films9,15,16,36,37, the PVP@Ag2Se/MC film in our work exhibited a relatively high PF of 2191.5 μWm−1K−2 at 400 K, with a nH of 5.26 × 1018 cm−3 (Fig. 2h).

The determination of in-plane κ in the polymer-based TE composite films remains challenging, particularly for composite films on flexible substrates38. Herein, both theoretical calculation and experimental measurement were conducted to evaluate the in-plane κ of PVP@Ag2Se/MC film. First, the calculated in-plane κ of 0.36–0.56 Wm−1K−1 for the PVP@Ag2Se/MC film was obtained using the reference-outlined series-parallel connected model39, and the calculation method was also described in the Supplementary Information Note S4. Subsequently, the in-plane κ of ~ 0.52 Wm−1K−1 at 300 K for the PVP@Ag2Se/MC film was measured using the 3ω method (Supplementary Information Note S5 and Fig. S7), and this value aligned well with the above-mentioned theoretical calculations. Such low κ could be attributed to several factors: (1) the pristine Ag2Se has a low κ of ~ 0.6–1.0 Wm−1K−19,40; (2) the PVP and MC, as typical polymers, have inherently low κ41,42; and (3) the pores, interfaces, and defects could effectively scatter the phonons (discussed later). Ultimately, the corresponding ZT of the PVP@Ag2Se/MC film was 0.94 at 300 K, which represented the highest reported value among all the flexible organic/inorganic composite films prepared via 3D printing43,44,45,46,47 (Fig. 2i).

Comprehensive characterizations to understand the obtained high thermoelectric performance

A series of comprehensive characterizations and in-depth analyses were conducted to investigate the factors responsible for the exceptional TE properties of the PVP@Ag2Se/MC film. The XRD patterns revealed that the pure Ag2Se phase existed in the Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films (Fig. S8), while the peaks of PEDOT:PSS, PVP, and MC were not detected, which might be related to the poor crystallinity and/or low content of the polymers20,48. Moreover, when compared with the as-synthesized Ag2Se NWs, PEDOT:PSS@Ag2Se NWs, and PVP@Ag2Se NWs, the intensities of the (112) and (121) peaks of Ag2Se in the Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films significantly decreased, whereas those of (002), (004), (013), and (014) became much stronger. The results suggested that the as-prepared films were highly oriented along (00 l) and (01 l) after the cold pressing and annealing process, which was beneficial for enhancing the TE properties of Ag2Se49.

The SEM images of all nanowires revealed one-dimensional structures, and the morphology of PEDOT:PSS@Ag2Se NWs exhibited some differences compared to those of the other nanowires (Figs. 3a and S9). Typical TEM images (Figs. 3b and S10) indicated that the PVP and PEDOT:PSS were successfully coated on the surfaces of Ag2Se NWs. For instance, a clear interface between the amorphous region (thickness ~ 10 nm, Fig. 3b) and the Ag2Se can be observed in the PVP@Ag2Se NWs (Fig. 3c).

Fig. 3: Electron microscopy analysis to understand the achieved high TE performance.
figure 3

a SEM image and TEM image (the inset) of PVP@Ag2Se NWs. b HRTEM image of PVP@Ag2Se NWs. The insets show the inverse fast Fourier transform image derived from the blue-framed area, and the enlarged green area for the PVP@Ag2Se NWs. c The enlarged red-framed area in (b). d HAADF-STEM image of PVP@Ag2Se/MC film, and e EDS mappings of Ag, Se, C, and N elements. f HRTEM image of PVP@Ag2Se/MC film with a typical heterointerface. g The enlarged orange-framed area in (f), with the inset showing the corresponding SAED patterns. h HRTEM image of PVP@Ag2Se/MC film showing a transition grain boundary. i Inverse fast Fourier transform image obtained from (h), suggesting typical dislocations.

Full size image

Figures S11 and 12 showed SEM images of Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films before and after the annealing process, respectively. After annealing, the morphologies of the three films changed significantly, and the Ag2Se NWs, PEDOT:PSS@Ag2Se NWs, and PVP@Ag2Se NWs underwent sintering to form conductive networks, which facilitated the carrier transport in the composite films. The Ag2Se particles on the surface of the PVP@Ag2Se/MC film caused by sintering were smaller than those of the other composite films, might because PVP acted as a strong capping agent to inhibit the growth of Ag2Se particles9,50. The corresponding EDS mappings of the Ag2Se/MC, PEDOT:PSS@Ag2Se/MC, and PVP@Ag2Se/MC films were shown in Figs. S13–15. The base elements of Ag, Se, C, and O originated from Ag2Se and MC, and the additional S or N elements were ascribed to PEDOT:PSS or PVP, respectively.

To analyze the internal microstructure of the composite films, lamellar TEM samples were prepared using focused ion beam (FIB) technology. Figure 3d showed the high-angle annular dark field scanning TEM (HAADF-STEM) image of PVP@Ag2Se/MC film. Small pores (e.g. white square in Fig. 3d) were observed in PVP@Ag2Se/MC film, likely due to the incomplete decomposition of MC (thermogravimetric analysis as shown in Fig. S16) and sintering of Ag2Se NWs. The pores can further enhance phonon scattering, so as to reduce the κ of the film. Figure 3e displayed the EDS mappings corresponding to the area shown in Fig. 3d, and the distribution of N was consistent with that of Ag and Se, mainly due to the surfaces of Ag2Se NWs being coated with PVP.

High-resolution TEM (HRTEM) images of Ag2Se grains in PVP@Ag2Se/MC film were shown in Fig. 3f, g, which exhibited high crystallinity. A clear heterointerface was observed between Ag2Se and polymer (Fig. 3f). Figure 3g displayed the enlarged orange square in Fig. 3f, and typically the (011) crystal plane of Ag2Se can be identified. The inset of Fig. 3g showed its selected area electron diffraction (SAED) pattern, revealing the zone axis is along the [01-1]. A transition grain boundary (purple dashed line) and a typical dislocation (green dashed circle) were observed in the HRTEM images of the PVP@Ag2Se/MC film in Fig. 3h, i, respectively. Such well-crystallized grains facilitated carrier transport, while transition grain boundaries and typical dislocations were beneficial for scattering phonons.

Overall, the comprehensive TEM analysis suggested that the excellent TE performance of PVP@Ag2Se/MC films arose from several key factors: (1) PVP coating effectively reduced the grain size and introduced more interfaces, favoring the formation of interstitial Ag, thereby significantly increasing the nH and σ; (2) The formed heterointerfaces enhanced carrier scattering, so as to increase the |S | ; and (3) The formed pores, transition grain boundaries, and dislocations were beneficial for scattering phonons and then decreasing the κ.

Evaluation of flexibility

Flexibility is crucial for practical applications, in addition to excellent TE performance. Figure 4a showed the changes of σ and S versus bending cycles for the PVP@Ag2Se/MC film at a r = 4 mm. After 1000 bending cycles, the σ and S remained ~ 93.76% and ~ 96.85% of the corresponding initial values (σ0 and S0), and the σ/σ0 and S/S0 exhibited minimal variation after 1000 bending cycles at larger r values (5.5, 7, 10, and 12 mm) (Fig. 4b). The results suggested that the PVP@Ag2Se/MC film had excellent flexibility and stability, positioning it at the forefront of reported TE films16,18,21,51,52,53,54 (Fig. 4c). By lifting a 100 g weight, the PVP@Ag2Se/MC film maintained good electrical connectivity even when subjected to different folding angles (Fig. 4d). The above demonstration confirmed the excellent flexibility of the PVP@Ag2Se/MC films, primarily ascribed to (1) the inherent flexibility of the organic binder MC, as-coated PVP, and nylon membrane, (2) the porous network inside the film formed by the incomplete decomposition of MC and sintering of Ag2Se NWs, and (3) the tight interfacial bonding between the nylon membrane and PVP@Ag2Se/MC film (as illustrated in Fig. 4e, f).

Fig. 4: Examination of flexibility.
figure 4

a σ/σ0 and S/S0 of PVP@Ag2Se/MC film versus bending cycles at r = 4 mm. b σ/σ0 and S/S0 of the PVP@Ag2Se/MC film after 1000 bending cycles at r = 4, 5.5, 7, 10, and 12 mm. c Flexibility comparison of PVP@Ag2Se/MC film and other reported flexible TE materials16,18,21,51,52,53,54 (CNT carbon nanotube, PPy polypyrrole, PEDOT poly(3,4-ethylenedioxythiophene)). d Digital photographs of the PVP@Ag2Se/MC film lifting a 100 g weight and serving as a flexible conductor with a corresponding circuit diagram. e HAADF-STEM image showing the interfacial bonding between the PVP@Ag2Se/MC film and nylon membrane, and f corresponding EDS mappings of Ag, Se, C, and O elements.

Full size image

3-leg PVP@Ag2Se/MC f-TEG fabricated via direct-ink printing

The traditional f-TEG fabrication processes involve cutting the composite film into TE legs and then assembling them on a substrate, while using the direct-ink printing, a 3-leg f-TEG prototype can be directly printed from the PVP@Ag2Se/MC slurry (see details in Fig. 5a and Supplementary Information Note S1) and the corresponding temperature-dependent TE parameters of the composite film after annealing at 250 °C for 1 h were shown in Fig. S17. Figure 5b displayed the variation of open-circuit voltage (Voc) with respect to the temperature difference (ΔT). The Voc can be calculated based on the Eq. (3):

$${V}_{{{{\rm{oc}}}}}=N\cdot \left|S\right|\cdot \Delta T$$
(3)

Herein, N denotes the number of the TE legs (N = 3) and S equals −145.47 μV/K at 300 K. The calculated Voc of PVP@Ag2Se/MC film and silver paste (the S of silver paste was ~ 1.76 μV/K at 300 K) was 15.75 mV and 0.19 mV at ΔT = 36.1 K, respectively. As a result, the theoretical Voc of the printed f-TEG can be calculated as 15.94 mV (the Voc of PVP@Ag2Se/MC film plus that of silver paste = 15.75 + 0.19 = 15.94 mV), which was slightly higher than the experimental Voc (15.66 mV), mainly attributed to the unavoidable thermal contact resistance and heat dissipation to the environment55.

Fig. 5: Typical printed PVP@Ag2Se/MC f-TEGs with extraordinary TE performance.
figure 5

a Digital photos of printed 3-leg f-TEG before and after the annealing process. b Actual and theoretical results of the relationship between Voc and ΔT. c Current-dependent output voltage and P under different ΔT. d PDmax comparison of the printed PVP@Ag2Se/MC f-TEG with other typical f-TEGs16,19,37,53,56,57,58. e ΔR/R0 versus bending cycles perpendicular to the TE leg length direction, where the r is 12.5 mm. The inset represents the schematic illustration of the bending measurement. f Real-time infrared thermal image of the 3-leg f-TEG (inset showing the f-TEG generated a Voc of 0.70 mV by harvesting the heat dissipated from the human forearm).

Full size image

The output power (P) is calculated using the Eq. (4):

$$P=\left({\frac{{V}_{{{{\rm{oc}}}}}}{{R}_{{{{\rm{ex}}}}}\,+\,{R}_{{{{\rm{in}}}}}}}\right)^{2}\times {R}_{{{{\rm{ex}}}}}$$
(4)

where Rin is the internal resistance of the f-TEG, and Rex is the external resistance. Figure 5c showed the current-dependent output voltage and P at different ΔT. When Rex = Rin, the maximal output power (Pmax) can be obtained, which can be described by Eq. (5):

$${P}_{\max }=\frac{{{V}_{{{{\rm{oc}}}}}}^{2}}{{4R}_{{{{\rm{in}}}}}}$$
(5)

The Pmax of the f-TEG reached 411.56 nW, 944.67 nW, 1735.49 nW, and 2842.57 nW, under the ΔT of 13.3 K, 20.5 K, 27.7 K, and 36.1 K, respectively. Thus, Rin was determined to be ~ 21.3 Ω, which aligned well with the calculated value based on the formula R = NL/(Aσ) = 20.0 Ω, where L is the effective length, and A denotes the cross-sectional area perpendicular to the length of the TE leg, respectively. The Au contact layer and Ag electrode employed for the connections primarily contributed to the low contact resistance of ~ 1.3 Ω for the f-TEG (see the zoomed-in illustration in Fig. S18). The maximal power density (PDmax) is determined by Eq. (6):

$${{PD}}_{\max }=\frac{{P}_{\max }}{{NA}}$$
(6)

An impressive maximum PDmax of 22.1 W/m2 was ultimately obtained when a ΔT of 36.1 K, surpassing that of most reported f-TEGs (Fig. 5d)16,19,37,53,56,57,58.

The resistance change ratio versus bending cycles for the directly printed PVP@Ag2Se/MC f-TEG, with the bending direction perpendicular to the strip length, was demonstrated in Fig. 5e. After 1000 bends at r = 12.5 mm, the resistance change ratio was less than 7%, demonstrating that the f-TEG has excellent flexibility. Moreover, by attaching the two sides of the printed f-TEG to the human forearm and thermal insulator, respectively, a Voc of 0.70 mV was generated at a ΔT of ~ 1.8 K (the corresponding real-time infrared thermal distribution was visualized in Fig. 5f).

Versatile PVP@Ag2Se/MC f-TEG for multi-scenario applications

In addition to traditional low-grade heat recovery, f-TEGs have also been applied to position identification, light-heat-electricity conversion, and respiratory monitoring. The printed f-TEG was integrated into a mask, and the Voc generated under different breathing conditions was captured for respiratory monitoring (Fig. 6a, c displayed the configuration and photograph of the mask). As illustrated in Fig. 6b, the frequency of Voc fluctuations was significantly higher during rapid breathing than during normal breathing. This type of mask, functioning as a breath monitor, is helpful for patients, especially those suffering from asthma, to assess their health condition.

Fig. 6: Multi-scenario applications of f-TEGs.
figure 6

a The specific structure of the printed f-TEG integrated into a mask for respiratory monitoring. b Voc versus time under normal and rapid breathing conditions. c Digital photographs of Voc generated during inhalation or exhalation by the f-TEG integrated into the mask. d The cycling of Voc for the printed s-TEG versus time under different light intensities of 1 sun, 2 suns, and 3 suns. e Voc versus time generated by exposing the printed s-TEG to natural sunlight at 1 pm and 5 pm, and f digital photos of Voc yielded by the printed s-TEG. g V oc versus time for the as-assembled annular s-TEG. h The assembled f-TEG generated a unique Voc when different regions were individually touched by a finger (Fig. S20c), and i digital photo showing a Voc ~ 2.33 mV generated when the region I was touched.

Full size image

Furthermore, with the TiO2 coating, the printed solar-thermoelectric generator (s-TEG, see details in the Supplementary Information Note S1) can convert light into electricity, mainly due to the significant difference in light absorption efficiency for the visible and near-infrared regions between the PVP@Ag2Se/MC film and TiO239, thereby expanding the applications of f-TEG. The cycling of Voc for the s-TEG versus time under different light intensities was shown in Fig. 6d. The Voc rose rapidly and then trended toward stability with the light on, while its value decreased quickly when the light was turned off. Such a sudden drop in Voc when the light was abruptly turned off also indicated the quick response of the s-TEG. The peak value of Voc rose from ~ 3.0 mV to ~ 9.1 mV, as the light intensity increased from 1 to 3 suns.

Additionally, the printed s-TEG was also tested at 1 pm and 5 pm under natural sunlight and shadow, and the Voc variation over time was illustrated in Fig. 6e. Notably, the values of Voc at 1 pm and 5 pm consistently increased as the exposure time from 0 to 30 s under sunlight, and then immediately exhibited a decreasing tendency when under shadows. Furthermore, the Voc recorded at 5 pm was substantially lower than that at 1 pm (Fig. 6f), mainly because of the lower light intensity at 5 pm compared to that at 1 pm, suggesting the good sensitivity of the s-TEG to light intensity.

As shown in Fig. S19a, the annular s-TEG was also assembled. When a sun visor was applied as shown in Fig. S19b, the uncovered part of TE legs in the center became the hot side under light, and the corresponding Voc versus time was displayed in Fig. 6g (inset showed the Voc variation in the first 60 s). The Voc retained ~ 97.0% of its maximum value (~ 13.5 mV) after 30 min, illustrating the outstanding stability and energy conversion efficiency of the annular s-TEG. The device structure was optimized to achieve more distinct position identification (Fig. S19c). When the assembled f-TEG was divided into nine parts and a random part was touched with a finger, a unique Voc was generated, as shown in Figs. 6h and S20. For example, a Voc of ~ 2.33 mV was generated when the region I was touched (Fig. 6i). Such a well-defined Voc distribution demonstrates that the f-TEG has a high sensitivity to the placement of the heat source.

Discussion

This study highlights a remarkable advancement in the development of flexible versatile polymer-coated Ag2Se thermoelectric composite films and devices for multi-scenario applications via a cost-effective direct-ink printing. At 400 K, the PVP@Ag2Se/MC film reached the peak power factor of 2191.5 μWm−1K−2, surpassing all previously reported flexible organic/inorganic composite films fabricated via the direct-ink printing. An effective strategy for improving the TE properties of flexible composite films was developed by introducing heterointerfaces, pores, boundaries, and dislocations through direct-ink printing, cold pressing, and annealing processes. The PVP@Ag2Se/MC film demonstrated excellent flexibility and stability, retaining 93.76% of its initial electrical conductivity after 1000 bends at r = 4 mm. The printed f-TEG achieved a power density of 22.1 W/m2 at a ΔT of 36.1 K, and its resistance change ratio was less than 7% after 1000 bending cycles at r = 12.5 mm. The use of direct-ink printing can enhance the design freedom and manufacturability of f-TEGs, and the printed f-TEGs exhibit huge application potential across multiple scenarios, including low-grade heat recovery, health monitoring, position identification, light-heat-electricity conversion, etc. The direct-ink printing and TE performance optimization strategies for flexible polymer-based films have universal applicability, offering promising solutions for energy harvesting and sensing in wearable electronics.

Methods

Film and device fabrication

Ag2Se NWs, PEDOT:PSS-coated Ag2Se NWs, and PVP-coated Ag2Se NWs were synthesized via wet chemical methods. The corresponding composite films were fabricated on the nylon membranes using the direct-ink printing. Subsequently, a sequential cold pressing and annealing process was performed to optimize the TE properties. Furthermore, three-leg f-TEGs were printed directly to assess the corresponding application potential across multiple scenarios. The experimental details are provided in the Supplementary Information.

Characterizations and performance measurements

The characterization and measurement details are shown in the Supplementary Information.

Theoretical calculation

The single parabolic band (SPB) model and Pisarenko curves were utilized to analyze the carrier transport performance. Meanwhile, a series-parallel connected model was applied to estimate the in-plane κ. Corresponding details are elaborated in the Supplementary Information.