Introduction

Solar-thermoelectric generators (STEGs) can efficiently convert the heat generated by sunlight into electricity, making them highly promising for applications in solar power generation, photodetection, and self-powered wearable electronics. Conventional STEGs are typically composed of thermoelectric (TE) modules integrated with photothermal (PT) conversion systems1. Despite extensive research efforts over several decades, the record solar-to-electricity conversion efficiency of STEGs based on bulk thermoelectric materials remains at ~9.6%2,3. In comparison, a significant gap still exists between STEGs and other emerging solar-energy conversion technologies, such as thermophotovoltaic (TPV) cells, which can achieve efficiencies exceeding 35% owing to the high operational temperature of thermal emitters4,5 and the utilization of highly efficient PV cells6,7,8. Consequently, STEGs still have a considerable distance to go before they can be considered viable for large-scale power generation applications. However, STEGs based on flexible thermoelectric materials offer distinct advantages, including facile fabrication, low costs, and adaptability to curved surfaces. Crucially, flexible STEGs prioritize consistent energy output rather than peak efficiency, which is a trade-off well suited to the requirements of low-power electronics and sensor networks. Such attributes render them highly suitable for wearable applications, where energy reliability and mechanical adaptability are essential.

According to the configuration, STEGs can be categorized into discrete and integrated types9. Discrete STEGs typically require an additional PT layer, such as a metal-nanoparticle coating or a metal-ceramic multilayer selective absorber,3,10 to achieve photothermal conversion. However, the insulating barrier layer located between the PT layer and TE materials introduces thermal loss, limiting the achievable temperature gradient. Additionally, the inclusion of a multilayered PT layer increases the complexity of the device fabrication process. In contrast, integrated STEGs utilize the TE materials as the light-absorbing terminal, thereby reducing the interface and simplifying the fabrication process. Nevertheless, this integrated strategy requires TE materials with strong intrinsic light absorption. Materials such as carbon nanotubes11,12 and organic semiconductors (e.g., PEDOT)13,14,15 exhibit broadband absorption and efficient light-to-heat conversion, but their relatively low TE performance restricts overall device output and makes simultaneous optimization of PT and TE properties challenging.

Wearable applications require higher room-temperature (RT) TE performance. Although n-type Bi2Te2.7Se0.3, p-type Bi0.5Sb1.5Te316,17. and emerging Mg3Bi2-based materials18,19. With superior TE performances at RT can be considered as potential candidates, Ag2Se is particularly well-suited for flexible and wearable thermoelectric applications due to its balanced combination of high TE performance at RT20,21, mechanical compliance22, and environmental stability. In recent years, flexible Ag2Se films have been fabricated using various processing routes, including magnetron sputtering23, vacuum filtration24, direct selenization25, and printing techniques26,27. Among these, magnetron sputtering is particularly advantageous, as it allows for precise control of film thickness, ensures excellent layer uniformity, and facilitates good reproducibility with large-area fabrication, compatible with existing industrial production lines.

A variety of high-performance Ag2Se-based flexible TE devices have been extensively investigated24,26,28,29, demonstrating their great potential for efficient energy conversion and wearable power generation. By integrating various PT conversion layers such as carbon black film30, TiO2 coating27,31, and polypyrrole (PPy)32, Ag2Se flexible TE devices can effectively achieve photo-thermal-electrical conversion for photothermal power generation and light detection. In addition, the narrow bandgap (~0.15 eV) endows Ag2Se film with broad-spectrum light absorption characteristics33, making it a promising candidate for integrated STEGs. Recent studies have demonstrated the feasibility of Ag2Se as a photothermal terminal in flexible STEGs9,34,35,36. Specifically, thermoelectric devices based on Ag2Se/Nylon composite films, without requiring additional PT conversion layers or heat sink components, achieved an open-circuit voltage (Uoc) of 8.8 mV under 1 sun irradiation9. However, the achieved Uoc remains modest relative to that of most flexible STEGs (Supplementary Table 1). The absence of strategies to further improve the photothermal performance of the Ag2Se-based films has constrained the overall output performance of the corresponding STEGs.

In this study, we designed a multilayer PT conversion terminal to realize an integrated STEG. As shown in Fig. 1a, the PT terminal directly employs Ag2Se film as the light-absorbing layer, utilizing its narrow bandgap to achieve broadband solar absorption. Photothermal conversion is accomplished through non-radiative recombination processes. By integrating an infrared-reflective metallic layer at the bottom and a dual anti-reflection coating on the surface, the structure enables high solar absorption while suppressing infrared emission, thereby maintaining an elevated operating temperature. The flexible STEG using this multilayer design as the photothermal terminal exhibited a maximum temperature difference of 19.6 K. The output power density reached 0.17 μW cm-2 under a light irradiance of 100 mW cm-2 (Fig. 1b), which is superior to recently reported flexible photo-thermoelectric devices11,30,37,38,39 (Fig. 1c). More importantly, this strategy is generalizable to other narrow-bandgap thermoelectric thin films, providing a versatile approach to extend the applicability of thermoelectric materials.

Fig. 1: Overview of the design and performance of the integrated STEG.
Fig. 1: Overview of the design and performance of the integrated STEG.The alternative text for this image may have been generated using AI.
Full size image

a Schematic of the ring-shaped STEG integrated with a multilayer photothermal structure. The right panel shows the working mechanism of the structured design, highlighting the broadband absorption enabled by narrow-bandgap materials, the non-radiative recombination-driven photothermal conversion mechanism, and the selective absorption characteristics of the multilayer structure. b Schematic diagram of the actual measurement condition of the STEG. c Comparison of output power density of the all-in-one STEG in this work and others, such as CNT11, printed Ag2Se film30 and TE fabrics37,38,39 based STEGs.

Results and Discussion

In this work, Ag2Se films were deposited on flexible polyimide (PI) substrate using magnetron sputtering, and the TE performance was optimized by adjusting the stoichiometry of the target (Supplementary Figs. 1, 2). The specific experimental scheme is shown in the Supplementary Note 1. As the Se content in the target increases, as shown in Fig. 2a, the carrier concentration (nH) decreases from 6.3 × 1018 cm−3 for Ag2Se film to 4.8 × 1018 cm−3 for Ag2Se1.02 film (with 2 at.% excess Se in the sputtering target). In contrast, the Hall mobility (μH) of the film is significantly enhanced. Notably, the Ag2Se1.02 film exhibits a room-temperature μH as high as 1414 cm2 V−1 s−1, surpassing those reported for most Ag2Se-based flexible thermoelectric films24,29,40. Such high μH primarily arises from the high orientation and large grain size (Supplementary Fig. 3), which lowers the grain-boundary barrier height and grain-boundary density (Supplementary Fig. 4), thereby weakening carrier scattering41,42,43. Despite the decrease in nH, the room-temperature (~300 K) electrical conductivity (σ) of the film increases with Se content due to the enhanced μH (Fig. 2b). The Seebeck coefficient (S) increased with increasing Se content from −132 μV K−1 to −152 μV K−1 (Fig. 2c). As shown in Fig. 2d, the effective mass of Ag2Se1+x films remains nearly constant at 0.28 mₑ, suggesting that the added Se has little influence on the band structure. The observed increase in S thus primarily originates from the reduced nH. It should be noted that when the Se content was excessively high, as in the Ag2Se1.03 sample, the nH shows an upward trend, while both the preferred orientation factor (Supplementary Fig. 3b) and in-plane mobility (Fig. 2a) decrease significantly. This degradation in film quality is likely due to the precipitation of elemental Se. Finally, when 2 at % of additional Se was added to the target, the room-temperature power factor (PF) of the film reached 25.7 μW cm−1 K−2 (Fig. 2e), comparable to state-of-the-art flexible Ag2Se films (Supplementary Table 2)23,24,25,26,28,29,40,42,44,45,46,47. Moreover, the film exhibits a resistance change of less than 2.5% after 1000 bending cycles at a bending strain of 1% (Fig. 2f), demonstrating excellent mechanical flexibility (experimental details are provided in the Supplementary Note 2 and Figs. 5−7)

Fig. 2: Thermoelectric and mechanical properties of Ag2Se thin films.
Fig. 2: Thermoelectric and mechanical properties of Ag2Se thin films.The alternative text for this image may have been generated using AI.
Full size image

a Measured carrier concentration (nH) and mobility (μH) at room-temperature. b Seebeck coefficient, and c electrical conductivity as a function of temperature. d Room-temperature Pisarenko relations of Ag2Se1+x (x = 0, 0.1, 0.2, and 0.3) thin films. e Power factor as a function of temperature. f Normalized resistance of the Ag2Se1.02 films as a function of bending cycles. The inset shows the bending process.

The photothermal performance of the pristine Ag2Se (hereafter referring to the TE performance optimized Ag2Se1.02 film) flexible film was evaluated by monitoring its temperature rise under Xe lamp illumination. Details of the measurement are provided in the Materials and Methods section. Upon light illumination, the photons with energies higher than the bandgap generate above-bandgap electron-hole pairs in the Ag2Se film. The excited carriers subsequently relax to the band edges, releasing the excess energy as heat through thermalization14,48. The temperature of the pristine film (including PI substrate) rose rapidly as soon as the Xe lamp was turned on, and stabilized at 62 °C after 70 s under a light irradiance of 100 mW cm−2 (Supplementary Fig. 8a). Moreover, the maximum stable temperature (Tmax) exhibited thickness-independent behavior (Supplementary Fig. 8b). In addition to Ag2Se, similar Tmax values (60–70 °C) were also observed in other flexible narrow-bandgap thermoelectric films, such as Bi2Te3, Sb2Te3, and Cu2Se (Supplementary Fig. 9). However, the achieved Tmax remains lower than that of blackbody-like PT films12,49. To further enhance the photothermal conversion efficiency of the Ag2Se film, we adopted a multilayer design inspired by selective-absorption coatings.

As shown in Fig. 3a, the overall structure of the multilayer designed Ag2Se (MD-Ag2Se) film consists of a metallic base layer, an intermediate absorption layer, and a surface antireflection layer. Both Ag and W are well-known infrared thermal radiation barrier materials50,51,52. At the bottom of the structure, they serve as infrared reflectors, effectively reducing the infrared radiation of the structured film. Meanwhile, the reflective property of the metal layer enhances the absorption and utilization of infrared light by the Ag2Se absorption layer, thereby improving the photothermal conversion efficiency. Moreover, due to its high thermal stability, the W layer also functions as an effective diffusion barrier, preventing interdiffusion between Ag and Ag2Se during the annealing process. To further enhance overall absorption, a double-layer antireflection coating was fabricated using low-refractive-index Al2O3 and SiO2 thin films. The gradient refractive index induced by this bilayer structure promotes destructive interference of reflected light, thereby suppressing surface reflection and enhancing the overall light absorption of the film53.

Fig. 3: Schematic and microstructure of MD-Ag2Se film.
Fig. 3: Schematic and microstructure of MD-Ag2Se film.The alternative text for this image may have been generated using AI.
Full size image

a Schematic of the MD-Ag2Se thin-film structure. b TEM cross-sectional view of the MD-Ag2Se film. c Enlarged view of region 1 in b, showing the thickness of the double anti-reflection layer. d High-resolution image and selected-area electron diffraction pattern of region 2 in b, where the zone axis is identified as [100]. e SAED pattern of region 1 in b, where the zone axis is determined to be [010]. f SAED pattern of the Ag layer. g EDS-mapping image of the entire cross-section. h Distribution of Si, Al, Ag, W, and Se elements. Scale bar is 500 nm.

The microstructure of MD-Ag2Se thin films was characterized in detail by transmission electron microscopy (TEM). From the TEM images in Fig. 3b, the interfaces between the layers are sharp, and the film has uniform thickness. The magnified view of region 1 in Fig. 3b reveals that the dual antireflection layers are amorphous, with the thicknesses of Al2O3 and SiO2 measuring ~20 nm and ~35 nm, respectively. Beneath the antireflection layer, the Ag2Se film exhibits high crystallinity. The HRTEM image of region 2 in Fig. 3b reveals a lattice spacing of 0.78 nm, characteristic of the orthorhombic β-Ag2Se (001) plane (Fig. 3d), consistent with the XRD results. Selected-area electron diffraction (SAED) patterns obtained from different regions of the Ag2Se layer (Fig. 3d-e) exhibit sharp, well-defined diffraction spots, confirming the film’s uniformity and high crystallinity, which are key factors contributing to its high in-plane carrier mobility. The SAED image of the bottom metallic layer (Fig. 3f) is indexed to cubic Ag. EDS mapping illustrates the elemental distribution across the multilayer structure (Fig. 3g, h), where the Ag signal intensity changes across the W layer, reflecting the presence of Ag2Se above and the Ag film below. The line-scan profiles exhibit sharp transitions at the interface (Supplementary Fig. 10), demonstrating the absence of elemental diffusion between the two layers and attesting to the effectiveness of the W layer as a diffusion barrier.

The optical images reveal that the MD-Ag2Se film exhibits a notably darker surface appearance compared to the intrinsic film (Supplementary Fig. 11). This distinction corresponds to a substantial increase in average absorptance of the structured film in the UV-Vis-NIR range (250-2500 nm), rising from 63.6% to 84.3% (Fig. 4a). The MD-Ag2Se film also exhibits high reflectance in the infrared range of 2.5-18 μm (Supplementary Fig. 12), indicating low heat loss. Simulations of electric field distribution (Fig. 4b) reveal that the multilayer design enhances absorption across the solar spectrum, verifying its pronounced anti-reflection behavior. Under Xe-lamp irradiation at 1 sun, the Tmax reaches 85 °C, representing a 1.37-fold enhancement relative to the pristine Ag2Se film (Supplementary Fig. 8a). As the Xe lamp intensity increases from 0.25 to 2.0 suns, the Tmax sharply increases from 47.2 °C to 145.7 °C (Fig. 4c), surpassing many photothermal materials such as carbon-based materials, conjugated polymers, MXene 2D materials, etc. (Supplementary Table 3)12,14,37,38,49,54,55,56,57,58,59, and approaching the performance of commercial selective solar absorber films60. Moreover, the Tmax demonstrates a high degree of linearity with the light intensity, indicating its potential for use in light intensity detection (inset of Fig. 4c).

Fig. 4: Photothermal properties of the MD-Ag2Se film.
Fig. 4: Photothermal properties of the MD-Ag2Se film.The alternative text for this image may have been generated using AI.
Full size image

a UV-Vis-NIR absorption spectra of the pristine and MD-Ag2Se films. b Simulated normalized electric field distribution of the MD-Ag2Se film. c Temperature-time curves of the MD-Ag2Se film under different light intensities; the inset illustrates the relationship between Tmax and light intensity. d Photothermal conversion efficiency η of the pristine and MD-Ag2Se films. e Temperature-time cycling behavior under repeated light on/off cycles. f Infrared thermal images of the MD-Ag2Se film patterned into an “HBU” shape under illumination at different times. g Temperature–time curves of the film under outdoor illumination. h Dependence of Tmax on the number of bending cycles.

The photothermal conversion efficiency (η) of the films was evaluated based on the following equation48,61,62:

$$\eta=\frac{hs\,({T_{\max }}-{T_{{{\rm{surr}}}}})}{I{A}_{\lambda }}$$
(1)

Where h is the heat-transfer coefficient, s is the surface area of the film, Tsurr is the ambient temperature, Tmax is the maximum equilibrium temperature of the film, I is the solar irradiance, Aλ is the absorptance of the PT film (details of the calculation are given in Supplementary Note 3 and Fig. 13). As shown in Fig. 4d, the η of the flexible MD-Ag2Se film was calculated to be 87.6%, more than twice of the pristine film (~31.2%). Figure 4e presents the temperature variations under light illumination (“on” and “off” states), indicating excellent photothermal stability and repeatability of the MD-Ag2Se film. Notably, the film reaches its peak temperature within 70 s, significantly faster than those reported for state-of-the-art solar thermal conversion materials49,56,58,63,64. This rapid response is further demonstrated in infrared thermal images of a structured Ag2Se film patterned into “HBU” under 1 sun irradiation (Fig. 4f).

The photothermal response of the structured Ag2Se film was continuously monitored in outdoor conditions in Baoding (China) on 7 May 2025 under cloudless and sunny conditions, with temperatures ranging from 15 to 28 °C, moderate humidity (45%), and low wind speed (≤ 2.5 m s−1). The experimental apparatus, schematically illustrated in Supplementary Fig. 14, mainly comprises a polystyrene box coated with aluminum foil to block external radiation. Figure 4g shows that the temperature variation of the samples consistently follows the solar irradiance. The Tmax of the MD-Ag2Se film reached 93 °C during the peak solar irradiance period (11:30 a.m.–1:30 p.m.), in contrast to 67 °C for the pristine film. The outdoor test results were superior to those obtained under Xe lamp illumination, likely due to the larger illuminated area that heated the entire test setup. Similarly, cyclic bending tests were conducted to evaluate the mechanical properties of the MD-Ag2Se film. As shown in Fig. 4h, after 10,000 bending cycles with a radius of 3.75 mm (1% strain), the photothermal performance (as indicated by the Tmax) exhibited negligible change, indicating excellent flexibility (Supplementary Fig. 15). In addition, after 1000 cycles of bending, the relative resistance change of the multilayer film was less than 1% (Supplementary Fig. 16). Such remarkable flexibility renders the film highly promising for applications in the field of wearable electronics.

To further validate the generality of this multilayer design strategy, similar structured modifications were applied to other representative narrow-bandgap thermoelectric thin films, such as Bi2Te3, Sb2Te3, and Cu2Se. Upon structural optimization, all films exhibited a significantly enhanced Tmax that increased linearly with light intensity (Supplementary Fig. 17). Under Xe lamp illumination, the Tmax was typically 20–30 °C higher than that of the pristine films (Supplementary Fig. 9). Notably, the Tmax of the structured Cu2Se film reached 99 °C under 1 sun and further increased to 160 °C when the light intensity reached 2 suns. These results demonstrate the general applicability of our strategy in enhancing the photothermal performance of narrow-bandgap thermoelectric thin films.

To systematically verify the universality of this design, we fabricated a flexible ring-shaped STEG containing nine pairs n-type Ag2Se and p-type Sb2Te3 TE legs. The TE performance of the Sb2Te3 film is shown in Supplementary Fig. 18. Prior to device assembly, finite element simulations were performed to optimize the overall device configuration. Owing to the disparity in TE properties between the n- and p-type legs, the output performance of the device is highly sensitive to the relative cross-section dimension65. As shown in Fig. 5a, a thickness ratio of 1.2 between the n-type and p-type legs yielded the highest output power. Further simulations examined the effect of the ratio (r) between the photothermal terminal length (l1) and the thermoelectric leg length (l2). As shown in Fig. 5b, increasing the proportion of the photothermal terminal caused a gradual decrease in the Uoc. This reduction occurs because expansion of the photothermal region shortens the non-illuminated segment, while the heat dissipation capacity of the cold end remains unchanged, leading to a smaller steady-state ΔT. The Pmax of the device is governed by the Uoc and the internal resistance (Rin) of the sector-shaped TE leg as follows:

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

The Uoc is associated with the ΔT, demonstrating a decreasing trend as the length ratio r increases (Supplementary Fig. 19a). The Rin of the sector-shaped TE leg can be calculated as follows:

$${R}_{{{\rm{in}}}}=\frac{\rho }{d\theta }{{\mathrm{ln}}}\frac{{r}_{2}}{{r}_{1}}=\frac{\rho }{d\theta }{{\mathrm{ln}}}\frac{{l}_{2}+{r}_{1}}{{r}_{1}}$$
(3)

where ρ is the electrical resistivity, d is the thickness of the TE films, θ, r1, and r2 denote the sector angle, inner radius, and outer radius of the sector-shaped TE leg, respectively (Supplementary Fig. 19b). The increasing r decreases l2, thereby reducing the Rin. Hence, both ΔT and Rin are functions of the leg-length ratio r. Consequently, Pmax initially increases with r and then decreases, reaching a maximum at r = 0.28. The final geometric configuration and the optical graph of the device were shown in Fig. 5c and Supplementary Fig. 20, and fabrication details were provided in the Materials and Methods section and Supplementary Fig. 21.

Fig. 5: Output performance of the ring-shaped STEG.
Fig. 5: Output performance of the ring-shaped STEG.The alternative text for this image may have been generated using AI.
Full size image

a The dependence of the Pmax on the ratio of the cross-sectional areas of the n- and p-type thermoelectric legs (An/Ap). b The dependence of Uoc and Pmax on the ratio (r) of the photothermal terminal length (l1) to the thermoelectric leg length (l2). c Schematic diagram of the geometric structure of the ring-shaped STEG. d Uoc-time response under different light intensities. e Uoc-time cycling behavior under repeated on/off illumination cycles. f I-U and I-P characteristics under varying illumination intensities. g Outdoor output performance of the ring-shaped STEG. h Output performance of the ring-shaped STEG integrated with external equipment under outdoor conditions.

Device performance was evaluated under simulated sunlight. During the test, only the central photothermal terminal was illuminated (Supplementary Fig. 22). Upon irradiation at 1 sun, the Uoc rose rapidly, reaching 49 mV within 60 s (Fig. 5d), in good agreement with the simulation results (Supplementary Fig. 23). The characteristic rise and fall times, corresponding to 10% and 90% amplitude, were 18.2 s and 24.6 s, respectively (Supplementary Fig. 24). When the light intensity increased to 1.5 and 2 suns, Uoc further increased to 66 mV and 79 mV, respectively. Under1 sun irradiation, the measured ΔT of 19.6 K was consistent with theoretical predictions based on the Seebeck coefficient and temperature difference (Supplementary Fig. 25). The STEG demonstrated stable behavior, nearly identical voltage responses, over five consecutive on/off light cycles under 1 sun irradiation (Fig. 5e). Figure 5f presents the I-U and I-P characteristics of the STEG under Xe lamp irradiation with varying light intensities from 1 sun to 2 suns, where Pmax increased proportionally with light intensity. The corresponding maximum output powers were 1.2, 2.1, and 3.0 μW under irradiation of 1, 1.5, and 2 suns, respectively. The maximum output power density (PDmax) was then calculated by dividing Pmax by the total surface area of the device. As shown in Fig. 1c, the PDmax reached 0.17 μW cm−2 under 1 sun irradiation and further increased to 0.42 μW cm−2 when the light intensity rose to 2 suns, surpassing most previously reported flexible photo-thermoelectric devices, such as fiber-based STEGs37,38,39, printed Ag2Se film-based STEG30, and carbon nanotube-based STEGs11.

A full-day outdoor test was carried out on 10 May 2025 in Baoding, China, from 8:00 to 18:00 to evaluate the real-world output performance of the ring-shaped STEG under natural sunlight. The experimental setup and weather conditions are presented in Supplementary Fig. 26. Notably, the device achieved a peak output power of approximately 1 μW at noon with the maximal solar irradiance (Fig. 5g). Further experiments revealed that the ΔT varied synchronously with the solar irradiance, resulting in a change in the Uoc and the Pmax (Supplementary Fig. 27-28).

To demonstrate integration into wearables, the ring-shaped STEG was mounted on a sun hat and a backpack, and its output performance was tested at noon on 21 September 2025 in Baoding, with an ambient temperature of ~24 °C and a wind speed of ≤0.5 m s−1. A solar radiometer and a multimeter were used to measure the solar irradiance and Uoc, respectively. As shown in Fig. 5h, the STEG delivered a Uoc of 14.5 mV under a sunlight irradiance of 525 W m−2 when attached to the sun hat. When it was further sewn onto a backpack, its Uoc increased to 23.7 mV under a light irradiance of 657 W m−2. Further studies indicated that both ΔT and Uoc exhibited small fluctuations under continuous wind disturbance. However, when the wind speed was low, a relatively high Uoc was maintained, whereas higher wind speeds led to a reduction in Uoc (Supplementary Fig. 29). The variations of ΔT and Uoc follow an identical trend. Nevertheless, such environmental perturbations can be effectively mitigated in future practical applications through improved system-level packaging and protection strategies.

These results demonstrate that the integrated STEG architecture proposed in this work offers significant advantages over conventional discrete devices and systems employing organic photothermal films. The structure enables strong coupling between PT and TE conversion processes within a unified platform, thereby improving heat utilization efficiency. Notably, in comparison to common flexible PT materials such as CNT11,12 and PEDOT13 films, the utilization of Ag2Se film results in significantly improved PF of 25.7 μW cm−1 K−2. In addition, the entire device fabrication relies solely on magnetron sputtering, which is highly compatible with MEMS fabrication and photolithography techniques, offering a scalable route toward wafer-scale production of STEGs. The integrated device delivers excellent output performance under outdoor conditions, enabling it to power wearable health-monitoring sensors (e.g., pulse or sweat-rate sensors). Moreover, its linear photothermal response and light-induced Uoc characteristics indicate substantial potential for applications in real-time outdoor light intensity sensing.

In summary, we presented a design strategy for integrated flexible solar-thermoelectric devices that combines the strong broadband absorption of narrow-bandgap thermoelectric thin films with a multilayer selective-absorption architecture. The multilayer-designed Ag2Se flexible films exhibited temperature rises exceeding 85 °C under both Xe lamp and outdoor sunlight illumination. Correspondingly, the photothermal conversion efficiency was enhanced from 31.2% for the pristine film to 87.6% for the multilayer-designed film. The as-prepared ring-shaped Ag2Se-based STEG fabricated using this strategy generated a large ΔT of 19.6 K, and delivered a maximum output power density of 0.17 μW cm−2 under 1 sun irradiation, outperforming most reported flexible STEGs. Moreover, it achieved an output power of ~1 μW under outdoor conditions and generated an open-circuit voltage of 14.5–23.7 mV when integrated with external equipment. These findings demonstrate the broad applicability of our strategy and highlight its potential to accelerate the practical deployment of high-performance, flexible STEGs for energy harvesting.

Methods

Fabrication of Se excess Ag2Se targets

Polycrystalline Ag2Se ingots were synthesized by sealing stoichiometric amounts of high-purity (≥99.99%) Ag and (≥99.99%) Se chunks in an evacuated quartz ampoule. The ampoule was gradually heated to 1323 K over a period of more than 10 hours and maintained at this temperature for 12 hours to ensure complete melting and homogenization. Subsequently, the melt was allowed to cool naturally within the furnace by shutting off the power supply. The resulting ingots were then cut into small pieces (less than 3 mm3). Subsequently, Se particles with specific excess atomic ratios of 1 at%, 2 at%, and 3 at% were introduced, and the mixture was ground into fine powder under ambient conditions using a pulverizer. The powder was subsequently consolidated into a dense sputtering target with a diameter of 50.8 mm and a thickness of 4 mm by spark plasma sintering (SPS) at 573 K under a uniaxial pressure of 30 MPa.

Deposition of flexible Ag2Se films

Ag2Se1+x (x = 0, 0.01, 0.02, 0.03) targets were connected to a direct current (DC) power source for sputtering. Prior to deposition, PI substrates with a thickness of 75 μm were sequentially cleaned using ultrasonic baths in acetone and ethanol for 20 minutes each. During film deposition, the Ag2Se targets were sputtered at a power of 100 W, and the Ar pressure was fixed at 0.4 Pa. The resulting films were designated Ag2Se, Ag2Se1.01, Ag2Se1.02, and Ag2Se1.03, corresponding to the compositions of the sputtering targets. To improve crystallinity, all deposited films were subsequently annealed in a nitrogen atmosphere at 448 K for 30 minutes.

Characterization

The Seebeck coefficient (S) and electrical conductivity (σ) were simultaneously measured by a commercial system (ZEM-3, Advance-Riko). The measurement uncertainty was 3% for electrical conductivity and 5% for the Seebeck coefficient. Hall coefficients (RH) were measured using the van der Pauw method with a four-probe configuration under a reversible magnetic field of 1.5 T and an electrical current of 10 mA. The Hall carrier concentration (nH) was calculated via nH = 1/(eRH). The absorption spectra were measured using a UV–Vis–NIR spectrophotometer (Agilent, Cary 5000). The cross-sectional morphology of the MD-Ag2Se film was characterized using a transmission electron microscope (JEOL ARM 200 F.).

Device fabrication

The fabrication process of the ring-shaped STEG, illustrated in Supplementary Fig. 21, involved several critical steps. First, the inner electrode, composed of an Ag/W bilayer, was deposited onto the substrate via magnetron sputtering. Subsequently, p-type Sb2Te3 legs were deposited and annealed under vacuum at 573 K for 1 h to improve crystallinity. Next, the n-type Ag2Se film was deposited and annealed at 448 K for 30 min. An Al2O3/SiO2 double-layer anti-reflection coating was then deposited on the central area. Finally, an Ag film was deposited as the outer electrode to connect the n- and p-type thermoelectric legs in series. To further enhance the effective temperature difference, the substrate regions not covered by the thermoelectric materials and electrodes were selectively removed. An optimized co-sputtering process was employed to prepare the Sb2Te3 thin films. The fabrication parameters for each layer and thermoelectric leg are provided in Supplementary Table 4.