Polyphenol‐Mediated Liquid Metal Composite Architecture for Solar Thermoelectric Generation

The development of advanced solar energy technologies, which efficiently convert solar energy to heat and then to electricity, remains a significant challenge in the pursuit of clean energy production. Here, this challenge is addressed by designing a photothermal absorber composed of liquid gallium particles and a natural polyphenol‐based coordination ink. The design of this composite takes advantage of the tuneable light absorption properties of the polyphenol inks and can also be applied onto flexible substrates. While the ink utilizes two types of coordination complexes to absorb light at different wavelengths, the liquid gallium particles with high thermal and electrical properties provide enhanced thermoelectric effect. As such, the photothermal composite exhibits a broad‐spectrum light absorption and highly efficient solar‐to‐heat conversion. A thermoelectric generator coated with the photothermal composite exhibits an impressive voltage output of ≈185.3 mV when exposed to 1 Sun illumination, without requiring any optical concentration, which sets a new record for a power density at 345.5 µW cm−2. This work showcases the synergistic combination of natural compound‐based light‐absorbing coordination complexes with liquid metals to achieve a strong photothermal effect and their integration into thermoelectric devices with powerful light harvesting capabilities.


Introduction
Solar thermoelectric generators (STEGs) present a promising pathway to harnessing solar energy. [1]With their ability to generate electricity from direct and diffuse sunlight and their potential for waste heat recovery from the infrared region of the long-term reliability and durability. [23]Additionally, LMs can capture waste heat and convert it into electricity. [22]For example, Yang et al. [12] reported a flexible 3D-structured absorbers based on LM, polyvinyl alcohol, polydopamine, and reduced graphene oxide (PVA/LMP-rGO).The STEG device was constructed via a controllable casting molding strategy featuring a micrograting absorber architecture.The resulting system display an attractive blend of characteristics, including robust broadband selective light absorption, excellent photothermal conversion efficiency, high heat flux density, and intriguing mechanical properties.However, the narrow light absorption range of LMs limits their direct application in the field of solar energy harvesting and typically requires their integration with different photoactive materials.Altogether, exploring such composite systems has just started to flourish and requires ongoing research to optimize their performance.
[37][38] For example, TA/iron (Fe) complexes are deep blue colored and absorb light in the range of 600-700 nm. [38]However, despite the progress in polyphenol-based functional systems, utilization of their light absorption property for solar energy harvesting has remained relatively unexplored. [39]or this study, we hypothesize that the combination of LMs and polyphenol coordination complexes can provide a general platform to integrate the light-absorbing properties of the phenolic complexes with the favorable thermal and electrical properties of LMs in a synergistic way to develop highly efficient and flexible STEGs.As such, we propose a composite system based on liquid Ga particles and a coordination-based polyphenol ink prepared via mixing TA ,Fe 3+ , and titanium (Ti 4+ ) ions.The resulting metal-phenolic ink (MPI) shows broad-spectrum light absorbance in the visible range due to the presence of two types of coordination complexes (TA-Fe 3+ and TA-Ti 4+ ).Combining this broad-spectrum absorber ink with LM particles, a highly efficient STEG device is fabricated.Unlike other LM-based composites for STEG applications, our approach is rapid, straightforward, and does not require any high-tech instrumentation.More importantly, the developed STEG device displays a record power density of ≈345.5 μW cm −2 under 1 Sun illumination.

Synthesis and Characterization of LM-MPI Composite
The photothermal composite architecture was developed using a two-step approach, as depicted in Figure 1.In the first step, a large LM (liquid Ga) droplet was subjected to a high-power soni-cation in an acidic methanol solution (containing 50 mm of HCl).This process produced submicron LM droplets, as schematically shown in Figure 1a.The average diameter of the particles was found to be ≈397 nm, as shown by the scanning electron microscopy (SEM) images in Figure 2a and Figure S1 (Supporting Information).In the second step, a MPI was prepared through the coordination interactions of TA (the structure of TA is shown in Figure S2, Supporting Information) with Fe 3+ and Ti 4+ ions.Then, the concentrated LM dispersion and MPI were deposited sequentially on a substrate (Figure 1) to obtain the final photothermal composite architecture.The MPI was designed so that two types of coordination complexes emerged in the ink that could absorb visible light in two different wavelength ranges (see for details in the following section), enabling broad-spectrum solar absorption properties.
The UV-vis absorption spectroscopy performed on the MPI ink revealed the presence of TA/Fe 3+ and TA/Ti 4+ complexes in the ink that displayed a ligand-to-metal charge transfer (LMCT) band at ≈570 and ≈385 nm (Figure 2b), [36] which identifies the coordination between the catechol groups from TA with Fe 3+ and Ti 4+ ions (bis/tris type in terms of gallol chelating sites, Figure 2b insets and Figure S3, Supporting Information), respectively. [40]he separate TA/Fe 3+ and TA/Ti 4+ complex solutions are also presented for comparison (Figure 2b). [40]he LMCT band for the iron-catecholate complexes in the MPI was not well-resolved since the extinction coefficient of iron-catecholate complexes are expected to be lower than that of titanium-catecholate complexes [40] in the mix that obscures signals from the former.Conversely, only TA solution was featureless in the same spectral region (350-900 nm).To get further evidence about the iron complexes in the MPI ink, Raman spectroscopy was performed as shown in Figure 2c.The peaks detected at 590 and 630 cm −1 , are known to be the characteristic peaks of Fe-O vibration due to the coordination interaction between Fe 3+ and phenolic oxygen of TA. [41] At the same time, the peak at 528 cm −1 indicates a bidentate chelation mode as observed previously. [42]The additional peaks at 1239, 1340, and 1499 cm −1 could be attributed to the catechol ring vibrations of TA [43] and the peaks at 1621 and 1719 cm −1 could be assigned to the stretching vibration mode of the benzene ring and the C=O stretching of carboxylate from TA, respectively. [43]-ray photoelectron spectroscopy (XPS) was performed to investigate the chemical composition of the LM dispersion and MPI.For the LM dispersion, the Ga 3d and Ga 2p spectra confirmed the presence of Ga in the dispersion.Specifically, the Ga 3d spectrum (Figure S4, Supporting Information) showed the presence of Ga 3+ species at binding energy (BE) of ≈20.7 eV, [44] and also showed the presence of the elemental Ga 0 (around 18.1 eV).[45,46] The Ga 2p spectrum further confirms these assignments, where the BE of ≈1116.3eV corresponds to the elemental Ga 0 of the Ga particle core, and at BE of ≈1118.51eV can be attributed to Ga 3+ species that originates from the thin oxide layer formed during the sonication.[45,47,48] For the MPI ink (Figure 2d), the presence of two characteristic peaks at BE of ≈458.9 and ≈465 eV, indicating the chemical states of Ti 2p 1/2 and Ti 2p 3/2 , respectively, that corroborates the existence of the 4+ oxidation state of Ti ions in the MPI, as also reported elsewhere.[36,49] Besides, the presence of Fe 3+ was also identified by XPS (Figure 2e) in the MPI with the two major peaks at BE of ≈712 and ≈725 eV that can be attributed to Fe 2p 3/2 and Fe 2p 1/2 , respectively.[50,51] Additionally, metal ion chelation caused a shift in the phenolic oxygen peak from 529.4 eV to a higher BE of 531.8 eV in the O1s spectra (Figure S5, Supporting Information), indicating the electron transfer process from TA to the metal ions.[35,52] These results are consistent with the Raman and UV-vis spectral analyses as presented and also with previous reports.[34,53,54] Next, the LM dispersion and the photothermal MPI were integrated into a thermoelectric module (TE module, details provided in the Experimental Section and Table S1, Supporting Information) via a sequential deposition method (Figure 2f) by dropcasting.The resulting composite coating thickness was observed to be 247 ± 15 μm as evident from the cross-sectional SEM image (Figure 2g).The corresponding energy dispersive X-Ray spectroscopy (EDS) analyses revealed the homogeneous distribution of Fe and Ti elements in the MPI coating layer (top) and the concentrated region of Ga element from the Ga particle dispersion layer (bottom) as shown in Figure 2h.Such an architecture with the photothermal composite can also be used in flexible substrates such as PET films and cellulose paper, as shown in Figure 2i.This suggests their possible usage for soft and flexible solar devices and electronics, that will be communicated in a future study.

Solar-Thermal Energy Conversion of LM-MPI Composite
To evaluate the photothermal performance of the photothermal composite (LM-MPI), the absorption capacity was assessed.As shown in Figure 3a, the LM particle layer displayed over 61% absorption capacity in the spectral range of 250-1015 nm.In the same spectral range, when the MPI layer was deposited over the LM layer, the LM-MPI showed much stronger absorption reaching ≈85.5% in absorption capacity.The reflectance of the MPI and LM-MPI was found to be low in the range of 4-14% that remained unaltered across the entire spectral range (Figure 3b).The MPI and LM-MPI had a low mid-infrared (IR) reflectance (2.5-20 μm) generating heat dissipation via thermal re-radiation.This indicates that the MPI, LM-MPI would lead to a considerable loss of absorbed heat due to high thermal re-radiation [55] in the infrared region.Meanwhile, LM presents an average solar reflectance in the range of 29-45%, ensuring sufficient reflection of sunlight and decreasing the solar heat accumulation on the surface for cooling applications.The absorption capacity for the LM, MPI, and LM-MPI on quartz and TE device is presented in Figures S6 and S7 (Supporting Information).
Theoretical calculations were also conducted to study the HOMO-LUMO gap of TA-Fe 3+ and TA-Ti 4+ metal complexes present in the MPI ink with optimized structures (Figure S8, Supporting Information).In this case, the HOMO-LUMO gap decreased in the following order TA-Fe 3+ (0.381 eV) > TA-Ti 4+ (0.109 eV), which supported the presence of a broad absorption band in the visible to NIR region.Overall, the HOMO−LUMO gaps were in good agreement with the trends observed from the optical band gaps (Figure S9, Supporting Information) deduced from the UV-vis absorption spectra.
The strong broadband light absorption, as demonstrated for the LM-MPI photothermal composite, can be attributed to the following main reasons: 1) the LMCT bands stemming from the TA/Fe 3+ and TA/Ti 4+ complexes in the MPI produces an enhanced absorption range in the visible spectrum, that results in a lower band gap of the MPI ink compared to TA, improving its light-harvesting capacity and the total photothermal effect. [56]Additionally, such LMCT absorption can endow a photothermal ability toward the conversion of light into heat, which is previously observed in other photothermal TA systems, [57][58][59][60] 2) the wide particle size distribution (20-1.4μm) of the LM particles may also contribute to the overall light absorption in some specific spectral region, [61,62] and 3) the rough surface (Figure S10, Supporting Information) and porous structure of the composite allow not only to trap light but also to enhance multiple scattering of the incident light in confined spaces that enables efficient absorption of broadband solar light. [63]he photothermal properties of the LM dispersion and the LM-MPI composite under light irradiation were investigated in a dry state using a xenon lamp to simulate the solar radiation (Figure S11, Supporting Information).First, the surface temperature distribution (monitored with an IR camera) under 1 Sun irradiation (100 mW cm −2 ) for a specific period was carried out for the LM-MPI composite deposited on a glass substrate.Figure 3c shows the time evolution of the temperature increase of the LM-MPI composite, LM dispersion, and MPI ink.It was found (Figure 3c,d) that the temperature of the LM-MPI composite (54.4 °C) was higher than the glass (40.8 °C) and Ga-coated glass (31.7 °C).This higher temperature difference (13.6 °C) between the LM-MPI composite and the non-coated glass (control) was observed within 40 min of simulated Sun irradiation.Furthermore, MPI ink and the LM-MPI composite exhibited a higher temperature (Figure 3c,d) than the bare substrate (glass) when heated simultaneously, indicating a strong light-to-heat conversion capability.
A temperature difference, known as the thermal gradient, is a fundamental requirement for the efficient thermoelectric energy conversion for TE devices.The larger the temperature difference between the hot and cold sides of the device, the higher the voltage generated. [64]As such, the photothermal performance was evaluated on a TE device (Figure S12, Supporting Information) via the surface temperature measurement (hot side).The top surface temperature of the LM-MPI device (Figure 3e,f) after  S3 (Supporting Information).
irradiation of 1 Sun for 1590 s was found to be 56.2 °C, representing 14.9 °C higher than the bare TE device.The MPI showed a similar temperature increase after irradiation.
Efficient light-to-heat conversion with uniform temperature distribution can be achieved for LM-MPI and MPI under 1 Sun irradiation.As shown in Figure S13 and Table S2 (Supporting Information), in addition to the enhanced voltage, the LM-MPI and MPI devices exhibited the most extended decay lifetime (about 22.10 and 22.05 s, respectively) compared to the bare generator (15.02 s) and LM only devices (16.70 s).
The photothermal performance was further evaluated by irradiating the MPI coating layer with an IR laser ( = 808 nm, power of 1 W).As shown in Figure S14 (Supporting Information), the surface temperature of the MPI coating rapidly increased from the onset of irradiation, exceeding 180 °C within 30 s and reached a maximum at 216 °C.This result indicates the excellent photothermal conversion ability of the MPI absorber as in line with the above discussion.Additionally, the photothermal conversion efficiency ( s ) was calculated from the obtained cooling curve for MPI (Figure S15, Supporting Information) using the formula (detailed in Note S1, Supporting Information) as described previously. [12,13]From the calculations, the photothermal conversion efficiency of MPI was found to be ≈93%, which is higher than some of the best performing materials reported in literature, for example, PVA/LMP-rGO composites (89.4%) [12] and LM/polyphenol/oxidized polyethylene composites (77.3%). [13]

Thermoelectric Performance of LM-MPI STEG
Finally, the thermoelectric performance of the LM-MPI composite was assessed in a TE device (Figure 4).Briefly, the electricity produced by the Seebeck effect was quantified per unit area of the TE device.As such, the TE module was placed between the LM-MPI composite and the thermostatic station to generate electricity when a temperature gradient is produced between the two surfaces.The LM-MPI STEG device delivered temperature differences of 0, 6.2, and 10.3 °C under different light intensities of 0, 1, and 2 kW m −2 (0, 1, and 2 Sun, respectively, Figure S16, Supporting Information).
The LM-MPI STEG device generated an output open-circuit voltage (V oc ) of 185.3 mV under 1 Sun irradiation (Figure 4a), which was 3.2 times higher than the bare TE device (57.6 mV).As shown in Figure 4b, under 1 Sun illumination, the LM-MPI produced a current of 29.8 mA, compared with a current output of 8.8 mA for the blank device (without absorber).Utilizing its remarkable photothermal properties, the LM-MPI STEG device exhibits a power density of 345.5 μW cm −2 (Figure 4c) under 1 Sun irradiation.][67][68][69][70][71] The observed power density for LM-MPI STEG represents a ≈tenfold enhancement relative to the bare TE module (31.7 μW cm −2 ) and a 10% improvement relative to the TE module covered with only MPI.The LM dispersion over the TE device thus increases the effective surface area for heat transfer between the MPI and TE device.It also increases the thermal and electrical conductivity, [72,73] enabling a higher current flow resulting in the enhanced power density generated in the LM-MPI composite.As previously reported, when an insulating polymer and an inorganic metal are combined, they can form a composite with exceptional thermoelectric properties. [74]he solar thermoelectric performance of the LM-MPI device could be further enhanced by varying the intensities of the solar light irradiation.At the solar illumination intensity of 0, 1, and 2 kW m −2 , the V OC of the device was found to be 1.5, 185.3, and 352.9 mV (Figure 4d), respectively.Meanwhile, the generated current increased to 0.03, 29.83, and 39.46 mA.The power density also increased to 870.5 μW cm −2 when irradiated at the intensity of 2 Sun (Figure 4f).
The maximum output power from the LM-MPI composite is calculated to be 3.455 W m -2 under 1 Sun illumination of 1 kW m −2 , around ten times as high as that of the pristine TE without the composite, representing an efficiency of 0.346%.The performance of this STEG exceeds the majority of solar thermoelectric generators reported in the literature without concentration or vacuum enclosure (Figure 4i and Table S3, Supporting Information).In addition, to verify the stability of the thermoelectric device, we performed a cyclic test on the system by turning on and off the simulated sunlight for five cycles.As illustrated in Figure 4g,h, the V oc and the I sc had a reproducible response, and there was no significant attenuation in the power density during the five cycles.These results suggest that the LM-MPI STEG device can generate electricity with durable performance and efficiency.During an outdoor experiment (Figure S17, Supporting Information), the resulting temperature difference between the two sides generates electricity with an average recorded voltage and current of 47.74 mV and 38.80 mA, respectively.The output power density in this case was calculated to be 115 μW cm −2 .This value was expected to be lower than the simulated ones because of the lower sun intensity on that particular day and heat losses by air convection.
We have also investigated the effect of some compositional and environmental parameters including the LM particle size, thickness of the LM particle layer, Fe 3+ :Ti 4+ ratio in the ink and humidity on the device performance.These data are presented in the Supporting Information.Within a range of 20-100 μm of the LM droplet layer thickness, the output power density did not vary significantly as shown in Figure S18 (Supporting Information).The particle size variation also did not have a significant impact (Figure S19, Supporting Information) on the power density possibly because of the polydisperse nature of the LM particle dispersion.With increasing Fe 3+ :Ti 4+ ratio in the MPI, the power density was found to increase slightly (Figure S20, Supporting Information).However, we note that increasing the ratio to a higher value led to the coagulation of the MPI, and this aspect was not investigated further.In addition, with increasing humidity (Figure S21, Supporting Information) we also observed slightly increasing output power densities, suggesting our device can be used in different environments.
The harnessed output power of the standard LM-MPI STEG device (345.5 μW cm −2 ) can be further increased by cascading more than one such solar thermoelectric modules.It is also possible to improve the performance with a thermal concentration strategy by enlarging the area of the absorber.By implementing various engineering techniques, such as incorporating optical lens array to achieve optical concentration, it is possible to further enhance the STEG performance by concentrating solar radiation onto the surface.Additionally, the use of vacuum enclosure can guarantee vacuum operation to minimize conductive and convective losses; as most heat loss is through air convection, leading to low opto-thermal efficiency.

Conclusions
In summary, our research highlighted a general platform to design an efficient photothermal absorber using a combination of LMs and metal-phenolic ink for solar energy harvesting.While the MPI ink coating provides a rough surface consists of charge transfer complex centers (catechol/iron and catechol/titanium), with a strong broadband light absorption capacity, LM particles layer provided the capability to absorb in the NIR region and boosts the conductive features of the composites.By integrating such a broadband photothermal absorber with a thermoelectric generator, we demonstrated a scalable approach to fabricate a LM-MPI STEG that generates electricity with a power density of 345.5 (under 1 Sun) and 870.5 μW cm −2 (under 2 Sun), respectively, that outperforms the contemporary STEG materials.The flexible, composite architecture proposed here should unlock a potential avenue for green solar harvesting.

Experimental Section
Materials: Gallium (Ga, beads, 99%) was purchased from Roto Metals, USA.Tannic acid, iron (III) nitrate nonahydrate, titanium (IV) bis(ammonium lactato) dihydroxide solution (Ti-BALDH, 50 wt% in H 2 O) were purchased from Sigma-Aldrich and used as received.Methanol was obtained from Chem-supply and HCl (32%) was purchased from RCI Labscan Ltd.High-purity (Mili-Q) water with a resistivity of 18.2 MΩ cm was obtained from an inline Milipore RiOs/Origin water purification system.
Synthesis of Ga Particles: The Ga ingot was first liquified on a hot plate at 60 °C, and an aliquot of Ga (500 μL) was added to 30 mL of methanol containing 50 mm of HCl.Then, the resulting dispersion was sonicated for 2 h (at 40% amplitude) using a probe sonicator (VCX 750, Sonics & Materials, Inc., tip no 630-0435).To prevent the sample heating during sonication, an ice-water bath was used.
Preparation of the Photothermal Composite Coating on Thermoelectric Devices: Two milliliters of the Ga particle dispersion (after sonication) were first deposited over a (hot side) thermoelectric device.Afterward, the coated thermoelectric device was left to dry at room temperature for 24 h.In the next step, the MPI (TA-Fe 3+ -Ti 4+ ) ink (2 mL) was poured over the Ga particle layer on the thermoelectric device, resulting in a double-layered thermoelectric device.The final samples were dried at 30 °C for 60 min in a vacuum oven and later in a hot plate at 26 °C for 24 h before usage.
To investigate the effect of different compositional parameters such as LM layer thickness, LM particle size, Fe 3+ -Ti 4+ ratio in the MPI ink, the thermoelectric devices were prepared in identical manner as above.The standard LM-MPI STEG device with an output power density of 345.5 μW cm −2 was used to assess the effect of humidity (Figure S21, Supporting Information).
Characterization: The SEM images were obtained using JSM-IT500HR InTouchScope (JEOL) at an accelerating voltage of 10 kV.Mapping of EDS was obtained using a JSM-IT500HR InTouchScope at an accelerating voltage of 20 kV.Raman spectroscopy (inVia Raman microscope, RENISHAW, 532 nm laser source) was used to characterize the MPI.For surface analysis, XPS was performed on an ESCALAB250Xi spectrometer (Thermo Scientific, UK).The incident radiation was monochromatic Al K X-rays (1486.68 eV) at 120 W (13.8 kV × 8.7 mA).All data were processed using Advantage software, and the energy calibration was referenced to the C 1s peak at 284.8 eV.The UV-vis absorption spectra were recorded on a Cary 5000 (Agilent, USA).The UV-vis-NIR spectra were recorded on a spectrophotometer (LAMBDA 1035, PerkinElmer, USA) with an integrating sphere of 150 mm, and the absorbance spectra was calculated from 100 -%R.The reflectance spectra in the IR band (1.6-25 μm) were measured by a Fourier transform infrared spectrometer (VERTEX 80v, Bruker, USA) with an Au-coated integrating sphere A562.The Thermimage package (v.4.1.3)in R was employed to analyze thermal images. [75,76]The resistance (R) of the samples was measured by a 2-point probe method at room temperature (Table S4, Supporting Information).The photothermal efficiency of the MPI coating layer was evaluated with an 808-nm optical fiber-coupled diode NIR laser (Hi-Tech Optoelectronics Co., Ltd.Beijing, China).Differential scanning calorimetry (Netzsch STA 449 F3 Jupiter) was used to determine the specific heat capacity (Cp) of the samples.
Thermoelectricity Generation: A commercial 68 W thermoelectric (TE) module (40 mm × 40 mm) with W/leads (Model: TEC1-12708/ZP9100; internal resistance: 3.0 ± 0.1 Ω; Jaycar Electronics, Sydney, NSW, AU) was used for the TE measurements.Specifically, the photothermal composite was deposited on the top side of the TE module, and the bottom side of the TE module was placed on an aluminum heat sink which was immersed in a water bath at 18 °C (EHEIM, model: 2373 51 0).
The TE performance was measured using a solar simulator (Sun-LiteTM, model: 11002-2, Abet Technologies, USA ASTM Class A with 100 W Xe arc lamp) as a light source, and a quartz disc (thickness:0.5 mm) was used as an optically transparent cover between the sample and the incident light source.The open circuit voltage and current were measured using a multimeter (SIGLENT, model: SDM3065X), and the surface temperature of the composite was recorded using an infrared camera (Model: FLIR ONE Pro, Teledyne FLIR LLC).The ambient relative humidity was monitored by a temperature and humidity recorder with a portable device (TA298, RSpro).
Calculations for Geometry Optimizations: All geometry optimizations performed in the present work are implemented with the Biovia Materials Studio DFT-Dmol3 software. [77,78]Geometry optimizations were obtained until the maximum force values were less than 0.004 Ha Å −1 .For geometry optimization, the generalized gradient approximation [79] was used with the functional proposed by Perdew and Wang (PW91), [80] for restricted spin.
DFT Calculations: The optimization of molecules was carried out using the DFT functional with basis set of B3LYP from the DMol3 code of material studio setting the criteria of ground state, restricted spin and basis set DN 4.4.After optimization, the value of highest occupied molecular orbital (HOMO), lowest unoccupied molecular orbital (LUMO), and ΔE (the energy required to promote an electron from the HOMO to the LUMO) were calculated from the optimized structure.
Solar Thermoelectric Conversion Efficiency Calculation: Typically, a solar thermoelectric generator's efficiency is calculated according to the following equation: where P output is the output power density, P input is the input power density of light.
The Decay Process of Voltage: The decay process of the V oc after the light switched off was investigated by fitting the decay curves with a oneorder exponential formula as follows: where V is the actual V oc in real-time, t is time, and  is decay lifetime, A and B are constants.

Figure 1 .
Figure 1.Schematic illustration of the assembly process of photothermal composite and deposition over TE device.a) Preparation of Ga particle dispersion.b) Synthesis of the MPI ink (TA-Fe 3+ -Ti 4+ ).c) The fabrication process of LM-MPI composite.d) The deposition of MPI over LM particle layer.e) The final LM-MPI photothermal composite architecture.f) Illustration of the LM-MPI STEG under sunlight.

Figure 2 .
Figure 2. Characterization of the Ga LM particles and the MPI (TA-Fe 3+ -Ti 4+ ) ink. a) SEM image of the Ga LMs dispersion.b) UV-vis absorption spectra of MPI, TA-Fe 3+ , TA-Ti 4+ , and pure TA. c) Raman spectra of TA-Fe 3+ -Ti 4+ ink and pure TA.XPS spectra of TA-Fe 3+ -Ti 4+ ink.d) Ti 2p core level.e) Fe 2p 3/2 core level.f) Photograph of LM-MPI over TE device.g) SEM cross-sectional image of the LM-MPI composite.h) EDS analyses of the cross-section of LM-MPI.i) Photograph of the LM-MPI composite over flexible substrates.

Figure 3 .
Figure 3. Photothermal performance of the composites.a) Absorbance spectra of the LM dispersion, LM-MPI, MPI, and Peltier in the 250-2000 nm wavelength range.b) Reflectance of LM particles and LM-MPI and MPI in the 2.0-25 μm band.c) Temperature profiles of LM, MPI, and LM-MPI composite over the glass under 1 Sun irradiation.d) Infrared-thermal images of the glass cover under 1 Sun irradiation.e) Temperature profiles of TE devices coated with LM, MPI, and LM-MPI composite under 1 Sun irradiation.f) Infrared-thermal images of the TE hot side under 1 Sun irradiation.

Figure 4 .
Figure 4. Solar thermoelectric performance of LM-MPI.a) Open circuit voltage (V OC ) generated from the integrated STEG under 1 Sun irradiation.b) Short-circuit current (I SC ) generated from the incorporated STEG under 1 Sun irradiation.c) Comparison of output power density under 1 Sun irradiation.d) V OC , e) I SC , and f) comparison of output power density under 2 Sun irradiation.Stability determination in cyclic tests: the performance of the STEG exposed to the irradiation of 1 Sun irradiation in terms of output g) V OC and h) I SC .i) Comparison of output power density to similar systems reported in literature.Details of these studies are presented in TableS3(Supporting Information).