Bio‐Inspired Mesoporous Solar Evaporator Enabled by Upcycled PET Authors: Sungdo Kim, Sourav Chaule, YuJi Han, Yong Soo Kim, Ji‐Hyun Jang Published: 2026 ABSTRACT ABSTRACTSolar‐driven interfacial steam generation (ISSG) has emerged as a promising strategy for decentralized freshwater production. However, developing scalable solar evaporators that combine high evaporation performance, environmental compatibility, and stable operation under varying solar angles remains challenging. Here, we report a compound‐eye‐inspired three‐dimensional solar evaporator based on graphite‐containing PET nanofoam (GPF). The GPF was produced by upcycling polyethylene terephthalate (PET) waste. The nonsolvent‐induced phase separation process produces a mesoporous and moldable structure with tunable water transport properties, enabling the construction of hemispherical multi‐surface evaporators. This architecture provides angle‐insensitive solar exposure and an enlarged water‐wetted evaporative interface within a fixed projected footprint. As a result, the optimized evaporator achieved a projected‐area‐normalized evaporation rate of 3.44 kg m− 2 h− 1 under 1‐sun illumination. In outdoor operation, the integrated evaporation‐condensation module produced a freshwater yield of up to 24.4 kg m− 2 over 14 h under natural sunlight. The enhanced performance arises from the combined effects of expanded evaporation‐active area, capillary‐sustained water transport, side‐surface evaporation, and internal heat redistribution. This result indicates that the 3D‐CE improves evaporation through coupled transport and thermal effects, rather than through surface‐area enlargement alone. This work presents a potential route for converting waste PET into functional solar evaporators for decentralized freshwater production.Keywords: bio‐inspired 3d‐compound‐eye structure, interfacial solar steam generation, nano‐porous structure, pet waste upcycling Introduction Polyethylene terephthalate (PET), widely used in applications ranging from food packaging to medical tools, generates approximately 30 million tons of wasssste annually [1]. PET waste is typically disposed of by landfilling or incineration; however, these methods can impose environmental burdens and social costs [2, 3]. Due to its chemical properties, PET exhibits high resistance to biodegradation, leading to the formation of microplastics (∼µm) that eventually infiltrate groundwater [4, 5]. Additionally, the incineration process releases greenhouse gases, including carbon dioxide. These adverse effects highlight the need to replace conventional PET disposal methods with more environmentally conscious alternatives. Global warming and freshwater scarcity are increasingly pressing environmental and societal challenges [6, 7, 8]. Less than 1% of Earth's water resources are accessible freshwater suitable for human use. This already limited fraction is increasingly strained by population growth and urbanization [9, 10, 11, 12]. As freshwater depletion becomes increasingly critical, the demand for and importance of desalination technologies continue to grow [13]. Interfacial solar steam generation (ISSG) is a solar‐driven desalination strategy that can operate with localized heat generation at the water–air interface [14, 15, 16]. In ISSG, minimizing thermal losses to the environment is essential for directing photothermal energy toward evaporation [17, 18, 19]. Efficient water transport sustains evaporation by distributing water across the solar evaporator surface through capillary action. The mechanism reduces the temperature gradient with the surrounding environment, minimizing energy loss [20, 21]. Consequently, solar evaporators that combine photothermal materials with porous structures have become a focus of research [22]. Carbon‐based foams are promising solar evaporator materials owing to their strong light absorption, porous structure, efficient vapor release, and buoyancy [23, 24]. Plastic waste can serve as a recyclable precursor for carbon‐foam solar evaporators [25]. Conventional methods for fabricating carbon foam from PET include template‐assisted heat treatment, carbonization of freeze‐dried polymers, and coating or decorating carbide materials. For instance, H. Bai et al. (2022) suggested a carbon foam‐based solar evaporator fabricated through template‐assisted low‐temperature heat treatment (≥ ∼350°C) [26]. The solar evaporator demonstrated notable performance (∼2.4 kg m− 2 h− 1) due to the enthalpy reduction properties with water transport capability. However, recycled PET‐based solar evaporators still show limited performance compared with other material platforms [27, 28, 29]. Three‐dimensional architectures have been widely explored to improve solar evaporator performance beyond conventional two‐dimensional designs [30, 31, 32]. For instance, spherical structures enable omnidirectional light absorption, while columnar designs can use cooler side surfaces to exchange heat with the surrounding environment [33, 34]. Surface‐patterned evaporators enhance solar absorptance and induce Marangoni convection, whereas cone‐shaped structures further improve energy capture by reabsorbing reflected light from inclined surfaces [35, 36]. These strategies have improved projected‐area‐normalized evaporation rates by expanding the effective evaporation interface and modifying heat and mass transfer. Insect compound eyes, with repeated optical units arranged on a curved surface, provide a useful motif for designing angle‐insensitive three‐dimensional interfaces [37, 38, 39]. In solar evaporators, these hemispherical multi‐surface geometries reduce the incident‐angle dependence of solar exposure. They also expand the water–air interfacial area within a fixed projected footprint. Evaporation occurs at the water–air interface. Thus, high evaporation performance requires an expanded interfacial area, efficient water supply, and effective vapor release [40]. Previous studies have improved evaporation by increasing the effective surface area relative to the projected sunlit area and enhancing wettability. However, most reported designs are still based on relatively simple geometries, such as cylinders, bridges, spheres, and hemispheres [41, 42, 43]. In this study, graphite‐based 3D‐foam (GPF) was fabricated from waste PET using the nonsolvent‐induced phase separation (NIPS) method at room temperature [44, 45]. The NIPS process enabled uniform dispersion of graphite within the PET matrix. Incorporating polyethylene glycol (PEG) generated a porous structure. The resulting graphite‐containing PET framework provided the moldability required for constructing macroscopic three‐dimensional solar evaporators. The compound‐eye‐inspired hemispherical multi‐surface architecture improved wettability, salt resistance, and self‐cleaning behavior. At the same time, it enabled angle‐insensitive solar exposure and expanded the water‐wetted evaporative interface. As a result, the optimized solar evaporator achieved a projected‐area‐normalized evaporation rate of 3.44 kg m− 2 h− 1 under 1 kW m− 2 illumination. In outdoor operation, the integrated evaporation‐condensation module produced a freshwater yield of up to 24.4 kg m− 2 over 14 h under natural sunlight. Results and Discussion Design and Fabrication of a Compound‐Eye‐Inspired Three‐Dimensional Solar Evaporator The compound‐eye‐inspired three‐dimensional solar evaporator (3D‐CE) was prepared by integrating a moldable GPF with a hemispherical multi‐surface architecture. GPF was prepared from waste PET through a NIPS process at room temperature (Figure 1a). Waste PET was first dissolved in a mixed solvent of trifluoroacetic acid (TFA) and dichloromethane (DCM), followed by the addition of graphite and polyethylene glycol (PEG). The mixed solvent system facilitates polymer‐chain relaxation at room temperature. This process is mainly associated with protonation of the ester carbonyl oxygen in PET and acid‐mediated interactions with PEG. Graphite was selected as the photothermal component owing to its comparable evaporation performance, low cost, and processing simplicity (Supporting Information S1). When deionized water was added as a nonsolvent, PET precipitated with uniformly dispersed graphite. Meanwhile, PEG served as a removable template for forming the porous framework [47]. The GPF fabrication process is shown in Movie S1, and the PEG‐induced domain formation within the PET precipitate is shown in Movie S2. Overview of the compound‐eye‐inspired three‐dimensional solar evaporator. (a) A PET waste upcycling process used to fabricate the GPF through the NIPS method. (b) An Illustration of the compound‐eye‐inspired hemispherical multi‐surface solar evaporator fabricated using GPF. The inset shows an image of the compound eye structure of a male Grapholita molesta (Adapted from Yang et al., Front. Physiol. (2024), CC BY 4.0) [46]. (c) Structural advantages of the proposed evaporator: (i) reduced angular dependence of solar exposure and (ii) tunable water transport capability controlled by PEG content. The solvent‐based NIPS process promoted homogeneous dispersion of the photothermal material within the PET matrix. This was supported by position‐dependent optical measurements and elemental mapping (Figure S2). During nonsolvent addition, the dissolved PET first formed granular precipitates, followed by the formation of a porous sheet‐like precipitate that bridged the surrounding granules. This sequential precipitation process preserved the molded shape of GPF and generated interconnected pores for capillary water transport. The resulting GPF retained the molded structure while transporting water through the porous network. This moldable and porous framework allowed the fabrication of complex GPF‐based architectures (Figure S3). Its scalability was also supported by the fabrication of a 20 cm‐diameter hemispherical structure (Figure S4). Using this moldable GPF framework, the 3D‐CE was prepared by assembling small hemispherical units on a larger hemispherical substrate (Figure 1b). The compound‐eye‐inspired architecture served as a structural motif for the hemispherical multi‐surface evaporator, rather than a biological explanation for evaporation enhancement (Figure S5). This geometry reduces the incident‐angle dependence of solar exposure and expands the water‐wetted evaporative interface within a fixed projected footprint (Figure 1c). Characteristics and Optimization of GPF The physicochemical properties and water‐transport behavior of GPF were examined to identify the optimized composition for solar evaporation. Figure 2a shows the Raman spectra of graphite, PET, and GPF‐P1.0. The PET reference exhibited characteristic vibrational bands associated with ester C = O stretching, CH2‐related vibration, and trans/gauche ethylene glycol modes. The graphite reference showed the characteristic D, G, and 2D(G′) bands of graphitic carbon [48, 49]. In the GPF‐P1.0 spectrum, PET‐derived vibrational bands were observed together with carbon‐related D, G, and broad 2D(G′) features. These spectral features support the coexistence of PET and graphitic carbon signatures in the composite. The PEG template was not detected in the Raman spectra or XRD pattern, indicating its removal during the washing process (Figure S7c). The predominantly physical interactions among PET, graphite, and PEG were further supported by component‐extraction experiments (Supporting Information S6). Characteristics and optimization of graphite‐containing PET foam (GPF). (a) Raman spectra of raw materials and GPF (GPF‐P1.0). (b) UV–Vis–NIR absorption spectra of PET, GPF, and GPF‐P1.0. (c) SEM images of GPF and GPF‐P1.0. (d) BET adsorption/desorption isotherms of GPF and GPF‐P1.0. The inset shows the corresponding pore‐size distribution. (e) Time‐dependent water contact angles of GPF and GPF‐P1.0. (f) Water content of GPF and GPF‐P1.0 as a function of time. The inset images show the saturated wet state of hollow hemispherical GPF‐P1.0. (g) Surface temperature profiles and (h) evaporation rate and estimated energy‐loss‐corrected photothermal conversion efficiency as a function of PEG content. (i) Long‐term cycling stability of the solar evaporator under 1‐sun illumination using seawater with a ∼ 3.5 wt% salinity. The top panel shows the evaporation rate measured over 20 repeated 10 h cycles using the same solar evaporator. The bottom panel shows the cumulative mass change over a total operation time of 200 h. (j) Salt crystallization resistance and self‐cleaning behavior of GPF‐P1.0. The upper images show salt crystallization resistance, with inset IR images indicating the surface temperature at each illumination intensity. The lower images show the self‐cleaning process under 1‐sun illumination after accelerated salt accumulation at approximately 20 sun. (k) Ion concentrations of seawater and collected freshwater. The dashed lines indicate the freshwater standards set by the WHO and EPA. The optical and pore‐structural characteristics of GPF were evaluated by UV–Vis–NIR spectroscopy (Figures 2b and S7d–f), SEM (Figures 2c and S7g–i), and BET analysis (Figure 2d). Absorption was calculated from diffuse reflectance and transmittance measurements using A = 100 − (R + T). Pristine PET showed limited absorption in the visible and near‐infrared regions. Graphite incorporation gave GPF strong broadband absorption, with an average value of approximately 95.2% over the 250–2500 nm range. GPF‐P1.0 maintained high broadband absorption of approximately 89.9% after PEG templating. SEM images showed that PEG separated the PET precipitates and produced a rough porous framework. BET analysis further revealed a larger pore volume and broader pore‐size distribution after PEG templating (Supporting Information S8). The specific surface area increased from 7.3 m2 g− 1 for GPF to 16.5 m2 g− 1 for GPF‐P1.0. These results indicate that PEG‐induced porosity provides a structural basis for capillary water transport but slightly reduces optical absorption by increasing diffuse reflectance (Figure S8). The accompanying change in mechanical strength with PEG content is summarized in Supporting Information S9. The water‐transport behavior of GPF was then examined by contact‐angle (Figure 2e) and water‐content measurements (Figure 2f). GPF showed an initial contact angle of 85° and required approximately 3.2 s for complete water absorption. In contrast, GPF‐P1.0 showed a lower initial contact angle of 78° and absorbed water within approximately 0.7 s. This faster wetting behavior is attributed to the increased pore volume and roughened framework, which facilitate capillary water uptake. Water‐content measurements further showed uniform water absorption throughout the hollow hemispherical GPF‐P1.0 structure. The absorbed water reached approximately 1.3 times the dry mass of GPF‐P1.0. This improved water uptake was also visualized in Movie S3. These results show that PEG templating promotes wettability and water uptake, which are critical for sustaining a water‐wetted evaporative interface (Supporting Information S10). The PEG content was optimized by balancing photothermal heating, capillary water supply, and water retention. As the PEG content increased, the surface temperature gradually decreased from 49.9 ± 0.5°C for pristine GPF to 34.9 ± 0.3°C for GPF‐P1.50 (Figure 2g). This decrease reflects the increased water content and heat capacity of the porous framework. The evaporation rate showed a non‐monotonic trend. It increased with PEG addition and reached a maximum at GPF‐P1.0, with an evaporation rate of 2.22 kg m− 2 h− 1 and an estimated energy‐loss‐corrected photothermal conversion efficiency of approximately 65% (Figure 2h). Further increasing the PEG content reduced the evaporation rate despite improved water uptake. These results indicate that excessive porosity can retain unnecessary water and increase heat capacity, whereas insufficient porosity limits capillary‐driven water supply [21]. Therefore, GPF‐P1.0 was selected as the optimized composition for subsequent 3D‐CE fabrication. The long‐term evaporation stability of GPF‐P1.0 was assessed under 1‐sun illumination using seawater (Figure 2i). The same solar evaporator was tested over 20 repeated 10 h cycles, corresponding to a total operation time of 200 h. The evaporation rate remained stable throughout the repeated cycles, and the cumulative mass change increased linearly with operation time. These results support the cycling stability of GPF‐P1.0 under seawater and 1‐sun conditions. Additional stability tests under different salinity conditions and after UV aging further confirmed the environmental and mechanical stability of GPF‐P1.0 (Supporting Information S11). Salt crystallization resistance and self‐cleaning behavior were further examined using GPF‐P1.0 (Figure 2j). No apparent salt accumulation was observed after 5 h of evaporation in 3.5 wt% saline water under 1–3 sun illumination. This result supports the salt resistance of GPF‐P1.0 under moderate solar intensities. To further evaluate salt removal under a severe condition, approximately 20‐sun illumination was applied for 1 h. This condition served as an accelerated salt‐stress test rather than a practical desalination condition. Under this intensified condition, salt crystals accumulated on the top and side surfaces of the solar evaporator. When the illumination intensity was reduced to 1 sun, the deposited salt gradually dissolved and was removed from the solar evaporator surface within 4 h. This self‐cleaning behavior is attributed to continuous capillary water supply through the porous GPF network. The supplied water promotes salt dissolution and transport away from the evaporation surface. Finally, the quality of the collected freshwater was evaluated by ion‐concentration analysis (Figure 2k). The concentrations of major ions, including Na+, K+, Mg2 +, Ca2 +, and Cl−, were substantially reduced after desalination and remained below the freshwater standards set by the WHO and EPA. Residual solvents and possible leachates were further examined by pH testing, GC‐MS, and optical absorbance measurements. No detectable acidic residues, DCM, or PET‐derived optical signatures were observed in the collected freshwater (Supporting Information S12). The Impact of Solar Evaporator Geometry on Solar Energy Capture Geometry‐dependent solar exposure was evaluated by combining CFD‐based incident‐flux analysis with projected‐area estimation (Figure S16). Disc‐ and hemisphere‐shaped evaporators with identical projected areas were compared as representative planar and curved geometries. According to Lambert's cosine law, the local incident flux on a tilted surface decreases as the angle between the surface normal and the solar beam increases. Thus, the curved surface of the hemisphere can exhibit a lower local incident flux per unit area than the disc at many solar elevation angles. However, its continuous curvature allows a broader surface region to remain exposed to sunlight under oblique illumination. This analysis suggests that the hemispherical geometry does not enhance angular response by increasing local solar flux. Rather, it redistributes incident solar energy over a wider curved surface, which can reduce localized overheating and moderate angle‐dependent thermal variations. To evaluate this geometric effect, disc‐ and hemisphere‐shaped GPF‐P1.0 evaporators with identical projected areas were tested under controlled incident angles (Figure 3a). The hollow geometry minimized water retention while preserving surface curvature (Supporting Information S14). At normal incidence, the disc reached a higher maximum surface temperature of 39.1°C ± 0.4°C than the hemisphere (35.5°C ± 0.3°C). This difference is consistent with the higher local incident flux on the planar surface under normal illumination. As the incident angle increased, the surface temperature of the disc decreased substantially. In contrast, the hemisphere maintained a nearly constant temperature, varying by only approximately 0.2°C for each 30° change in incident angle. This reduced angular dependence is consistent with the curved surface of the hemisphere, which maintains a light‐exposed region under oblique illumination. Geometry‐dependent solar exposure, thermal response, and evaporation behavior of three‐dimensional solar evaporators. (a) IR thermal images of disc‐ and hemisphere‐shaped solar evaporators under different incident angles. The insets show the corresponding top‐view temperature distributions. (b) Projected‐area‐normalized evaporation rates of disc‐ and hemisphere‐shaped solar evaporators at different incident angles. (c) Photographs of large‐area disc‐ and hemisphere‐shaped solar evaporators with a diameter of 10 cm. (d) Outdoor IR thermal images of the disc‐ and hemisphere‐shaped solar evaporators. The marked temperature values represent the average surface temperatures within the dashed regions. (e) Surface‐temperature profiles along the indicated paths on the disc‐ and hemisphere‐shaped solar evaporators. (f) Time‐dependent average surface temperatures of the disc‐ and hemisphere‐shaped solar evaporators during outdoor operation from sunrise to sunset, together with the solar altitude. (g) Time‐dependent surface‐temperature distribution of the 3D‐CE solar evaporator under outdoor sunlight. The different angular responses were also reflected in the projected‐area‐normalized evaporation rates (Figure 3b). The hemisphere‐shaped evaporator maintained higher evaporation rates than the disc at all tested incident angles, decreasing only from 2.73 ± 0.06 kg m− 2 h− 1 at 0° to 2.36 ± 0.13 kg m− 2 h− 1 at 60°. In contrast, the disc‐shaped evaporator decreased from 2.22 ± 0.05 to 1.11 ± 0.05 kg m− 2 h− 1 over the same angular range. The hemispherical geometry can reduce the local incident flux per unit surface area. However, it provides a larger water‐wetted evaporative interface within the same projected footprint. This enlarged interface helps offset the local flux reduction. This balance allows the hemisphere‐shaped evaporator to maintain higher footprint‐level evaporation under changing solar incidence. The angle‐dependent evaporation behavior of the completed 3D‐CE is further summarized in Figure S18. The geometric effect was further examined under natural sunlight using large‐area disc‐ and hemisphere‐shaped evaporators with a diameter of 10 cm (Figure 3c). The measurement was conducted at noon, when the solar altitude angle was 36.81°. Under this condition, the hemisphere exhibited a higher average surface temperature within the irradiated region than the disc. The average surface temperature was 37.6°C for the hemisphere and 32.7°C for the disc (Figure 3d). This result differs from the CFD‐based incident‐flux analysis, which suggested that the disc can receive a higher local incident flux at the corresponding solar altitude. Thus, the outdoor thermal response cannot be explained solely by the local solar‐flux distribution. Spatial temperature profiles showed a temperature increase near the side surface of the disc. In contrast, the hemisphere reached its highest surface temperature around the 40 mm position along the curved surface (Figure 3e). In addition to geometry‐dependent solar exposure, substrate contact can influence conductive heat dissipation. The smaller contact area of the hemisphere may reduce heat loss to the support and contribute to its higher average surface temperature. Outdoor time‐dependent measurements further showed that the hemisphere remained hotter than the disc from sunrise to sunset (Figure 3f). The average temperature difference was approximately 3.8°C. This result supports the reduced angular dependence of the hemispherical geometry under natural solar‐angle variation. At the same time, the lower local flux distributed over a broader curved surface can moderate localized overheating and produce a more spatially distributed thermal response. Time‐dependent IR images of the 3D‐CE showed a shift in the heated region across the architecture as the sun moved (Figure 3g). The surface temperature remained above the ambient temperature throughout the measurement. Therefore, the thermal response of the hemispherical geometry should not be attributed to increased local solar flux or net environmental heat gain. Instead, it reflects geometry‐mediated redistribution of solar exposure and structure‐dependent heat dissipation, including possible differences in contact area with the supporting substrate. Evaporation Performance and Outdoor Freshwater Production of the 3D‐CE We next optimized the 3D‐CE architecture by controlling the size and number of hollow hemispherical units assembled on a larger hollow hemispherical substrate. The fabrication procedure for the 3D‐CE and its module components is described in Figure S19. Before varying the number of attached units, the diameter of the hollow hemispheres was optimized because it affects local curvature, surface‐area amplification, and water‐wetted interfacial area. Among the tested diameters, the 5 mm hemispherical unit produced the highest evaporation rate (Figure S21). We then varied the number of attached hemispheres to examine the effect of unit density on evaporation performance. The resulting 3D‐CE configurations are shown in Figure 4a. Evaporation performance and outdoor freshwater production of the 3D‐CE. (a) Schematic illustration of 3D‐CE solar evaporators with different numbers of attached hemispherical units. (b) Projected‐area‐normalized evaporation rate of the 3D‐CE as a function of the number of attached hemispherical units. (c) Schematic illustration of thermal‐energy distribution and heat‐loss management in hemispherical‐ and 3D‐CE‐based solar evaporators. (d) Photograph and schematic of the outdoor evaporation‐condensation module equipped with a 3D‐CE solar evaporator. (e) Photograph of the scaled‐up 3D‐CE solar evaporator. (f) Time‐resolved outdoor performance over four daytime cycles. (g) Stepwise enhancement in projected‐area‐normalized evaporation rate from GPF to the compound‐eye‐inspired solar evaporator. (h) Comparison of projected‐area‐normalized evaporation rate with representative PET‐recycling‐based ISSG devices. The evaporation rate was normalized by the projected irradiation area to evaluate the footprint‐level device performance. To clarify the area basis, Table S5 summarizes the area‐normalized evaporation performance of each 3D‐CE configuration using both projected and total geometric surface areas. As the number of attached hemispheres increased, both the mass change and the projected‐area‐normalized evaporation rate increased up to the 16‐hemisphere structure (Figure 4b). The optimized 3D‐CE achieved a projected‐area‐normalized evaporation rate of 3.44 kg m− 2 h− 1 under 1‐sun illumination. When normalized by the total geometric surface area, the corresponding evaporation rate was approximately 1.13 kg m− 2 h− 1. Therefore, the 3.44 kg m− 2 h− 1 value should be interpreted as footprint‐level device performance within a fixed projected area, rather than intrinsic interfacial evaporation flux. Further increasing the number of hemispheres to 19 reduced the evaporation rate despite the larger geometric surface area. This non‐monotonic trend indicates that the performance enhancement is not governed solely by surface‐area enlargement. Rather, the optimal structure requires a balance between water‐wetted area amplification, capillary water supply, and vapor release. The structural‐optimization analysis further supports this interpretation (Figure S22). Excessive hemispherical‐unit density partially restricted local vapor‐transport pathways around the 3D‐CE and reduced the dark evaporation rate. Figure 4c illustrates how the 3D‐CE translates geometry‐induced area amplification into productive evaporation through geometry‐mediated thermal‐energy distribution. In the hemispherical GPF‐P1.0 evaporator, absorbed solar energy is distributed across a single curved water‐wetted interface. By contrast, the 3D‐CE divides and expands this interface into multiple hemispherical units within the same projected footprint. This multi‐surface architecture generates position‐dependent solar flux and surface‐temperature gradients across the 3D structure. These gradients can promote conductive heat transfer through the continuous GPF framework. As a result, absorbed heat can contribute to evaporation over a broader water‐wetted interface, rather than being confined to strongly illuminated regions. This interpretation is supported by the conduction‐related evaporation analysis, which suggests that solar‐flux‐induced temperature differences within the 3D structure can assist evaporation over a broader water‐wetted interface (Supporting Information S19). An estimated energy‐balance comparison was further performed by considering radiative, convective, and conductive heat‐loss terms. Compared with the hemispherical GPF‐P1.0 evaporator, the estimated radiative, convective, and conductive heat losses were slightly lower for the 3D‐CE, decreasing from 11% to 10%, from 9% to 8%, and from 0.8% to 0.6%, respectively (Tables S7–S9). Based on these estimated heat‐loss terms, the energy‐loss‐corrected photothermal conversion efficiency was estimated to be 71.1% for the 3D‐CE, compared with 67.9% for the hemispherical GPF‐P1.0 evaporator (Table S10). This analysis should be interpreted as an estimated energy‐balance comparison rather than a rigorous experimental closure of the complete heat budget. Together, these results suggest that the enhanced projected‐area‐normalized evaporation rate of the 3D‐CE results from coupled geometric, transport, and thermal effects. These effects include expanded water‐wetted evaporation area, capillary‐sustained water supply, temperature‐gradient‐assisted heat redistribution, and moderated heat dissipation. Therefore, the performance enhancement should not be attributed to geometric surface‐area enlargement alone. The optimized 3D‐CE architecture was scaled up for outdoor freshwater production. The outdoor module integrated the enlarged 3D‐CE solar evaporator with a water‐supply structure and floating support (Figure 4d). The scaled‐up evaporator was prepared by attaching 16 optimized hemispherical units to a 10 cm‐diameter hemispherical substrate, preserving the compound‐eye‐inspired multi‐surface geometry at the module scale (Figure 4e). Outdoor freshwater production was evaluated using the scaled‐up 3D‐CE module over four daytime cycles from 06:00 to 19:00 under natural sunlight (Figure 4f). The control experiments in Supporting Information S21 show that the vapor‐collection configuration affects the local operating environment. Thus, the outdoor performance should be interpreted as module‐level freshwater production rather than intrinsic open‐condition evaporation. The water collection rate followed the solar‐intensity profile and reached a maximum of 1.74 kg m− 2 h− 1. The maximum projected‐area‐normalized freshwater yield reached 24.4 kg m− 2 over 14 h of daytime operation, corresponding to approximately 70% of the total vapor generation of 34.83 kg m− 2. The stepwise performance evolution highlights the coupled contribution of material design and geometry (Figure 4g). PEG templating improved water transport, whereas the hollow hemispherical and 3D‐CE architectures progressively expanded the water‐wetted evaporative interface within the projected footprint. The highest projected‐area‐normalized evaporation rate of the 3D‐CE therefore results from the combined effects of capillary water supply, multi‐surface geometry, and internal heat transfer, rather than surface‐area enlargement alone. The 3D‐CE was benchmarked against representative PET‐recycling‐based ISSG devices (Figure 4h), and structure‐based comparisons with reported three‐dimensional solar evaporators are summarized in Supporting Information S23. These comparisons position the PET‐upcycled 3D‐CE as a competitive footprint‐level ISSG device architecture that combines projected‐area‐normalized evaporation rate with waste‐polymer valorization. Conclusion This study demonstrates a PET‐upcycling platform for interfacial solar steam generation. Through nonsolvent‐induced phase separation, waste PET is converted into a tunable graphite‐PET foam that enables precise control of porosity, wettability, and moldability. By constructing a compound‐eye‐inspired 3D architecture, we show that the improved projected‐area‐normalized evaporation rate arises from coupled geometric, water‐transport, and thermal effects, rather than reduced evaporation enthalpy. The optimized device further demonstrated salt resistance and operational stability under repeated seawater‐evaporation conditions. Beyond performance metrics, this work provides design guidelines for geometry‐driven solar evaporation and presents a route for waste‐PET valorization into functional desalination systems. Future efforts toward solvent recovery or solvent replacement will be important for improving the environmental profile of the current TFA/DCM‐based process. Experimental Section Preparation of GPF PET and PEG are added to a mixture of trifluoroacetic acid and dichloromethane. Once the polymers are fully dissolved, graphite powder is introduced and dispersed into the solution. All additives are stirred at 300 rpm for approximately 20 min to ensure uniform dispersion, after which distilled water is added. The mixture is further stirred until PET precipitates form, after which it is shaped using a mold. Detailed information is provided in Supporting Information S1 and Movie S1. Fabrication of Solar Evaporator The solar evaporator consists of 3D‐CE, the water supply, and the floating support. The 3D‐CE and the water supply are based on the GPF fabrication method but differ in the materials added. The 3D‐CE is fabricated using GPF‐P1.0, while the water supply is made from GPF‐P1.25 without the addition of graphite. The floating support is fabricated using a polymer solution under the GPF‐P0.0 condition without graphite and is immersed in an ethanol bath to induce phase separation. The fabrication process for each component is detailed in Supporting Information S16. Characterization The morphology of GPF was measured using SEM (Seceng SNE‐4500 M and Hitachi High‐Tech Corporation SU7000). Pore characteristics were measured using BET (Micromeritics ASAP 2420). The crystallinity of GPF was observed using an x‐ray diffraction analyzer (BRUKER D8 Advance) with a Cu‐Kα (λ = 0.154 nm) radiation source. Raman spectroscopy data were obtained using a micro PL/Raman setup (Dong Woo Optron MonoRa500i) with a wavelength of 473 nm and an input power of approximately 2 mW. The ion concentration of seawater and desalinated water was measured using inductively coupled plasma (Agilent 5800 ICP‐OES) and ion chromatography (Metrohm Dual Channel Advanced Ion Chromatograph). To calculate light absorption, reflectance and transmittance (measured with an Agilent Cary 5000) were used. Absorption was calculated as A = 100−R−T (%). Time‐lapse images of water droplet absorption by GPF were obtained using a contact angle meter (Phoenix 300). A reference PV cell (PV Measurements PVM1056) and a sourcemeter (KEITHLEY 2450 SourceMeter) were used to measure the average solar intensity at different times. Mass changes due to evaporation were measured using a high‐precision electronic balance (Vibra AJH‐220E‐D), and a precision electronic balance (Ohaus SPX6201) was used in outdoor environments. The surface temperature of the interfacial solar desalination device under solar irradiation was monitored using a thermal imaging camera (Testo 868) with a resolution of 320×240. A centrifuge (LABOGENE 1248) was used to extract materials from the GPF. The surface area value used for the evaporation rate calculation in Figure 4c was derived using the Fiji Is Just ImageJ software. The CFD simulations are performed using SOLIDWORKS software (Dassault Systèmes). Outdoor Experimental Conditions The outdoor experiments were conducted in Ulsan, South Korea. (1) Experiments for Figure 3c–f took place on November 6, 2024, from 6:00 (sunrise at 6:50 AM) to 19:00 (sunset at 5:23 PM). The average temperature on that day was 9.4°C (maximum: 14.6°C, minimum: 5.6°C), with an average cloud cover of 0. (2) Experiments for Figure 3g werssse conducted on June 14, 2024, with an average temperature of 25.4°C (maximum: 31.3°C, minimum: 20.2°C) and an average cloud cover of 5.3. (3). Experiments for Figures 4f and S24 were conducted on May 16 to 19, 2026. Saline Water Source Information The saline water used to evaluate the performance of the 3D‐CE‐structured solar evaporator was collected from natural seawater. The source location is Ilsan‐dong, Dong‐gu, Ulsan, Republic of Korea (35°29'45“N, 129°25'51”E). Evaporation Rate Calculation The performance of 3D‐CE is evaluated by projected‐area‐normalized evaporation rate (kg m−2 h−1). The evaporation rate and efficiency of the solar evaporator are calculated based on the mass change observed under light emitted by a Xenon lamp (CERMAX, PE300BFA). The equations used for the calculations (Equations 1 and 2) are sequentially presented below. where, m.(kgm−2h−1) is projected‐area‐normalized evaporation rate, Δm (g) represents water mass reduction over 1 h (3600 s), A (m2) is an area exposed to light, and t is irradiation time, which is based on a 1 h standard. Mass change was monitored from the interval in which the mass‐change rate shows a linear trend. Photothermal Conversion Efficiency Calculation In solar desalination, the photothermal conversion efficiency is defined as the percentage obtained by dividing the total enthalpy used for evaporation by the absorbed solar energy. where m. is the net evaporation rate, hfg is the total enthalpy of liquid–vapour phase change (sensible heat and phase‐change enthalpy), Cp is the specific heat of water, ΔTis the temperature rise of the evaporating water, Ais the illuminated area, and q solaris the absorbed solar flux. Because the efficiency calculated by Equation (2) can exceed 100% under certain conditions, we estimate radiative, convective, and conductive heat‐loss terms and report an estimated loss‐corrected photothermal efficiency (Section S20). Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supporting File 1: smll74735‐sup‐0001‐SuppMat.docx. Supporting File 2: smll74735‐sup‐0002‐movieS1.mp4. Supporting File 3: smll74735‐0003‐movieS2.mp4. Supporting File 4: smll74735‐sup‐0004‐movieS3.mp4.