Flexible Self‐Cleaning Janus Emitter for Transparent Radiative Cooling in Enclosed Spaces

Abstract

Passive daytime radiative cooling is a sustainable and cost‐efficient strategy that works by reflecting solar radiation and emitting heat to the cold universe through infrared radiation. However, cooling enclosed spaces that retain heat due to the greenhouse effect remains a significant challenge, particularly for transparent radiative coolers. Here, the Janus Transparent Radiative Cooler (JTRC) is introduced, which simultaneously achieves strong solar spectrum reflectivity, high infrared emissivity, and high heat absorptivity within enclosed spaces, all while maintaining transparency for practical real‐world applications. Uniquely functioning as a Janus device, with one side acting as a selective emitter and the other as a broadband emitter, this design effectively extracts internal heat while blocking external heat. As a result, it minimizes heat accumulation in enclosed spaces, effectively addressing the greenhouse effect. During the daytime, the JTRC achieved a temperature reduction of 20 °C compared to a conventional transparent radiative cooling method. This transparent, flexible, self‐cleaning, and high‐performance cooler demonstrates significant advancements over conventional designs and highlights its potential for practical applications in vehicles, buildings, and electronic devices requiring thermal management.

A Janus transparent radiative cooler is introduced to address the challenge of heat buildup in enclosed spaces. By combining selective and broadband infrared emitters on opposite sides, the device effectively extracts internal heat while blocking solar gain. With a 20 °C temperature reduction and flexible, self‐cleaning design, it offers practical potential for thermal management in buildings, vehicles, and electronics.

1. Introduction

The growing global demand for energy‐efficient and environmentally sustainable technologies, along with climate change, has amplified interest in radiative cooling.^{[ 1} , ² , ³ , ⁴ , ^{5 ]} This technique, capable of lowering temperatures without external energy input, offers a promising solution to the increasing need for energy‐intensive cooling systems. Radiative cooling operates by leveraging the principles of blackbody radiation, enabling thermal energy to escape directly into outer space through the atmospheric window (AW, 8–13 µm), a spectral range where Earth's atmosphere is highly transparent to infrared (IR) radiation.^{[ 6} , ⁷ , ^{8 ]} This method, particularly during the daytime, requires not only high emissivity in the AW but also high reflectance in the solar spectrum (0.3–2.5 µm) to minimize heat absorption from sunlight.

Extensive research has improved the thermal efficiency of daytime radiative cooling (DRC) devices through advanced designs using multilayers,^{[ 9} , ¹⁰ , ¹¹ , ¹² , ^{13 ]} micro/nanostructures,^{[ 14} , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ^{19 ]} and random particle/porous structures.^{[ 15} , ²⁰ , ²¹ , ²² , ²³ , ^{24 ]} Notable work includes the Janus emitter radiative cooler, which simultaneously emits thermal energy to outer space while absorbing accumulated heat from an enclosed space, further enhancing temperature reduction within the space.^{[ 25 ]} However, these DRC technologies often exhibit opaque white or metallic gray appearances due to their high reflectance in the visible spectrum (VIS, 0.4–0.7 µm). This visual characteristic, while functional, limits its aesthetic appeal and applicability in architectural or consumer‐facing designs. Enhancing the aesthetics of radiative coolers, including incorporating colors or transparency, has become an important objective for commercialization and broader adoption.^{[ 26} , ²⁷ , ²⁸ , ²⁹ , ^{30 ]}

Transparency in radiative cooling systems presents significant potential for broadening their applications, particularly in windows, glass facades, and coatings.^{[ 31} , ³² , ³³ , ³⁴ , ³⁵ , ^{36 ]} Previous efforts to design efficient transparent radiative coolers (TRC) have demonstrated varying levels of success. For instance, Zhu et al.^{[ 14 ]} developed a patterned SiO₂ photonic crystal TRC that achieved a temperature reduction of 1.3 °C compared to a planar SiO₂ substrate. However, its performance was limited under full solar exposure. In a similar vein, Kim et al.^{[ 12 ]} introduced a TRC combining a selective Bragg reflector (BR) with a polydimethylsiloxane (PDMS) emitter, which achieved partial solar reflectance while maintaining VIS transparency. Despite this, a portion of the solar energy was still absorbed, which reduced the overall cooling efficiency. To further improve performance, Ko et al.^{[ 19 ]} employed a window screen reflector with a 90 µm hole in the Ag film, enabling the reflection of the unwanted solar spectrum in specific regions. However, this design is prone to high temperatures due to the greenhouse effect, as it allows solar radiation to enter but prevents the escape of long‐wave thermal radiation, leading to potential overheating.

In this work, we designed and prepared a Janus transparent radiative cooling (JTRC) device. The JTRC is a simple sandwich structure consisting of the selective emitter (SE), IR reflector, and broadband emitter (BE). Three radiative coolers (SE, SE‐IR reflector, and JTRC) were fabricated, and the JTRC demonstrated superior cooling performance by effectively reflecting near‐infrared (NIR) radiation from the outside and absorbing internal heat, maintaining the lowest internal temperature compared to conventional glass window and other radiative cooling samples during outdoor experiments. The potential applications were further evaluated with a self‐assembled monolayer (SAM) coating, which enhanced the JTRC surface's hydrophobicity for self‐cleaning without affecting its optical or radiative properties, ensuring durable cooling performance. These strategies advance innovative passive cooling methods for surface and space applications, while simple, cost‐effective fabrication techniques offer a practical, scalable solution for radiative cooling devices.

2. Results

2.1. Concept of Janus Transparent Radiative Cooling

Solar radiation penetrating through windows can substantially raise the internal temperature of a stationary vehicle.^{[ 25 ]} The JTRC acts as an efficient thermal conduit by absorbing broadband thermal radiation from the interior through its bottom side and emitting heat as IR waves from its top side to outer space. (Figure 1a). The JTRC device consists of a 50‐µm‐thick ethylene tetrafluoroethylene (ETFE) layer, a 271‐nm‐thick IR reflector layer, and a 700‐µm‐thick PDMS layer. ETFE film is commonly used in large‐scale architecture and industry due to its durability, lightweight, and transparency. ETFE was selected as SE due to its high durability in the outdoor environment and superior selectivity in the AW region.^{[ 37 ]} PDMS is a commonly used material in laboratory settings and can be easily deposited with adjustable thickness.^{[ 38} , ^{39 ]} In this study, we deposit it to a thickness of 700 µm to function as a broadband emitter (Figure S2, Supporting Information).

Figure 1.

Janus Transparent Radiative Cooler (JTRC) concept and design. a) Schematic of the JTRC, consisting of a selective emitter (SE), an IR reflector made of a dielectric/metal/dielectric structure and a Bragg reflector, and a broadband emitter (BE), which transmits visible, reflects near‐infrared, exhibits high emissivity in the atmospheric window (AW), and absorbs heat from the enclosed space below. b) Ideal transmittance and reflectance spectra of JTRC with the light orange shaded area represent the spectral irradiance taken from the standard direct spectrum (AM 1.5D). Ideal emissivity spectra within c) AW (blue shaded area) and d) blackbody radiation (dark orange shaded area) regions. Schematic and working principle of e) SE (only ETFE) cooler, f) SE‐IR reflector (ETFE and IR reflector) cooler, and g) JTRC (SE, IR reflector, and BE) on the enclosed space, respectively.

Ideally, the top side of the JTRC should exhibit selective emission aligned with the AW window to minimize interference from solar and ambient radiation (Figure 1c). Simultaneously, the bottom side should demonstrate broadband absorption of interior thermal radiation for optimal performance (Figure 1d). The IR reflector layer primarily reflects the solar spectrum and effectively separates the top and bottom sides in the far‐infrared region, allowing the JTRC to operate independently as both a SE and a BE (Figure 1b). While the BE serves as a thermal absorber in this device, it also functions as an effective emitter in accordance with Kirchhoff's law (Note S3, Supporting Information). Accordingly, we adopt the terminology established in a prior study^{[ 25 ]} to ensure consistency with the designation of the SE.

Figure 1e describes the conventional transparent radiative cooling method. This method is not suitable for daytime radiative cooling, as it allows a significant amount of solar energy to enter both the VIS and NIR regions. The method in Figure 1f still allows solar energy to enter the VIS range, but it blocks the NIR region, making it transparent to the eye while remaining partially usable during the daytime. However, it also reflects outgoing radiation, which affects its overall efficiency. Figure 1g describes our proposed design. This approach enhances emissivity on the inner surface, facilitating the escape of trapped heat by increasing the internal radiative flux, which is further optimized through natural convection.

2.2. Thermal Exchange of the Cooler and Optimization of the Infrared Reflector

The thermal interaction of a radiative cooler^{[ 40 ]} can be described by:

$$P_{cool}\left(T,T_{amb}\right)=P_{rad}\left(T\right)-P_{sun}-P_{atm}\left(T_{amb}\right)-P_{cc}\left(T,T_{amb}\right)$$ (1)

In this context, P_cool represents the total cooling power, P_rad denotes the power emitted by the cooler, P_sun refers to the solar power absorbed by the cooler, P_atm is the atmospheric power absorbed by the cooler, and P_cc accounts for the non‐radiative energy transferred to the cooler via convection and conduction. The first three thermal flux components are described as follows:

$$P_{rad}\left(T\right)=\int d\mathrm{\Omega }cos\theta \underset{0}{\overset{\infty }{\int }}d\lambda I_{BB}\left(T,\lambda \right)\epsilon \left(\lambda ,\theta \right)$$ (2)

$$P_{sun}=cos\theta \underset{0}{\overset{\infty }{\int }}d\lambda I_{Sun}\left(\lambda \right)\epsilon \left(\lambda ,{\theta }_{Sun}\right)$$ (3)

$$P_{atm}\left(T_{amb}\right)=\int d\mathrm{\Omega }cos\theta \underset{0}{\overset{\infty }{\int }}d\lambda I_{BB}\left(T_{amb},\lambda \right){\epsilon }_{atm}\left(\lambda ,\theta \right)\epsilon \left(\lambda ,\theta \right)$$ (4)

here, ∫dΩ represents the integral over all solid angles, θ is the polar angle, I _BB/I _sun is the blackbody and solar irradiance, and ϵ/ϵ_atm refers to the emissivity of the cooler and the atmosphere, respectively. The term ϵ_atm can be further defined as ɛ _atm =  1 − t(λ)^1/cosθ , where t(λ) is the atmospheric transmittance in the zenith direction. The non‐radiative term P _cc is given by:

$$P_{cc}\left(T,T_{amb}\right)=h_{cc}\left(T_{amb}-T\right)$$ (5)

here, h_cc represents the overall heat transfer coefficient for convection and conduction at the radiative cooler. The primary objective is to maximize P _cool while improving the visibility of the transparent cooler. Visibility denoted as T̅_vis, is defined as the average transmittance over the wavelength range of 400–700 nm. Maximizing P _cool corresponds to maximizing R̅_NIR, where R̅_NIR is defined as the average reflectance over the wavelength range of 700–2500 nm.

The reflector consists of a dielectric‐metal‐dielectric (DMD) layer with a thin metal film between dielectric layers and an additional bottom BR layer. The dielectric material in the DMD layer, with significantly higher permittivity than the metal, limits surface plasmon coupling and reduces absorption, enabling transparency at specific frequencies and strong NIR reflection.^{[ 41} , ⁴² , ^{43 ]} To achieve high permittivity contrast, titanium dioxide (TiO₂) and silver (Ag) were used. Alternating layers of silicon dioxide (SiO₂) and TiO₂ were added to form the bottom BR layer, enhancing NIR reflection. A discussion of the materials used is provided in Note S2 (Supporting Information). With the Ag layer fixed at 22 nm (Figure S3, Supporting Information), the remaining thicknesses were optimized for high VIS transparency and strong NIR reflection using rigorous coupled‐wave analysis (RCWA) simulations.

To determine the optimal thickness of the dielectric layers in the DMD structure (D1/22 nm/D2) for achieving high solar reflectance (R̅_NIR) and VIS transmittance (T̅vis), the thicknesses D1 and D2 were systematically varied from 0 to 60 nm in 1 nm intervals. T̅vis and R̅_NIR were calculated as functions of D1 and D2 to identify the optimal configuration (Figure 2a,b). The optimal values for D1 and D2 were selected to maximize R̅_NIR while ensuring T̅_VIS > 0.72, meeting both the legal minimum for car windows and practical user requirements.^{[ 42 ]} The optimized thicknesses were determined to be 30 nm/22 nm/29 nm, and the corresponding spectrum is presented in Figure 2c.

Figure 2.

Optimization of the infrared reflector. Simulated contour plots of a) T̅_VIS and b) R̅_NIR for varying thicknesses of two TiO₂ layers (D1 and D2) with Ag fixed at 22 nm. c) Optical property spectrum of the optimized TiO₂/Ag/TiO₂ structure. Simulated contour plots of d) T̅_VIS and e) R̅_NIR for varying thicknesses of the SiO₂ and TiO₂ layers (D3 and D4), with D1 at 30 nm, Ag at 22 nm, and D2 at 29 nm. f) Optical property spectrum of the optimized TiO₂/Ag/TiO₂/SiO₂/TiO₂ structure. The refractive indices of the ETFE film and PDMS are set to 1.4.

To optimize the thickness of the BR layer composed of SiO2/TiO₂ (D3/D4), the values of D3 and D4 were varied from 0 to 300 nm in 10 nm intervals. T̅_VIS and R̅_NIR were calculated as functions of D3 and D4 to identify the optimal configuration (Figure 2d,e). The optimal values for D3 and D4 were selected to maximize R̅_NIR while ensuring that T̅_VIS > 0.72. The optimized thicknesses were determined to be 30 nm/22 nm/29 nm/90 nm/100 nm, with the corresponding spectrum shown in Figure 2f.

2.3. Preparation and Characterization of JTRC

Three types of radiative coolers, namely SE, SE‐IR reflector, and JTRC are fabricated. The fabricated coolers are shown in Figure 3a. The SE displays a neutral color and clear transparency. However, the SE‐IR reflector and JTRC appear to be a slightly blueish tone, resulting from the interference effect caused by light reflecting between different interfaces. This can be varied depending on different refractive index and film thickness.^{[ 44 ]} Despite this, the SE‐IR reflector and JTRC still remain transparent, and the effect of color is within an acceptable range (Figures S4 and S5, Supporting Information).

Figure 3.

a) Photograph of the three fabricated radiative cooler samples (5 cm × 5 cm each). Logos used with permission from Pohang University of Science and Technology, and Korea University. b) Scanning electron microscope (SEM) image of the infrared reflector (scale bar: 100 nm). c) Solar wavelength spectrum of the three fabricated samples. d) Long wavelength infrared spectrum of the three fabricated samples.

Figure 3b presents a scanning electron microscope (SEM) image of the JTRC, showing the DMD structure stacked with the BR layers. Then, we measure the transmittance, reflectance, and absorptivity spectra of the radiative coolers within the solar range of 0.3–2.5 µm. The optical responses in the solar wavelength of all the coolers are presented in Figure 3c. The SE demonstrates a high transmittance across the entire solar spectrum with T̅_vis accounts for 91.3%. However, it exhibits near zero reflectivity throughout the solar range (R̅_NIR = 4.4%), leading to the absorption of incoming solar flux. This functionality is ideal for nighttime radiative cooling but is not effective for daytime use. In contrast, the SE‐IR reflector and JTRC offer high reflectance in the NIR region while maintaining high transmittance in the VIS spectrum, making them effective for daytime cooling applications. The SE‐IR reflector exhibits T̅_vis of 56.9% and R̅_NIR of 84.7%, whereas JTRC shows T̅_vis of 57.8% and R̅_NIR of 79.6%. The SE‐IR reflector and JTRC exhibit similar spectral responses in the solar region, as both are designed with the same optimized multilayer structure reflector.

Figure 3d shows the emissivity of the coolers in the long wavelength infrared (LWIR) region for both the top and bottom sides of the coolers. The emissivity on the top side of all the coolers is similar, as they all feature the same top emitter layer. They exhibit high emissivity in the AW, making them highly effective for radiating heat. On the bottom side, the absorptivity of SE exhibits similar behavior to the top side, as it consists of the only ETFE emitter (${\overline{\epsilon }}_{\mathbf{top}}$= 52.8%, ${\overline{\epsilon }}_{\mathbf{bottom}}$ = 52.8%). In contrast, the SE‐IR reflector shows near‐zero absorptivity due to its IR reflector broadband reflection property in the infrared region (ɛ_top = 69.5%, ɛ_bottom = 8.4%). As a result, only the JTRC demonstrates a high broadband absorptivity, enabling it to effectively absorb heat from the enclosed space, thereby enhancing the cooling performance (${\overline{\epsilon }}_{\mathbf{top}}$ = 71.2%, ${\overline{\epsilon }}_{\mathbf{bottom}}$= 93.7%). Since all IR regions are reflected within the enclosed space by an IR reflector, the only pathway for heat to escape is through heat absorption. The higher emissivity of JTRC (${\overline{\epsilon }}_{\mathbf{bottom}}$= 93.7%) compared to the SE‐IR reflector (${\overline{\epsilon }}_{\mathbf{bottom}}$= 8.4%) suggests its effective prevention of the greenhouse effect. To compare the selective emissivity between these three coolers, we introduce a selective ratio (γ), which is defined as the ratio of the average AW emissivity (8 to 13 µm) to non‐AW emissivity (2.5 to 8 and 13 to 15 µm). For the top‐side, γ of SE, SE‐IR reflector, and JTRC are 2.19, 1.66, and 1.57, respectively. While γ of the bottom‐side are 2.19, 0.99, and 1.03 for SE, SE‐IR reflector, and JTRC, respectively. We can see that the selective ratio of the top and bottom sides of SE behave similarly since the cooler consists only of the emitter. While the SE‐IE reflector shows efficient cooling from the top, but suppressed bottom emission due to IR reflection. Lastly, JTRC presents balanced emissivity, suggesting heat absorption from the enclosed space while still emitting outward.

2.4. Cooling Performance and Global Cooling Potential

Next, we numerically assess the cooling performance of the radiative coolers placed on top of an enclosed space using thermal balance equations. It is important to note that these thermal balance equations do not account for the greenhouse effect, as the control volume of the thermal equation is defined as the entire chamber. Hence, the presence of BE does not influence the calculated cooling performance. It should be emphasized that since the SE‐IR reflector and JTRC share the same optimized IR reflector, their optical properties in the solar range are expected to be identical. However, fabrication variations may introduce slight structural differences, leading to minor deviations in solar reflection between the samples, as shown in Figure 3c. This results in a difference in their numerical cooling powers, primarily due to variations in P _sun from differences in optical properties in the solar spectrum regime, as their emissivity in the LWIR region on the top‐side remains similar. For clarity, we will focus on comparing only SE and JTRC in this section, while the cooling power calculations and plots for the SE‐IR reflector will be provided in Figure S6 (Supporting Information). Furthermore, it is important to highlight that the greenhouse effect, influenced by the presence of BE, significantly impacts cooling performance in practical conditions. This aspect will be addressed and demonstrated in the later section.

By setting h _cc = 8 Wm⁻²k⁻¹, T_amb = 303 K, we analyze the four thermal flux components of the coolers with T varies from 300–380 K, as shown in (Figure 4a). P _rad, P _atm, and P _cc of SE and JTRC/SE‐IR reflectors are not significantly different. However, the P _sun of SE is substantially higher at 852 W m⁻², compared to just 410 W m⁻² for the JTRC/SE‐IR reflector because of a high reflectance in NIR. Based on those thermal flux components, we can calculate the P _cool of the coolers using Equation (1). The calculated P _cool is shown in Figure 4b. Since the cooling performance of the coolers is greatly affected by P _sun, the P _cool of the JTRC/SE‐IR reflector is significantly higher than SE at the same T. The x‐intercept of the graph indicates the equilibrium temperature of the coolers, which is the point where the cooling power equals zero. Additionally, P _cc, which represents non‐radiative heat transfers through conduction and convection, is influenced by h _cc. Since h _cc values depend on weather conditions, we plot equilibrium temperature against different h _cc to illustrate how different environmental conditions affect the cooling performance (Figure 4c). The equilibrium temperature of the JTRC/SE‐IR reflector is significantly lower than that of SE, with a difference of ≈60 K. Additionally, the equilibrium temperature decreases as h _cc increases, indicating that the coolers can exchange heat more efficiently with their surroundings under higher h _cc conditions.

Figure 4.

Radiative cooling performance of SE and JTRC on the enclosed space. a) Thermal fluxes of the SE (dashed) and JTRC (solid) and b) Pcool when hcc = 8 W m⁻² K⁻¹. c) Equilibrium temperatures at which the cooling power is zero for different hcc. d) Worldwide climate database of solar irradiance, including six selected cities. e) Differences in cooling performance between the SE and the JTRC worldwide, including six selected cities. The calculations are based on the average Earth's surface temperature and solar irradiance data for July 2022 sourced from the NASA Langley Research Center (LaRC) POWER Project, which is supported by the NASA Earth Science/Applied Sciences Program.

The cooling power of the radiative cooler is significantly influenced by regional weather patterns and climate conditions, leading to significant variations in cooling performance across different geographic locations. To illustrate this, we present the cooling power of our radiative coolers on a worldwide scale accounting for local climate conditions from different regions in the globe. The cooling flux for the cooler, when positioned in a space chamber, is calculated for each terrestrial location. To perform such calculations, we utilize data on the earth's surface temperature and solar irradiance in July 2022, provided by the NASA Langley Research Center (LaRC) POWER Project.^{[ 45 ]} Figure 4d–i presents the variation in solar irradiance across different locations. By setting the cooler temperature equal to the earth's surface temperature, we can calculate the cooling flux for all three samples of the cooler. Detailed data on the earth's surface temperature and global cooling flux of the three radiative coolers is shown in Figure S7 (Supporting Information). Moreover, solar irradiance for the selected six major cities worldwide is shown in Figure 4d (ii). We also present cooling power differences (ΔP _cool) between SE and JTRC, visualizing them on a global map alongside the cooling powers of six selected cities, displayed on the right (Figure 4e (i, ii)), as the ETFE‐IR reflector provides similar cooling power to the JTRC. The JTRC outperforms the SE in cooling power due to its ability to reflect solar flux in the NIR range while absorbing heat within the chamber space. This advantage is evident from the predominantly positive ΔP _cool values across most regions. JTRC also exhibits positive cooling flux in all six selected cities and most regions worldwide, confirming its suitability for global deployment. In contrast, the ETFE cooler shows negative cooling flux in the majority of the global regions, highlighting its limitations. The maximum ΔP _cool reaches up to ≈250 W m⁻², with Riyadh nearly achieving this value among the six cities. Additionally, Rio de Janeiro records the highest cooling power (68.6 W m⁻²) among them.

2.5. Radiative Cooling Performance of JTRC

Outdoor temperature measurements were conducted using three different samples, denoted as #1 (SE), #2 (SE‐IR reflector), and #3 (JTRC). These measurements were carried out during autumn weather on the rooftop of Korea University in Seoul, South Korea. The experimental setup, illustrated in Figure 5a,b, utilized a modified chamber originally designed for radiative cooling studies. The chamber was customized to evaluate the cooling performance of the JTRC,^{[ 19} , ^{41 ]} allowing for a detailed assessment of its ability to maintain lower internal temperatures. In this setup, the samples were placed on top of an acrylic box situated inside the chamber. A thermocouple was positioned at the center of the acrylic box to monitor internal temperature accurately. To minimize solar heating effects and external influences, specific surface treatments were applied to the acrylic box: the exterior was painted white to reflect sunlight, while the interior was painted black. This configuration allowed the sunlight transmitted through the samples to be absorbed by the black‐painted surface, thereby influencing the internal temperature of the box. For control and comparison purposes, measurements were taken both with and without the samples in place. The condition without any sample was denoted as T _chamber, and a transparent glass sample was used as an additional control to benchmark performance against a conventional material. A low‐density polyethylene (LDPE) film was placed over the chamber to block wind and minimize parasitic heat transfer, ensuring that heat exchange occurred primarily through radiative and conductive pathways.

Figure 5.

Outdoor temperature measurement. a) A schematic diagram and b) a photograph of the measurement system. c) Temperature data profile measured in Seoul, South Korea, on September 24, 2024. d) Average temperature of each sample between 11:00 AM and 1:00 PM.

The temperature data from the clear day on September 24, 2024, is provided in Figure 5c. The temperature profiles reveal significant differences between the samples, particularly during peak solar radiation hours. The environmental conditions during the experiment are summarized in Figure S8 (Supporting Information). Sample #1, the ETFE, lacked the ability to block solar radiation effectively. As a result, the internal temperature of the acrylic box increased significantly when exposed to direct sunlight. Despite this limitation, the selective emission properties of ETFE in the AW led to a moderate reduction in temperature compared to the broadband‐emitting glass sample. This suggests that while ETFE can partially manage radiative cooling, its performance is limited in the absence of additional reflective layers.

Sample #2, incorporating a multilayer reflector designed to selectively reflect NIR radiation, demonstrated a noticeable improvement. The multilayer structure effectively reflected a significant portion of the NIR sunlight, which constitutes a substantial part of the solar spectrum. Consequently, the internal temperature of the acrylic box was significantly lower than that of Sample #1. This result highlights the importance of selectively reflecting NIR radiation to enhance cooling performance.

Sample #3 featured both a multilayer reflector and an additional back emitter designed to absorb internal heat. This combination proved to be the most effective, as it not only reflected incoming NIR radiation but also facilitated the absorption and dissipation of internal heat. The result was the lowest internal temperature among all samples, underscoring the synergistic effects of combining selective reflection and heat absorption.

Between 11:00 and 13:00, the average temperatures recorded were 59.7 °C for Sample #1, 52.5 °C for Sample #2, and 48.5 °C for Sample #3, as shown in Figure 5d. These results clearly demonstrate the superior cooling performance of Sample #3, even under intense solar conditions. Moreover, despite the greenhouse effect caused by sealing the acrylic box with Sample #3, it maintained a slightly lower temperature than the open‐structured T _chamber. Typical transparent radiative cooling devices tend to exhibit higher average temperatures than the inner ambient due to the greenhouse effect.^{[ 19 ]} Accordingly, both the SE and SE‐IR reflector samples demonstrate similar results. However, the JTRC sample records a lower average temperature than the inner ambient, which is T _chamber, showcasing the successful application of the BE. This configuration enables the system to maintain an internal temperature that is 17.9 °C lower than a conventional glass window, 11.2 °C lower than a conventional transparent radiative cooler, and 4 °C lower than an IR‐reflecting transparent radiative cooler.

Furthermore, on April 1, 2024, we conducted the same experiment in Riyadh, Saudi Arabia, the city identified in the previous section as exhibiting the highest cooling power difference (Figure S9, Supporting Information). Interior temperature measurements indicated that Sample #3 demonstrated a temperature reduction of 20 °C compared to Sample #1, 7 °C compared to Sample #2, and 2.3 °C compared to T _chamber. Despite spatial and temporal constraints during the hazy spring weather, Sample #3 achieved a maximum temperature drop of up to 20 °C. Even greater temperature reductions are expected under clear summer weather conditions.

2.6. Application Potential Assessment

For effective radiative cooling, the surface must be exposed to the open sky. This exposure makes the surface vulnerable to contamination by dust and other pollutants. Such contamination can hinder the performance of radiative coolers by reducing VIS light transmittance, increasing solar absorption, or impairing heat dissipation. Therefore, controlling surface contamination is critical in radiative cooling applications. To address this, we aimed to make the surface hydrophobic, thereby enabling a self‐cleaning function. The surface of the JTRC is composed of ETFE, with an initial water contact angle (WCA) of ≈90° (Figure 6a). To enhance its hydrophobicity, a SAM coating was applied.

Figure 6.

a) Water contact angle (WCA) of JTRC before and after self‐assembled monolayer (SAM) coating. Reflectance and transmittance of b) front side and c) back side of JTRC before and after SAM coating. The emissivity of d) front side and e) back side of JTRC before and after SAM coating. f) Processes of durability tests for JTRC. g) Reflectance and transmittance, h) emissivity, and i) WCA of JTRC after durability tests.

After SAM coating, the WCA increased to 125°, which is sufficient to achieve a self‐cleaning effect.^{[ 41} , ⁴⁶ , ^{47 ]} However, surface coatings can potentially alter optical properties, which may negatively impact radiative cooling performance. Thus, the optical characteristics of the JTRC were measured before and after the SAM coating. Figure 6b,c show the optical properties of the JTRC's front side (ETFE) and back side (back emitter) in the solar spectrum, respectively, before and after SAM coating. The results indicate no significant differences in either transmittance or reflectance for both sides. Additionally, Figure 6d,e presents the emissivity of the JTRC before and after SAM coating. The selective emission on the front side and the broadband emission on the back side were maintained, confirming that the SAM coating did not alter the radiative properties. In summary, the SAM coating effectively enhanced the hydrophobicity of the JTRC surface, increasing its self‐cleaning capability, while preserving the optical and radiative characteristics necessary for efficient radiative cooling (Video S1, Supporting Information).

To evaluate the durability of the JTRC, a rubbing test and a UV exposure test were conducted (Figure 6f). The optical properties of the JTRC were examined before and after these tests. As shown in Figure 6g, the reflectance and transmittance of the pristine sample, the sample after the rubbing test (After process 1), and the sample after both the rubbing and UV exposure tests (After process 2) exhibited no significant differences. Furthermore, the atmospheric window emissivity remained unchanged after both tests, as shown in Figure 6h, confirming that the radiative properties of the JTRC were preserved. Additionally, the WCA of the SAM‐coated JTRC surface remained similar after process 1 and process 2, indicating that the hydrophobicity was maintained as shown in Figure 6i. These results demonstrate that the JTRC possesses both mechanical and UV resistance, ensuring its durability in practical applications.

3. Conclusion

In summary, we propose a Janus transparent radiative cooler device composed of three key components: SE for high emissivity in the AW, IR reflector for high solar flux reflectance, and BE for efficient heat absorption within enclosed spaces. The IR reflector, designed with a DMD structure integrated BR, is optimized to selectively enhance transparency in the VIS spectrum while effectively blocking solar flux NIR range. This design achieves an effective cooling performance while maintaining transparency. Specifically, JTRC achieves a visible transmittance of ≈60% and a NIR reflectance of ≈80%. JTRC also addresses several limitations of traditional TRC. While conventional TRC can suffer from heat accumulation within enclosed spaces due to the greenhouse effect, JTRC integrates BE to mitigate internal heat buildup and enhance overall cooling performance, achieving up to a 20 °C temperature reduction relative to the conventional TRC. Its flexibility enables the application to various surfaces, thereby broadening its range of practical uses. Furthermore, the self‐cleaning property of JTRC helps maintain long‐term cooling efficiency by preventing dust and pollutant accumulation. We compare previous TRC works with our JTRC, presenting their T̅_VIS, R̅_NIR, and other functionalities. A detailed comparison is provided in Table S1 (Supporting Information).

It is worth noting that higher visible light transmittance can allow more solar radiation to pass through, potentially inducing a photothermal effect^{[ 48 ]} that leads to partial diminishment of the cooling performance. However, the purpose of JTRC is to balance effective cooling performance and optical transparency for practical applications where transparency is essential. Moreover, JTRC incorporates BE at its bottom, which is specifically designed to minimize heat accumulation within the enclosed space and enhance cooling performance.

Future work could potentially leverage deep learning^{[ 30} , ^{49 ]} or other optimization methods^{[ 13} , ⁵⁰ , ^{51 ]} to enhance the structural design of the IR reflector, achieving higher reflectance in the UV and NIR ranges by efficiently exploring a broader design space. Furthermore, the thickness of the silver layer can be adjusted to control both VIS transmittance and IR reflectance. If the focus is on transmittance, a thinner silver layer can be used, whereas, for improved cooling performance, a thicker silver layer may be preferred.^{[ 43 ]} This approach enables customization based on specific demands by simply controlling the deposition thickness, unlike previous methods that require additional photolithography.^{[ 5} , ¹⁹ , ²⁵ , ^{52 ]}

ETFE film is a widely adopted commercial material in architectural applications, while PDMS, frequently utilized in research settings, offers ease of processing due to its substantial thickness of 700 µm. Although the emitter in our current work is based on isotropic polymeric materials with a planar geometry, resulting in omnidirectional emissivity,^{[ 37} , ⁵³ , ^{54 ]} future integration of microstructured layers could enable angle‐selective thermal emission and directional radiative cooling, offering new possibilities for energy‐efficient thermal management in environments with geometric constraints.^{[ 18} , ⁵⁵ , ⁵⁶ , ^{57 ]} For enhanced market feasibility, the BR layer in the IR reflector can be omitted, suggesting that with further refinement and maturation of the DMD structure mass production technology, this design holds significant potential for application in various industries, including automotive, building windows, and electronic devices requiring thermal management,^{[ 58} , ⁵⁹ , ^{60 ]} due to its high transparency, flexibility, and self‐cleaning properties.

4. Experimental Section

Simulation

An in‐house MATLAB code using RCWA was employed to design the multilayer film and analyze its optical properties. The complex refractive index n of TiO₂, SiO₂, and Ag varied with λ (Figure S1, Supporting Information). The 2D calculations assumed a periodic structure with a 500 nm x‐periodicity and infinite in the y‐direction. The surrounding medium, ETFE, and PDMS, had a refractive index of 1.4 on both sides. The Fourier expansion was truncated in the fifth order. A normally incident plane wave, polarized along the x‐direction, was simulated over a wavelength range of 300 to 2500 nm in 11 nm intervals.

Fabrication of the Radiative Cooling Device

The JTRC was fabricated using an electron beam (E‐beam) evaporator (ULVAC Technologies, EI‐5, Massachusetts, USA). A multilayer structure was deposited on an ETFE film in the following order: TiO₂ (25.4 nm), Ag (19.0 nm), TiO₂ (28.6 nm), SiO₂ (88.9 nm), and TiO₂ (103 nm). To form the back emitter, the deposited film was placed in a petri dish, and PDMS was poured over it. The PDMS was then cured in an oven at 50 °C.

Enhancing Hydrophobicity for Self‐Cleaning via SAM Coating

To improve the hydrophobicity of the surface, a self‐assembled monolayer (SAM) coating was applied. The SAM solution was prepared by mixing (heptadecafluoro‐1,1,2,2‐tetrahydrodecyl) trichlorosilane (purchased from Sigma–Aldrich) and hexane at a 1:1000 volume ratio, followed by gentle stirring with a magnetic stirrer for 30 min. Subsequently, the JTRC samples were immersed in this solution for 10 min to allow the SAM layer to form on their surfaces. After the immersion step, the samples were removed and rinsed alternately with hexane and deionized water to ensure thorough cleaning.

Properties of JTRC Characterization

The reflectance, transmittance, and absorbance of the samples in the solar spectrum range (0.3–2.5 µm) were measured using a UV–vis–NIR spectrophotometer (SolidSpec‐3700, Shimadzu, Japan) with an integrating sphere accessory. The baseline was calibrated using a Spectralon white reflector (Certified Spectralon White Diffuse Reflectance Standard, Edmund Optics Korea, South Korea). To measure the optical properties at wavelengths beyond 2.5 µm, a Fourier‐transform infrared (FT‐IR) spectrometer (Nicolet IS‐50, Thermo Fisher Scientific, USA) equipped with a mid‐IR integrating sphere accessory (Mid‐IR IntegratIR, PIKE Technologies, USA) was used. To certify surface hydrophobicity, WCA was measured via Phoenix‐MT(T) (Gyeonggi‐do, South Korea).

Outdoor Measurements

The internal and external temperatures of the chamber were measured using K‐type thermocouples with copper wires, and the data were recorded using a data logger (OM‐CP‐OCTTEMP2000, Omega Engineering, USA). Environmental conditions, including solar radiation intensity, wind speed, and relative humidity, were monitored and recorded using a weather station (HD52.3D, DeltaOHM, Italy).

Durability Tests

Durability tests were conducted, including a rubbing test and a UV exposure test. In the rubbing test, the surface of the JTRC was rubbed 50 times using a cotton fabric. For the UV exposure test, the JTRC was exposed to a 2 W UV light‐emitting diode (LED) for 10 h.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

J.P., H. Lim, and H.K. contributed equally to this work. J.P. and D.C. conceived the idea and initiated the project. J.P. designed the samples, and H.K. conducted numerical simulations and analysis. H. Lim and D.C. fabricated the samples, and H.L. conducted measurements. J.P., H. Lim, and H.K. mainly wrote the manuscript. All authors contributed to the discussion, analysis, and writing of the manuscript, and gave approval of the final manuscript. H. Lee and J.R. guided the entire project.

Supporting information

Supporting Information

Supplementary Video 1

Park J., Lim H., Keawmuang H., Chae D., Lee H., Rho J., Flexible Self‐Cleaning Janus Emitter for Transparent Radiative Cooling in Enclosed Spaces. Small 2025, 21, 2501840. 10.1002/smll.202501840

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.