Both sub-ambient and above-ambient conditions: a comprehensive approach for the efficient use of radiative cooling

Abstract

In real-world radiative cooling applications, cooling surface temperatures may periodically fluctuate between sub- and above-ambient conditions. Traditional radiative cooling surfaces with ‘static’ spectral properties cannot realize high-efficiency cooling owing to different spectral requirements for different working scenarios. Herein, we report an infrared self-adaptive radiative cooling (ISRC) approach to selectively regulate emission spectra in the range out of the atmospheric window, resulting in a broadband emitter or an atmospheric window-selective emitter. A bilayer structure that consists of an upper microporous SiO₂ fiber layer and a bottom poly(N-isopropylacrylamide) hydrogel layer was developed. Through directional transportation of a broadband emission liquid (i.e., water) in thermo-response hydrogels, the switch of the spectra between selective infrared emissions (∼0.85) under the sub-ambient cooling condition and broadband emissions (∼0.92) under the above-ambient cooling condition was achieved. Improved temperature reductions of ∼4.1 °C (sub-ambient condition) and ∼12.4 °C (above-ambient condition) were measured compared to ‘static’ spectral radiative coolers. In addition, we implemented the simultaneous maximum improvement of daytime photovoltaic (12%) and nighttime thermoelectric (80%) power with the ISRC for round-the-clock electricity generation. The proposed ISRC approach demonstrates a comprehensive way to the efficient use of radiative cooling.

---

Broader context

Radiation flux generally propagates from a high-temperature heat source to a low-temperature heat source. Objects on the earth's surface can harness the coolness using the outer space (∼3 K) as a renewable and sustainable energy source. However, only a fraction of infrared emissions (8–13 μm) can pass through the atmosphere, and the emissions out of 8–13 μm wavelengths are absorbed and reflected back. In the real world, the temperature of a cooling object periodically switches between above-ambient and sub-ambient temperature conditions due to intermittent heating fluxes such as daily/seasonal changes in solar intensity and periodic start-stops of internal heat generation devices. Therefore, net infrared radiation exchange with the atmosphere may impose heating penalties for sub-ambient objects but offer cooling benefits for above-ambient objects. This work presents a comprehensive approach for the efficient use of radiative cooling by adaptively activating radiation heat exchange with the atmosphere under above-ambient and sub-ambient conditions. We demonstrate improved temperature reductions of ∼4.1 °C and ∼12.4 °C compared to ‘static’ spectral radiative coolers under different cooling conditions.

Introduction

Radiative cooling, through which terrestrial surfaces (∼300 K) release heat into space (∼3 K), has potential applications in buildings,^1–3 thermoelectrics,^4,5 water harvesting,^6–8 and personal thermal management.^9,10 The effectiveness of radiative cooling is determined by spectral properties, including selective mode or broadband mode (Fig. 1a) and working conditions, in particular the temperature of the cooling object (T_c). When T_c is higher than ambient temperature (T_amb), that is, under the above-ambient condition, the net thermal radiation flux outside of the atmospheric window (8–13 μm) (P_atm) can be a cooling benefit. Otherwise, under the sub-ambient condition, P_atm can be a heating penalty (Fig. 1b).

Fig. 1 Concept of self-adaptive radiative cooling with the selective regulation of infrared spectrum out of the atmospheric window. (a) Schematic of self-adaptive radiative cooling with infrared spectrum selective regulation. (b) Spectral distribution of radiation intensity. P_solar, P_rad, and P_atm refer to the net heat exchange in the solar spectrum, atmospheric window, and mid-infrared range out of the atmospheric window, respectively. (c) Thermal circuits of selective mode, broadband mode, and ISRC. The automatic opening and closing of the radiation circuit between the ISRC and the atmosphere depend on the direction of temperature gradient. T_c, T_s, and T_sky refer to the temperature of the cooling object, outer space, and atmosphere, respectively. (d) Net radiative cooling power of ideal radiative coolers calculated using eqn (1). The inset illustrates the spectra of the selective cooler (orange curve) and broadband cooler (azure curve).

There have been persistent efforts to obtain selective emissions for sub-ambient radiative cooling since 1959¹¹ using compounds such as polyvinyl fluoride,¹² silicon monoxide,¹³ and silicon oxynitride.¹⁴ During the last decade, several groups have reported daytime radiative cooling with both selective materials, such as photonic crystals,¹⁵ metamaterials¹⁶ and ceramics,^17,18 and broadband materials, including P(VdF-HFP)¹⁹ and lignocellulose.²⁰ In general, broadband materials are preferred for above-ambient cooling conditions and selective materials are preferred for sub-ambient cooling conditions. However, real-world cooling objects are usually accompanied with internal heat sources or parasitic heat absorption,²¹ such as power plant condensers,²² telecommunication base stations,²³ and solar cells.^24,25 Therefore, a more complex cooling condition is the fluctuating temperature of a radiative cooling object between above-ambient and sub-ambient conditions due to intermittent heating fluxes such as daily/seasonal changes in solar intensity and periodic start-stops of internal heat generation devices.

In this work, to meet the requirements of the above-mentioned fluctuating cooling conditions, we propose a concept of self-adaptive radiative cooling with the selective regulation of the infrared spectral range out of the atmospheric window. The selective regulation of solar radiation has been proven to be meticulous thermal management based on stimulus-responsive materials such as hydrogels^26,27 and vanadium dioxide,²⁸ which can regulate solar transmittance in the visible and near-infrared ranges, respectively. Another kind of spectral selective regulation is the simultaneous modulation of solar heating and radiative cooling within the full solar spectrum and long-wave infrared range.²⁹ Wang et al.³⁰ and Tang et al.³¹ successfully developed a thermo-responsive smart window and coating with dynamic thermal management properties, respectively. Yet, the selective regulation of infrared spectrum (range out of the atmospheric window) for the self-switching of the selective/broadband cooling mode has not been reported in the literature so far (Table S1, ESI†).

Fig. 1c shows the schematic of the proposed infrared self-adaptive radiative cooling (ISRC) approach. Similar to the adaptive control of electric circuit (Fig. S1, ESI†), when T_c is higher than T_amb, the infrared spectrum of the ISRC adaptively switches to the broadband mode, radiating heat to the atmosphere and generating larger cooling power. When T_c is lower than T_amb, the infrared spectrum adaptively switches to the selective mode, rejecting heat from the atmosphere and pursuing a larger temperature reduction (Fig. 1d). The automatic opening and closing of the radiation circuit between the ISRC and the atmosphere depend on the direction of the temperature gradient. The experimental results demonstrate enhanced temperature reductions of ∼12.4 °C and ∼4.1 °C under above- and sub-ambient conditions, respectively, compared to the “static” spectrum radiative coolers, i.e., SiO₂ fibers and commercial black paints.

Results

Structure of the ISRC

To meet the requirements for spectrally self-adaptive radiative cooling, we designed the ISRC by adopting the following principles: (i) the radiative cooler should possess high solar reflectivity and selective infrared emissivity, i.e., selective mode when T_c < T_amb; (ii) the infrared spectrum should be near-blackbody in the broadband mode, but the magnitude of solar reflectivity almost stays the same as T_c changes and (iii) the spectrum can promptly switch triggered by ambient stimulation, without any active energy input. Hydrogel, as a representative of stimulus-responsive material, can smartly change its properties in response to environmental stimulations including heat, light, and moisture.^32,33 Drawing inspiration from the automatic transportation of water molecules driven by osmotic pressure differences³⁴ in the swelling and shrinking processes of thermo-responsive hydrogels, we propose a bilayer structure for the ISRC, which consists of an upper microporous SiO₂ fiber layer and a bottom poly(N-isopropylacrylamide) (PNIPAM) hydrogel layer (Video S1, ESI†).

Fig. 2a shows the working principle of the ISRC. When T_c is below the phase transition temperature τ_c, the PNIPAM hydrogel layer is in swelling condition and the SiO₂ fiber layer is dry. Due to the instinct high bandgap (∼9.0 eV) and microporous structure for multi-scattering, the upper SiO₂ fiber shows a bright white appearance. In the infrared range, the special vibrations of –O–Si–O– (800 cm⁻¹ and 1080 cm⁻¹) guarantee the selectively high extinction index of SiO₂, particularly at 8–10 μm wavelength (Fig. 2b). Consequently, the spectrum of the ISRC manifests the selective mode when T_c < τ_c. When T_c is higher than τ_c, the PNIPAM hydrogel layer shrinks and the water molecules are transported to the upper SiO₂ fiber layer. The water molecules possess a broader linewidth for phonon polaritons in the infrared range (1594.7 to 3755.9 cm⁻¹), which is why the sky possesses low transmissivity out of the atmospheric window. In addition, the extremely low ultraviolet and visual light absorptivity of water molecules in the troposphere leads to the present solar spectrum on the terrestrial surface (Fig. S2, ESI†). It means that the spectral absorptivity of the water molecules well mismatches with the solar spectrum. The adoption of water herein provides broadband emissivity, together with low increased solar absorptivity. After moistening, the solar reflectivity for a single wet SiO₂ fiber layer will decrease inevitably because of the reduction of refractive index contrast, while the solar irradiation through the upper fiber layer can be further reflected by the microstructure of the bottom shrinking PNIPAM hydrogel. In this way, the ISRC can contain a broadband infrared emissivity, simultaneously maintaining a high solar reflectivity. In particular, the reversible phase transition of the hydrogel layer meets the third requirement for ISRC, the spontaneity of which can be explained by the reduction in Gibbs free energy (0 > ΔG_mix = ΔH_mix − T_cΔS_mix). When T_c < τ_c, the interaction between hydrophilic amide groups and hydrophobic isopropyl groups within the gel network restricts molecular mobility, with enthalpy reduction (ΔH_mix < 0) dominating the ΔG_mix. When T_c > τ_c, the coiling of polymer chains increases system disorder, with entropy increasing (T_cΔS_mix > 0) dominating ΔG_mix. It is also possible to adjust the threshold temperature τ_c by modifying the composition of hydrogels to alter the system's enthalpy and entropy. For example, the general τ_c of 32 °C can be decreased to ∼25 °C or increased to ∼40 °C by adding commoners (5–10 wt%) such as alkyl methacrylate and acrylamide to affect the conformational change of polymer chains.³⁵ Although τ_c may not be in real-time synchronization with T_sky, it falls within the temperature range of most cooling objects with dynamic heat sources.

Fig. 2 Structure of the ISRC. (a) Working principle of the ISRC for infrared spectral selective regulation. (b) Complex refractive index of SiO₂ and H₂O. (c) Front view (left) and side view (right) of the developed ISRC. Blue-dyed water is well dispersed in the hydrogel layer when it is in swelling condition. When the ISRC is heated and the hydrogel layer is in shrinking condition, the water is transported to the top and the hydrogel layer shows a white appearance. It should be noted that owing to the high scattering capability of the SiO₂ fiber, the upper layer remains white even when wetted by the blue-dyed water.

Fig. 2c shows the morphology of the prepared ISRC with a size of 4 cm × 4 cm. To control the upper-directional water transportation and accumulation on the top of the ISRC, the bottom sides of the PNIPAM hydrogel layer are chemically bonded with a polydimethylsiloxane (PDMS) tray using a benzophenone solution (20 wt% in acetone), leaving only the top side for shrinking and swelling. The PNIPAM hydrogel layer was synthesized by free-radical polymerization of N-isopropylacrylamide (NIPAM) monomer and N,N′-methylenebis(acrylamide) (BIS) within the PDMS tray in an ice-water bath. The upper microporous SiO₂ fiber was prepared by electrospinning with a precursor solution of methyltrimethoxysilane, phosphoric acid, and polyvinyl alcohol (Fig. S3, ESI†) (see the details in Methods). To avoid the dissipation of water molecules during the experiment, the ISRC was finally sealed in a 7 μm–thick PE film using a commercial impulse thermal sealing machine (Fig. S4–S6, ESI†). We demonstrated the upper directional water transportation property with the appearance and disappearance of the blue-dyed water, as well as the measured resistance change during the heating and cooling processes (Fig. S7, ESI†). Moreover, the prepared ISRC possesses credible flexibility, satisfying the application in surfaces with different textures (Fig. S8, ESI†).

Spectral properties of the ISRC

Fig. 3a and b shows the micromorphology of the prepared SiO₂ fiber layer and the PNIPAM hydrogel layer respectively. To obtain high solar reflection, the diameter distribution of the SiO₂ fiber follows a log-normal distribution (the mean μ and variance σ² of which are 0.63 and 0.23 respectively) with a peak diameter of 0.75 μm. According to the Mie scattering theory, this non-uniform diameter distribution enables the peak scattering efficiencies in the whole solar spectrum region (Fig. S9, ESI†), while the scattering efficiencies attenuate with the surrounding medium transforming from gaseous air (Δn = n_SiO2 − n_air = 1.53–1.0 = 0.53) to liquid water (Δn = n_SiO2 − n_H2O = 1.53–1.33 = 0.20) (Fig. 3c). We modeled the weighted average solar reflectivity of SiO₂ fibers for different layer thicknesses and refractive index contrast Δn (i.e. medium) by using the Ray Tracing Monte Carlo Method.³⁶ As illustrated in Fig. 3d, the solar reflection capability becomes progressively significant with the increase in the layer thickness and Δn. We selected the 400 μm–thick SiO₂ fiber as the upper layer of the ISRC, owing to its solar reflectivity above 0.8 even under the wet condition. For the infrared switching after wetting, the synergistic effect of instinct vibrations of the O–H bond and improved impedance matching enables the broadband infrared absorption (emission). The effective extinction coefficient and emissivity for different fiber conditions were calculated based on the Maxwell–Garnett–Mie theory (Fig. 3e and f).³⁷ Comparing with the SiO₂ fiber under dry conditions, the wet one with a larger extinction coefficient shows a significantly increased bandwidth of infrared emission. Specifically, we incorporated 20 micrometer SiO₂ particles (the inset of Fig. 3a) into the original fiber to achieve selective infrared emission in the dry state by backward reflecting the infrared radiation beyond the atmospheric window (Fig. S10 and S11, ESI†). The desired broadband infrared spectrum can be effectively tailored in the wet state, when the thickness of the hydrogel layer reaches 2 mm. As shown in Fig. 3g, the emissivity of the ISRC at 5 μm wavelength increases from 0.236 to 0.94 with the increase in the hydrogel thickness. Additionally, the microporous shrinking hydrogel layer with a high refractive index of ∼1.5 (Fig. S12, ESI†) remedies the attenuated solar reflection of the upper SiO₂ fiber layer, especially in the visual band, where the solar reflectivity surpasses 0.91 (Fig. 3h).

Fig. 3 Spectral properties of the ISRC. Scanning electron microscopy images of the SiO₂ fiber layer (a) and PNIPAM hydrogel layer (b). The freeze-drying of hydrogels during imaging may result in pore sizes appearing larger than they actually are at high temperatures. (c) Calculated scattering efficiencies of the SiO₂ fiber with different sizes when dry and wet. (d) Calculated solar reflectivity (0.3 to 2.5 μm) of the SiO₂ fiber (refractive index n ≈ 1.53) with different thicknesses and media surrounding (n from 1.0 to 1.53). Calculated effective electromagnetic extinction coefficient (e) and emissivity (f) of a 400 μm–thick SiO₂ fiber when dry and wet. (g) Measured emissivity of the ISRC with different thicknesses of shrinking hydrogel layers at 5 μm wavelength. (h) Measured solar reflectivity of the 2 mm–thick shrinking PNIPAM hydrogel (inset). (i) Measured spectrum of the ISRC in selective and broadband modes. Measured solar reflectivity and infrared emissivity outside the atmospheric window in response to the temperature (j) and heating–cooling cycles (k).

Fig. 3i presents the measured spectrum of the ISRC for different cooling modes, comprising a 400 μm–thick SiO₂ fiber layer and a 2 mm–thick PNIPAM hydrogel layer. In the selective mode, the ISRC possesses a high solar reflectivity of ∼0.94 (weighted with the AM1.5 solar spectrum) and a selective infrared emissivity of ∼0.85 (weighted with the 300 K blackbody infrared spectrum). When it switches to the broadband mode, a considerable solar reflectivity of ∼0.90 was maintained, as well as a broadband infrared emissivity of ∼0.92. Fig. 3j and Fig. S13 (ESI†) show the solar reflectivity and infrared emissivity modulations at different temperatures. The mode switching starts at 32 °C along with a selective infrared emissivity modulation (i.e., out of the atmospheric window) ΔE_IR of ∼0.40, implying the strong temperature-responsive characteristic of the PNIPAM hydrogel layer. At above 36 °C, the infrared emissivity reached ∼0.92, indicating an unprecedentedly selective ΔE_IR value of ∼0.52. Such a dramatic transition is desired for promptly connecting the radiation circuit between the ISRC and the atmosphere. Notably, the mode switching temperature shifts to ∼28 °C in the cooling process, because the water molecules’ directional transportation from the hydrophilic SiO₂ fiber to the swelling hydrogel is more difficult than transportation from the shrinking hydrogel to the hydrophilic SiO₂ fiber. In contrast, the solar reflectivity barely changes with the temperature (fluctuating around ∼0.90), sustainably blocking a large amount of solar irradiation for daytime cooling. Moreover, to account for its stability, 200 times heating–cooling cycles were conducted using a home-made thermoelectric system for a long-time operation test. As shown in Fig. 3k, the attenuations for solar reflectivity and infrared emissivity in different modes are minor, both less than 0.02 (Table S2, ESI†).

Self-adaptive radiative cooling of the ISRC

To experimentally investigate the self-adaptive radiative cooling performance of the ISRC under sub- and above-ambient conditions, we performed outdoor cooling tests with on–off external heating, imitating the fluctuating cooling condition in the real world. As shown in Fig. 4a and b, the setup consists of a foam box covered with Al foil to insulate the sample from the ambient heat conduction, convection, and incident solar irradiation. The polyimide heater was controlled by a direct current power.

Fig. 4 Self-adaptive radiative cooling performance with the ISRC. Photo (a) and schematic (b) of the setup used to measure radiative cooling performance. Experimental result (c) and weather profile (d) for the above-ambient cooling experiment with external heating, compared to the selective radiative cooler, i.e., SiO₂ fiber. Experimental result (e) and weather profile (f) for the sub-ambient cooling experiment without external heating, compared to the broadband radiative cooler, i.e., black paint.

In the above-ambient cooling experiment, the ISRC was compared with the selective cooler, i.e., single SiO₂ fiber, which can achieve impressive sub-ambient cooling performance (see Fig. S14, ESI†). Particularly, the single SiO₂ fiber for the control group was overlaid on a successive stack of the same PE film, PNIPAM hydrogel, and PDMS tray as those of the ISRC, to eliminate the influence of the thermal resistance. The experiment was conducted in a dry and sunny day (23 August, 2023) in Yinchuan, China (Fig. 4c and d). At the beginning of the test, the temperature of the ISRC was close to that of the SiO₂ fiber, with a selective mode. When heated above the T_amb (∼26 °C) with a power of 0.8 W at 9:00, the spectrum of the ISRC automatically switched to the broadband mode, and it maintained a stable temperature reduction of ∼7.9 °C compared to the SiO₂ fiber. As heating power increased, the cooling performance became progressively significant, achieving a maximum temperature reduction of ∼12.4 °C with a heating power of 1.4 W. This impressive above-ambient cooling performance guarantee the applications in high-temperature cooling objects such as photovoltaics and electronic equipment.

In the sub-ambient cooling experiment, the ISRC was compared with a commercial black paint, which has a near black-body infrared spectrum (Fig. S15, ESI†). Without external heating, the ISRC was maintained the selective mode. As shown in Fig. 4e, the ISRC achieved a sub-ambient temperature reduction of 6.0–7.8 °C at clear night (Fig. 4f) even without wind cover. Contrasted with the black paint, the ISRC also revealed an improved temperature reduction of ∼4.1 °C. To demonstrate the overall cooling performance, daytime sub-ambient cooling and nighttime above-ambient cooling control experiments (Fig. S16, ESI†), and theoretical calculations were also performed (Fig. S17 and S18, ESI†).

Simultaneously maximizing photovoltaic and thermoelectric power generation with the ISRC

Developing materials, devices, and systems that enable round-the-clock electricity generation using renewable energy sources is crucial for the electrification of off-grid areas. Radiative cooling of photovoltaics (PVs) can improve generating efficiency during the day, and radiative cooling for the cold side of the thermoelectric generator (TEG) can produce power at night. However, PV cooling falls under above-ambient conditions, and the cold side of TEG acquires a larger sub-ambient temperature reduction. In this work, we maximized daytime photovoltaic and nighttime thermoelectric power simultaneously with the ISRC by developing a PV–TEG hybrid power system. During the day, the ISRC in the broadband mode dissipates PV parasitic heat efficiently by virtue of water circulation (Fig. 5a). At night, the ISRC in the selective mode cools the TEG's cold side to a sub-ambient temperature, enlarging the temperature difference between the hot/cold sides (Fig. 5b).

Fig. 5 Simultaneous maximization of photovoltaic and thermoelectric power generation with the ISRC. Heat flow between the radiative cooling subassembly and PV subassembly (a) and TEG subassembly (b). The arrow color transitioning from red to blue signifies a change in temperature from high to low. (c) Schematic and photos of the PV–TEG hybrid power system. The measured power and improved generating efficiency of PV (d) and TEG (f). (e) PV temperature reduction. (g) Temperature difference for the TEG sides. (h) Round-the-clock electricity generation performance of the PV–TEG hybrid power system compared to the reported materials and devices in the state of the art (ref. 38–47).

Fig. 5c shows the schematic of the PV–TEG hybrid power system. Specifically, the radiative cooling subassembly consists of a radiative cooler and a coiler with a size of 4 cm × 4 cm. The PV subassembly with the same dimension is composed of a commercial polycrystalline silicon PV cell and a coiler, which is indirectly connected with the cooling subassembly by a water pipe. The TEG subassembly constitutes a commercial Bi₂Te₃ TEG with fins, directly attached to the cooling subassembly. For daytime PV operating, the cooling water cycle is pumped with an extremely low water flow of ∼0.01 l min⁻¹. At night when the pump stops, the TEG obtains cold from the cooling subassembly while extracting heat from the environment. We built three PV–TEG hybrid power systems for a comparative study, respectively, the ISRC module, the selective module, and the commercial module. The selective module has an identical setup to the ISRC module, in addition to replacing the ISRC with the selective SiO₂ fiber. The commercial module simulates the power generation performance of a typical commercial PV and TEG. As a result, it lacks cooling water circulation and employs black paint as the cold side for the TEG.

We conducted a comparative study on 23 August 2023. Fig. 5d shows the measured PV power (P_P) at noon time (10:00–14:00) and the improved relative generating efficiency (η_P) of the ISRC module. The ISRC module achieved a maximum P_P value of ∼150 W m⁻² and an average η_P value of ∼12.0% compared to the commercial module. This creditable power enhancement profited from the radiative cooling water, which realized an unprecedented maximum temperature reduction of 23.9 °C and an average temperature reduction of 18.3 °C (Fig. 5e). Due to the extra radiation channel out of the atmospheric window, the ISRC module also performed an improved temperature reduction of ∼50%, and an η_P value of ∼3.1% compared to the selective module. Fig. 5f shows the measured TEG power (P_T) at night (21:00–00:00) and the improved relative generating efficiency (η_T) of the ISRC module. The ISRC module generated a P_T value of 18.8–35.2 mW m⁻², maintaining a similar fluctuation trend to the selective module. Compared to the commercial module, the temperature difference of the TEG hot/cold sides increased 32% (Fig. 5g), and the η_T value reached ∼80%. The developed PV–TEG hybrid power system presents significant advancements compared to materials/devices in the state of the art that utilize renewable energy for round-the-clock electricity generation (Fig. 5h), including thermoelectrics,^38–43 thermochemicals,⁴⁴ and moisture-electrics.^45–47 The simultaneous maximization of the PV and TEG power demonstrated the efficient use of self-adaptive radiative cooling under both sub-ambient and above-ambient conditions.

Discussion

By 2020, 773 million people globally have no access to electricity, and 670 million people would still lack access to electricity in 2030 according to the forecast of the International Energy Agency.⁴⁸ The annual electrification growth rate in underdeveloped regions like Sub-Saharan Africa is less than 1%, which raises a great concern. Maximizing the PV and TEG power simultaneously with the ISRC could potentially promote the global electrification process and increase the proportion of renewable energy in the power mix. We simulated the global annual increase of PV and TEG energy of the ISRC module compared to the commercial module based on the typical meteorological year weather files⁴⁹ (Fig. 6). The mathematic models for the PV and TEG (Fig. S19 and S20) are given in the ESI,† and were verified with the measured date (Fig. S21 and S22, ESI†). Due to the typically abundant solar resources and tropical climates in underdeveloped electrification regions such as Africa, Central and Southern Asia, the ISRC contributes significantly to the enhancement of power generation performance (Fig. 6a and c, Fig. S23 and S24, ESI†). We also distinguished the energy improvement based on 12 climate zones⁵⁰ (Table S3, ESI†) in Fig. 6b and d. The zones with larger enhancements in PV energy also experience proportionately larger increasing ratios. The increased PV energy in the tropical zone with very high solar radiation (e.g., Bafata) can even reach 20 kW h m⁻² (η_P of 11%), while the TEG energy gains maintain over 60% across all climate zones. The maximum increased TEG energy accesses 40 W h m⁻² in dry and hot regions, and the maximum increased ratio is close to 100% in humid and hot regions. The underlying mechanism of this climatic effect essentially stems from the higher atmospheric infrared transmittance in regions with lower precipitable water.² Additionally, we conducted an economic analysis to show the system's applicability (Fig. S25 and S26, ESI†).

Fig. 6 Global annual increase of PV and TEG energy with the ISRC. Increased PV energy (a) and TEG energy ratio (c) vs. the commercial module. Increased energy and ratio in 12 climate zones for PVs (b) and TEG (d). The 12 climate zones are derived from the Köppen–Geiger–photovoltaic climate classification. The first letter indicates the temperature-precipitation zones: A, tropical; B, desert; C, steppe; D, temperate; E, cold; and F, polar. The second letter indicates the irradiation zones: K, very high; H, high; M, medium; and L, low irradiation.

It should be emphasized that the application of the ISRC is not limited solely to the PV–TEG hybrid power system. Through thoughtful design, it can be applied to versatile cooling systems, such as achieving nighttime sub-ambient cooling for cooling energy storage and daytime above-ambient cooling for heat dissipation. By refining material anti-aging design and manufacturing processes, it is possible to establish large-scale cooling systems in the future.

Conclusions

In summary, we proposed a concept of self-adaptive radiative cooling with infrared spectral selective regulation. For the first time, an infrared self-adaptive radiative cooler (ISRC) was developed, simultaneously processing a selective infrared emissivity of ∼0.85 and a broadband emissivity of ∼0.92. Under above- and sub-ambient conditions, improved temperature reductions of ∼12.4 °C and ∼4.1 °C respectively were demonstrated compared to the ‘static’ spectrum radiative coolers. Furthermore, an improved daytime PV efficiency of 12% and an improved nighttime TEG efficiency of 80% were demonstrated simultaneously. In off-grid regions, a PV–TEG hybrid system with the ISRC can increase the daytime electricity production by an additional 20 kW h m⁻² while achieving a doubling of thermoelectric power at night. This work paves the way for the comprehensive utilization of radiative cooling under both sub- and above-ambient conditions in real-world applications.

Methods

Preparation of the ISRC

A PDMS tray was prepared by mixing an elastomer base and a curing agent in a weight ratio of 10/1 in a negative glass mould. The curing action was then performed in an oven at 60 °C for 3 hours. After curing completely, the PDMS tray was removed from the mould. To chemically crosslink the PDMS tray with the PNIPAM hydrogel layer, the internal surface of the former was modified in advance by submerging it in a benzophenone solution (20 wt% in acetone) for 15 min. Subsequently, the PDMS tray was rinsed with methanol and dried in a vacuum chamber.

The PNIPAM hydrogel layer was synthesized by free-radical polymerization with a NIPAM monomer, a BIS cross-linker, an APS initiator, and a TEMED accelerator in the prepared PDMS tray. First, 3.39 g of NIPAM monomer and 49 mg of BIS were dissolved in DI water to make a 30 ml homogeneous aqueous solution. The solution was then purged with N₂ for 20 min and precooled below 10 °C. Subsequently, 180 μL APS initiator and 80 μL TEMED accelerator (10 wt% in DI water) were added to the monomer solution. After mixing uniformly, the solution was poured into the PDMS tray and placed in an ice-water bath to facilitate the polymerization at low temperatures. Launched with UV irradiation for 4 hours (40-W lamp, 365-nm wavelength), the polymerization was over and the hydrogel layer was grafted onto the PDMS tray.

A microporous SiO₂ fiber layer was fabricated by electrostatic spinning technology. First, an electrospinning precursor solution was prepared using methyltrimethoxysilane, a polyvinyl alcohol solution (10 wt% in DI water), and phosphoric acid in a mass ratio of 1/30/0.004. Second, the solution was stirred for 2 hours under vacuum to disperse uniformly. Then, the precursor solution was electrospun at a voltage of 20 kV, a spinning distance of 18 cm, and a feeding rate of 4 ml h⁻¹, using a 19-gauge needle tip. Finally, the prepared SiO₂ fibers were calcined in a muffle furnace at 800 °C for 2 h to remove the residual organic components.

Optical characterizations

The spectrum properties of the developed ISRC were characterized using an UV-Vis-NIR spectrometer (Lambda 950, PerkinElmer) at 0.3–2.5 μm and a Fourier transform infrared spectrometer (Nicolet IS50, ThermoFisher) at 2.5–20 μm.

Radiative cooling performance calculation

The theoretical net radiative cooling power P_cool(T_c) for the cooler with the temperature T_c can be calculated based on the first law of thermodynamics as follows:

P_cool(T_c) = P_rad + P_atm − P_solar − P_cond+conv (1)

where P_rad and P_atm refer to the net IR radiative heat exchange within and outside the atmospheric window, respectively. P_solar is the absorbed solar irradiation during the day, and P_cond+conv represents the conductive and convective heat transfer with surroundings. When P_cool(T_c) equals zero, the cooler gets the maximum temperature reduction.

Author contributions

H. T. and D. Z. proposed the concept, developed the ISRC, and designed the experiments. H. T. performed the simulations and theoretical calculations. H. T., C. G., and F. F. conducted the outdoor tests. H. T., C. G., H. P., and Q. X. performed the characterizations. H. T. and D. Z. prepared the manuscript and contributed to the interpretation of the results.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by National Natural Science Foundation of China under grant number 52276178. The authors gratefully acknowledge technical discussions with Prof. Ruzhu Wang at Shanghai Jiao Tong University. The authors are also thankful to anonymous reviewers for their valuable comments and feedback.

References

1. D. Zhao, A. Aili, Y. Zhai, S. Xu, G. Tan, X. Yin and R. Yang, Appl. Phys. Rev., 2019, 6, 021306.
2. D. Zhao, A. Aili, Y. Zhai, J. Lu, D. Kidd, G. Tan, X. Yin and R. Yang, Joule, 2019, 3, 111–123.
3. E. A. Goldstein, A. P. Raman and S. Fan, Nat. Energy, 2017, 2, 17143.
4. A. P. Raman, W. Li and S. Fan, Joule, 2019, 3, 2679–2686.
5. W. Ren, Y. Sun, D. Zhao, A. Aili, S. Zhang, C. Shi, J. Zhang, H. Geng, J. Zhang, L. Zhang, J. Xiao and R. Yang, Sci. Adv., 2021, 7, eabe0586.
6. M. Zhou, H. Song, X. Xu, A. Shahsafi, Y. Qu, Z. Xia, Z. Ma, M. A. Kats, J. Zhu, B. S. Ooi, Q. Gan and Z. Yu, Proc. Natl. Acad. Sci. U. S. A., 2021, 118, e2019292118.
7. I. Haechler, H. Park, G. Schnoering, T. Gulich, M. Rohner, A. Tripathy, A. Milionis, T. M. Schutzius and D. Poulikakos, Sci. Adv., 2021, 7, eabf3978.
8. X. Huang, J. Mandal, X. Huang, J. Mandal, J. Xu and A. P. Raman, Joule, 2022, 6, 2762–2775.
9. X. Gong, M. Ding, P. Gao, Y. Ji, X. Wang, X. Y. Liu, J. Yu, S. Zhang and B. Ding, Nano Lett., 2023, 23, 11337–11344.
10. B. Gu, Q. Xu, H. Wang, H. Pan and D. Zhao, ACS Nano, 2023, 17, 18308–18317.
11. A. K. Head, Australian Pat., 239364, 1959.
12. S. Catalanotti, V. Cuomo, G. Piro, D. Ruggi, V. Silvestrini and G. Troise, Sol. Energy, 1975, 17, 83–89.
13. C. G. Granqvist and A. Hjortsberg, J. Appl. Phys., 1981, 52, 4205–4220.
14. T. S. Eriksson and C. G. Granqvist, Appl. Opt., 1983, 22, 3204–3206.
15. A. P. Raman, M. A. Anoma, L. Zhu, E. Rephaeli and S. Fan, Nature, 2014, 515, 540–544.
16. Y. Zhai, Y. Ma, S. N. David, D. Zhao, R. Lou, G. Tan, R. Yang and X. Yin, Science, 2017, 355, 1062–1066.
17. X. Zhao, T. Li, H. Xie, H. Liu, L. Wang, Y. Qu, S. C. Li, S. Liu, A. H. Brozena, Z. Yu, J. Srebric and L. Hu, Science, 2023, 382, 684–691.
18. D. Zhao and H. Tang, Science, 2023, 382, 644–645.
19. J. Mandal, Y. Fu, A. C. Overvig, M. Jia, K. Sun, N. N. Shi, H. Zhou, X. Xiao, N. Yu and Y. Yang, Science, 2018, 362, 315–319.
20. T. Li, Y. Zhai, S. He, W. Gan, Z. Wei, M. Heidarinejad, D. Dalgo, R. Mi, X. Zhao, J. Song, J. Dai, C. Chen, A. Aili, A. Vellore, A. Martini, R. Yang, J. Srebric, X. Yin and L. Hu, Science, 2019, 364, 760–763.
21. P. Poredoš, H. Shan, C. Wang, Z. Chen, Z. Shao, F. Deng, H. Liu, J. Yu and R. Wang, Energy Environ. Sci., 2024, 17, 2336–2355.
22. A. Aili, D. Zhao, G. Tan, X. Yin and R. Yang, Appl. Energy, 2021, 302, 117515.
23. Z. Cui, C. Guo and D. Zhao, Build. Simul., 2022, 15, 1775–1787.
24. H. Tang, Z. Zhou, S. Jiao, Y. Zhang, S. Li, D. Zhang and J. Zhang, Sol. Energy Mater. Sol. Cells, 2022, 235, 111498.
25. L. Zhu, A. Raman, K. X. Wang, M. A. Anoma and S. Fan, Optica, 2014, 1, 32.
26. Y. Zhou, S. Wang, J. Peng, Y. Tan, C. Li, F. Y. C. Boey and Y. Long, Joule, 2020, 4, 2458–2474.
27. X. H. Li, C. Liu, S. P. Feng and N. X. Fang, Joule, 2019, 3, 290–302.
28. Y. Cui, Y. Ke, C. Liu, Z. Chen, N. Wang, L. Zhang, Y. Zhou, S. Wang, Y. Gao and Y. Long, Joule, 2018, 2, 1707–1746.
29. J. Wang, G. Tan, R. Yang and D. Zhao, Cell Rep. Phys. Sci., 2022, 3, 101198.
30. S. Wang, T. Jiang, Y. Meng, R. Yang, G. Tan and Y. Long, Science, 2021, 374, 1501–1504.
31. K. Tang, K. Dong, J. Li, M. P. Gordon, F. G. Reichertz, J. J. Urban, J. Yao, R. Levinson and J. Wu, Science, 2021, 374, 1504–1509.
32. L. Tang, L. Wang, X. Yang, Y. Feng, Y. Li and W. Feng, Prog. Mater. Sci., 2021, 115, 100702.
33. C. Lin, J. Hur, C. Y. H. Chao, G. Liu and S. Yao, Sci. Adv., 2022, 8, eabn7359.
34. H. Shan, P. Poredoš, Z. Ye, H. Qu, Y. Zhang, M. Zhou, R. Wang and S. C. Tan, Adv. Mater., 2023, 35, 2302038.
35. R. Y. O. Yoshida, T. Okano, Y. Sakurai and K. Sakai, J. Biomater. Sci., Polym. Ed., 1995, 6, 585–598.
36. H. Tang, C. Guo, Q. Xu and D. Zhao, Micromachines, 2023, 14, 219.
37. H. Tang, S. Li, Y. Zhang, Y. Na, C. Sun and D. Zhao, J. Cleaner Prod., 2022, 380, 135035.
38. K. Gao, J. Yang, H. Shen, Y. Liu, Y. Li and M. Zhang, Sol. RRL, 2022, 6, 2100975.
39. L. Yu, Z. Xi, S. Li, D. Pang, H. Yan and M. Chen, Appl. Energy, 2022, 322, 119403.
40. Y. Tian, X. Liu, F. Chen and Y. Zheng, Sci. Rep., 2020, 10, 20903.
41. J. Liu, D. Li, W. Ma, Y. Chen, C. Dou, D. Meng, Q. He, X. Li, X. Deng and H. Cai, Opt. Express, 2023, 31, 14495.
42. A. M. Alajlan, A. A. Almethen and H. Qasem, Appl. Phys. Lett., 2022, 121, 073903.
43. Y. Liu, S. Hou, X. Wang, L. Yin, Z. Wu, X. Wang, J. Mao, J. Sui, X. Liu, Q. Zhang, Z. Liu and F. Cao, Small, 2022, 18, 2106875.
44. H. Zhang, Z. Wang, H. Li, M. Salla, Y. Song, S. Huang, S. Huang, X. Wang, K. Liu, G. Xu, J. Huang, C.-W. Qiu and Q. Wang, Joule, 2023, 7, 1515–1528.
45. S. Yang, X. Tao, W. Chen, J. Mao, H. Luo, S. Lin, L. Zhang and J. Hao, Adv. Mater., 2022, 34, 2200693.
46. J. Bai, Y. Huang, H. Wang, T. Guang, Q. Liao, H. Cheng, S. Deng, Q. Li, Z. Shuai and L. Qu, Adv. Mater., 2022, 34, 2103897.
47. H. Wang, Y. Sun, T. He, Y. Huang, H. Cheng, C. Li, D. Xie, P. Yang, Y. Zhang and L. Qu, Nat. Nanotechnol., 2021, 16, 811–819.
48. UNSD, WHO, IEA, IRENA and World Bank, The Energy Progress Report 2022, 2023.
49. L. K. Lawrie and D. B. Crawley, Development of Global Typical Meteorological Years (TMYx), 2020. Available at https://climate.onebuilding.org/.
50. J. Ascencio-Vásquez, K. Brecl and M. Topič, Sol. Energy, 2019, 191, 672–685.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ee04261h

---