A radiative cooling structural material

See allHide authors and affiliations

Science  24 May 2019:
Vol. 364, Issue 6442, pp. 760-763
DOI: 10.1126/science.aau9101

A stronger, cooler wood

One good way to reduce the amount of cooling a building needs is to make sure it reflects away infrared radiation. Passive radiative cooling materials are engineered to do this extremely well. Li et al. engineered a wood through delignification and re-pressing to create a mechanically strong material that also cools passively. They modeled the cooling savings of their wood for 16 different U.S. cities, which suggested savings between 20 and 50%. Cooling wood would be of particular value in hot and dry climates.

Science, this issue p. 760


Reducing human reliance on energy-inefficient cooling methods such as air conditioning would have a large impact on the global energy landscape. By a process of complete delignification and densification of wood, we developed a structural material with a mechanical strength of 404.3 megapascals, more than eight times that of natural wood. The cellulose nanofibers in our engineered material backscatter solar radiation and emit strongly in mid-infrared wavelengths, resulting in continuous subambient cooling during both day and night. We model the potential impact of our cooling wood and find energy savings between 20 and 60%, which is most pronounced in hot and dry climates.

Buildings account for more than 40% of the total energy demand and 70% of electricity use in the United States, leading to an annual national energy bill of more than $430 billion. Heating and cooling accounts for ~48% of this energy use, making it the largest individual energy expense (1). In general, cooling is more challenging than heating because of the second law of thermodynamics (2). As a result, passive radiative cooling has become attractive for improving building energy efficiencies by providing a perpetual path to dissipate heat from these structures through the atmospheric transparent window into the ultracold universe with zero energy consumption. Nocturnal radiative cooling has been investigated on pigmented paints, dielectric coating layers, metallized polymer films, and even organic gases because of their intrinsic thermal emission properties (26). Daytime radiative cooling is more challenging, as natural high-infrared emissive materials also tend to absorb visible wavelengths, though advances include using precision-designed nanostructures (7, 8) or hybrid optical metamaterials (9) to tailor material spectrum responses for continuous cooling. However, it remains a challenge to both manufacture and apply these structures at the size and scale required for construction purposes.

Wood has been used for thousands of years and has emerged as an important sustainable building material to potentially replace steel and concrete because of its economic and environmental advantages (10). We engineered wood by complete delignification followed by mechanical pressing to render a structural material (Fig. 1, A and B) with daytime subambient cooling effects (figs. S1 to S8). We used scanning electron microscopy (SEM) to show that the wood exhibits multiscale cellulose fibers or fiber bundles (Fig. 1C and figs. S9 to S11). Our cooling wood is composed of cellulose nanofibers partially aligned in the tree’s growth direction (Fig. 1D and fig. S11); these fibers are nonabsorbing in the visible range (figs. S12 to S15). The multiscale fibers and channels (fig. S16) function as randomized and disordered scattering elements for an intense broadband reflection at all visible wavelengths (Fig. 1E and figs. S17 and S18). Meanwhile, the molecular vibration and stretching of cellulose in cooling wood facilitate strong emission in the infrared (Fig. 1F). The heat flux emitted by the cooling wood exceeds the absorbed solar irradiance, resulting in passive subambient radiative cooling for both day and night. The delignified and mechanically pressed wood also delivers mechanical strength and toughness that are, respectively, ~8.7 and 10.1 times the strength and toughness of natural wood. These findings establish cooling wood as a multifunctional structural material that may provide a path for improving the energy efficiency of buildings.

Fig. 1 Cooling wood demonstrates passive daytime radiative cooling.

Photos of a board of (A) natural wood and (B) cooling wood. (C) SEM image of the cooling wood showing the aligned wood channels. (D) SEM image of partially aligned cellulose nanofibers of the cooling wood. (E) Schematic showing the wood structure strongly scattering solar irradiance. (F) Schematic of infrared emission by molecular vibration of the cellulose functional groups. (G) Setup of the real-time measurement of the subambient cooling performance of the cooling wood.

The largely disordered mesoporous cellulose structures render the cooling wood extremely hazy. A reflective, hazy surface can effectively scatter incident light in a hemispherical solid angle, which is particularly desirable for building applications to avoid visual discomfort caused by strong specularly reflected light (11). We show the reflection haze spectra of the cooling wood with an incident angle of 8°, demonstrating that the material has an extremely high reflection haze of 96% on average (fig. S17). The high, diffusive reflectance in the solar radiation range leads to the bright whiteness of the cooling wood (Fig. 2A) (12). The higher reflection when the incoming polarization direction is along the fiber alignment direction is attributable to the strong scattering (fig. S18). We investigated the emissivity spectra of the cooling wood in the infrared range from 5 to 25 μm, i.e., covering the spectroscopically important wavelength range for room-temperature blackbodies (Fig. 2B). The cooling wood exhibits high emissivity (close to unity) in the infrared range, emitting strongly at all angles and radiating a net heat flux through the atmospheric transparency window (8 to 13 μm) to the cold sink of outer space in the form of infrared radiation. Thus, the cooling wood is black in the infrared range, a marked difference from its appearance in the solar spectrum, where it is white (i.e., simultaneously displaying a lack of absorption and high reflectivity). The infrared emissivity spectrum response shows negligible angular dependence (from 0° to 60°). The average emissivity across the atmospheric window is also greater than 0.9 for emission angles between ±60° (Fig. 2C), indicating a stable emitted heat flux when the cooling wood is aimed at different angles in relation to the sky, as it would be in practical applications. Figure S19 shows the Fourier transform infrared absorbance of the cooling wood. The strong emission from 8 to 13 μm is mainly contributed by the complex infrared emission of OH association and C–H, C–O, and C–O–C stretching vibrations between 770 and 1250 cm−1 (11). The cellulose exhibits the strongest infrared absorbance by OH and C–O centered at ~1050 cm−1 (9 μm) (11), which coincidently lies in the atmospheric transparency window (13). The high emissivity across the rest of the infrared spectrum results in radiative heat exchange between the cooling wood and the atmosphere (such as in the second atmospheric window between 16 and 25 μm), which further increases the overall radiative cooling flux when the surface temperature is close to that of the ambient environment (14).

Fig. 2 Optical characterization and thermal measurement of cooling wood.

(A) Absorption of the natural and cooling wood in the solar spectrum. (B) Infrared emissivity spectra of the cooling wood between 5 and 25 μm at different emission angles. (C) Polar distribution of the average emissivity across the atmospheric window of the cooling wood. (D) Schematic of the thermal box used to characterize the radiative cooling power and cooling temperature. PE, polyethylene. (E) Twenty-four–hour continuous measurement of the 200 mm–by–200 mm cooling wood. (Top) Measurement in Box one: Direct measurement of the radiative cooling power of the cooling wood. The heater was on, and a feedback control program maintained the wood temperature at the same temperature as the ambient environment. At this condition, the heating power is the same as the total radiative cooling power because all other heat fluxes are zero because of the zero-temperature difference. (Middle) Measurement in Box two: Steady-state temperature of the cooling wood. (Bottom) Temperature difference between the ambient surroundings and the cooling wood.

We demonstrated the subambient radiative cooling performance of the cooling wood during both day and night over 24-hour continuous thermal measurement in Cave Creek, Arizona (33°49′32″ N, 112°1′44″ W; 585-m altitude). We tested two sets of cooling wood, 200 mm by 200 mm in size, in two thermal boxes in parallel to monitor the subambient radiative cooling temperature directly as well as the cooling power with the assistance of a feedback-controlled heating system (Fig. 2D) (9). We elevated the two thermal boxes 1.2 m over the sunlight-shaded ground to avoid heat conducted from the ground to the boxes and overestimation of the thermal couples for ambient temperature measurement (fig. S3B). We found that the cooling wood had radiative cooling powers of 63 and 16 W/m2 during the night and daytime (between 11 a.m. and 2 p.m.), respectively, leading to an average cooling power of 53 W/m2 over the 24-hour period. We measured the steady-state radiative cooling temperature of the cooling wood synchronously in the second box, in which the Kapton heater was turned off. The cooling wood exhibits a radiative cooling temperature below ambient during both night and daytime (Fig. 2E). The average below-ambient temperature was >9°C during the night and >4°C during midday (between 11 a.m. and 2 p.m.). Both the natural wood and the cooling wood exhibit similar thermal conductivities between their top and bottom surfaces (fig. S20), and these values are higher than that of thermal insulation wood (15) because of the densified structure created by mechanical pressing. We observed the scattered clouds during the measurement, which slightly reduced the net radiative cooling effects (16). In addition, we used fluorosilane treatment, which can be used to make the wood superhydrophobic with a water contact angle of ~150° (fig. S21) and further improves the weatherability and protects the cooling wood from water condensate.

The cooling wood is also mechanically stronger and tougher than natural wood because of the larger interaction area between exposed hydroxyl groups of the aligned cellulose nanofibers in the growth direction after lignin removal (Fig. 3A) (17). The cooling wood demonstrates a tensile strength as high as 404.3 MPa, which is ~8.7 times that of natural wood. An improved toughness of 3.7 MJ/m3 was also observed, which is 10.1 times that of natural wood (Fig. 3B). We observed a simultaneous enhancement in mechanical toughness (fig. S22), which is desirable in structural material design (1719). We attributed this to the energy dissipation enabled by repeated hydrogen-bond formation and/or breaking at the molecular scale in the delignified and mechanically pressed material.

Fig. 3 Cooling wood as a multifunctional structural material.

(A) Schematics showing the origin of the high mechanical strength from the molecular bonding of the aligned cellulose nanofibers. The (B) tensile strength and (C) specific ultimate strength of the cooling wood are compared with those of natural wood and some common metals and alloy (2123). (D and E) Scratch-hardness characterization of the natural wood and the cooling wood in three different directions. A, B, and C denote directions parallel, perpendicular, and at a 45° angle to the tree growth direction, respectively. (F) Performance comparison of cooling wood and natural wood. Error bars in (C) and (D) indicate measurement variations among the samples.

The ratio of mechanical strength to weight is a critical parameter in buildings, especially because of cost considerations (20). The specific tensile strength of the cooling wood reaches up to 334.2 MPa cm3/g (Fig. 3C), surpassing that of most structural materials, including Fe–Mn–Al–C steel, magnesium, aluminum alloys, and titanium alloys (2123). The mechanical scratch hardness of the cooling wood also shows great improvement compared with that of the untreated natural wood. As characterized by a linear reciprocating tribometer (fig. S23), the scratch hardness of the cooling wood reaches up to 175.0 MPa in direction C, which is 8.4 times that of natural wood (Fig. 3, D and E). Compared with natural wood, the scratch hardness of the cooling wood also increased by a factor of 5.7 and 6.5 in directions A and B, respectively. The flexural strength of cooling wood is ~3.3 times as high as that of natural wood (fig. S24, A to C). The axial compressive strength of the cooling wood is also much higher than that of natural wood. The cooling wood shows a high axial compressive strength of 96.9 MPa, which is 3.2 times as high as that of natural wood (fig. S24, D to F). Cooling wood also exhibits a toughness that is 5.7 times as high as that of natural wood (fig. S24, G and H).

The cooling wood is superior to natural wood for building efficiency applications in terms of continuous cooling capability and mechanical strength (Fig. 3F). The properties of cooling wood, including continuous subambient cooling, high mechanical strength, bulk structure, low density, sustainability, and bulk fabrication process, make it attractive as a structural material when compared with other radiative cooling materials (79, 2427). Raman et al. (7) demonstrated a photonic approach to meet the stringent demands of high thermal emission in the mid-infrared and strong solar reflection using seven alternating layers of HfO2 and SiO2 of varying thicknesses. However, the material is difficult to execute at the scale required for buildings. Another metamaterial thin film was demonstrated to have the potential for scalable manufacturing (9) but cannot be used as a structural component. The influence on radiative cooling performance from local weather conditions, including wind speed, precipitable water, and cloud cover, has been investigated on large-scale radiative cooling metamaterial and systems (16). Durability for long-term outdoor applications must be considered if the cooling wood is to be utilized as a structural material on the external surfaces of buildings in the future. Surface treatment methods could improve the resistivity of the cooling wood against water (28), fire (29), ultraviolet exposure (30), and biological factors (31) to satisfy the need for long-term outdoor durability.

The combination of the visible white (i.e., high solar reflectance) and infrared black (i.e., high infrared emissivity) properties of the cooling wood leads to a highly efficient radiative cooling material (Fig. 4, A and B). The mechanical strength also allows the cooling wood to be used as both roof and siding material without other mechanical support. We used EnergyPlus version 8 and the parameters listed in table S1 to model the potential energy savings of using cooling wood on exterior surfaces (wall siding and roofing membranes) of buildings. Our energy model accounts for a total heat balance on both the internal and external building enclosure surfaces, the heat transfer through the building enclosures, and heat sources and sinks, such as internal loads generated by equipment, occupants, and lighting. This modeling is governed by energy-balance equations for both the outside and inside surfaces of the building, as shown in table S2, which are solved simultaneously. To determine an annual rate of energy consumption, we solved the governing equations iteratively with an hourly time step over a year. The internal boundary conditions used an indoor air temperature set point of 24°C, and the external boundary conditions used hourly weather data for a typical meteorological year (32). These models use ray tracing for all components of radiative heat transfer, including direct and indirect fluxes, and fluxes reflected from both the ground and surrounding building surfaces.

Fig. 4 Modeling energy savings by installing cooling-wood panels on roofing and external siding of midrise apartment buildings.

(A) When used as a building material, the cooling wood exhibits high solar reflectance and high infrared emissivity. (B) Photo of a 5-cm-thick piece of cooling wood. (C) Total cooling energy savings per year and (D) percentage among all 16 cities. (E) Average cooling energy savings and percentage among all 16 cities. (F) Total predicted cooling energy savings of midrise buildings extended for all U.S. cities based on local climate zones.

The building models that we used in this study are midrise apartment buildings across the United States, based on data from old (built before 1980) and new (built after 2004) structures provided by the U.S. Department of Energy Commercial Reference Buildings database (33). This building type is the most suitable among the reference buildings because of the importance of weather-related loads on the total building energy consumption (34). The energy modeling process established a baseline energy-consumption pattern for these old and new buildings and then modified the wall siding and roof membrane material properties on the basis of the cooling-wood performance to predict an energy-consumption pattern (figs. S25 and S26).

Sixteen cities in the United States were selected for this study: Albuquerque (NM), Atlanta (GA), Austin (TX), Boulder (CO), Chicago (IL), Duluth (MN), Fairbanks (AK), Helena (MT), Honolulu (HI), Las Vegas (NV), Los Angeles (CA), Minneapolis (MN), New York City (NY), Phoenix (AZ), San Francisco (CA), and Seattle (WA) (35). These cities are representative of all U.S. climate zones, allowing us to extend the results of this study to the entire country. The modified building models use cooling wood in place of common wood siding, which is a layer of the roofing and siding assembly, to determine the passive cooling power generated as a result of the local weather.

We determined the total cooling energy-saving patterns for the selected 16 cities and the percent savings relative to the baseline (Fig. 4, C and D). The midrise apartments built before 1980 and after 2004 are end members for assessing the energy savings, and buildings built in between will be between these two bounds. We found that an average of ~35% in cooling energy savings can be obtained for old midrise apartment buildings, and an average of ~20% can be obtained for new midrise apartments (Fig. 4E).

The energy savings from the installation of the cooling wood on the exterior surface of these buildings show that, on average for old and new midrise apartments, Austin (22.9 MJ/m2), Honolulu (28.2 MJ/m2), Las Vegas (21.1 MJ/m2), Atlanta (17.1 MJ/m2), and Phoenix (32.1 MJ/m2) would have the highest energy savings among the selected 16 cities. Phoenix had the highest potential cooling savings because of its hot and dry climate. Therefore, cities in the Southwest may be the most suitable for the installation of this material to reduce energy consumption for cooling. However, if the cooling wood remains exposed during the winter months, the heating energy cost would subsequently increase. The offset of the increased heating energy costs and a more detailed analysis of the overall energy savings can be found in fig. S26. We predicted the cooling energy savings of midrise buildings extended for all U.S. cities on the basis of local climate zones. The results show that cities with hot and dry climates have the largest potential cooling energy savings. The energy-savings effect of cooling wood has the potential to relax the energy load associated with conditioning indoor spaces that accounts for 31% of the total building primary energy consumption (36). We also evaluated the effect of neighboring structures on the energy performance (figs. S27 to S30). Surrounding buildings decrease the cooling energy demand of the building covered with cooling wood because of the shading that the surrounding structures provide. Therefore, the potential cooling energy savings obtained by using cooling wood changes, on average, from 35% for an isolated building to 51% for the highest urban density in pre-1980 buildings and changes from 21 to 39% for post-2004 buildings.

We developed a multifunctional, passive radiative cooling material composed of wood that can be fabricated by using a scalable bulk process to engineer its spectral response. The cooling wood exhibits superior whiteness, which originates from the low optical loss of the cellulose fibers and the material’s disordered photonic structure. The energy emitted within the infrared range of the cooling wood overwhelms the amount of solar energy received. We confirmed this cooling effect by real-time temperature measurements of natural and cooling-wood samples, in which the materials were exposed to the sky. Additionally, cooling wood is 8.7 times as strong as and 10.1 times as tough as natural wood. The intrinsic lightweight nature of the cooling wood has a specific strength three times that of widely used Fe–Mn–Al–C structural steel. This multifunctional, scalable cooling-wood material holds promise for future energy-efficient and sustainable building applications, enabling a substantial reduction in carbon emission and energy consumption.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S30

Tables S1 and S2

References (3739)

References and Notes

Acknowledgments: Funding: This project is not directly funded. L.H. and T.L. acknowledge the support of the A. James & Alice B. Clark Foundation and the A. James School of Engineering at the University of Maryland. X.Y. acknowledges the support of the Gordon and Betty Moore Foundation. Author contributions: T.L., Y.Z., and S.H. contributed equally to this work. L.H., T.L., Y.Z., S.H., and X.Y. designed the experiments. T.L., S.H., W.G., R.M., J.So., J.D., C.C., A.V., and A.M. performed the material preparation and characterization as well as mechanical measurements and analysis. Y.Z., Z.W., X.Z., A.A., X.Y., and R.Y. contributed to the thermal and optical measurement and analysis. M.H., D.D., and J.Sr. performed the modeling for building efficiency. Y.Z., Z.W., and T.L. went to Arizona for field tests. L.H., T.L., Y.Z., and X.Y. collectively wrote the manuscript. Competing interests: L.H., T.L., and S.H. are the inventors on a patent currently pending at the international stage (WO 2019/055789; filed 14 September 2018). All the other authors declare that they have no competing interests. Data and materials availability: All data are available in the manuscript or the supplementary materials.

Stay Connected to Science

Navigate This Article