with fluctuations in concentration of less than
0.4% (fig. S1) (32). The hybrid metamaterial
films are translucent because of the scattering
of visible light from the microsphere inclusions
(fig. S2) (32). Additionally, when backed with a
200-nm-thick reflective silver coating, the hybrid
metamaterial has a balanced white color (fig. S2)
(32). The strongly scattering and nonspecular
optical response of the metamaterial will avoid
back-reflected glare, which can have detrimental visual effects for humans and interfere with
aircraft operations (33).
We demonstrated real-time, continuous radiative cooling by conducting thermal measurements using an 8-in-diameter, scalably fabricated
hybrid metamaterial film over a series of clear
autumn days in Cave Creek, Arizona (33°49′32′′N,
112°1′44′′W, 585 m altitude) (Fig. 4, B and C).
The metamaterial was placed in a foam container that prevents heat loss from below. The
top surface of the metamaterial faced the sky
and was directly exposed to the air (fig. S3) (32).
We kept the surface temperature of the meta-
material the same as the measured ambient
temperature using a feedback-controlled electric
heater placed in thermal contact with the meta-
material so as to minimize the impacts of con-
ductive and convective heat losses. The total
radiative cooling power is therefore the same
as the heating power generated by the electric
heater if there is no temperature difference be-
tween the surface and the ambient air. With the
feedback control, the surface temperature follows
the measured ambient temperature within ±0.2°C
(Fig. 4C). The average cooling power around noon
reaches 93 W/m2, with normal-incidence solar
irradiance greater than 900 W/m2. We observed
higher average nighttime radiative cooling than
during the day. However, the cooling power peaks
after sunrise and before sunset, when the am-
bient temperature is changing rapidly and solar
irradiance is incident at large oblique angles. To
further demonstrate the effectiveness of radia-
tive cooling, we also used water as a cold storage
medium and show cold water production with
the scalably fabricated hybrid metamaterial (fig.
S6) (32). Although we did not determine the re-
liability and lifetime of the glass-polymer hybrid
metamaterials for outdoor applications, applying
chemical additives and high-quality barrier coat-
ings may enhance their outdoor performance.
Many polymeric thin films are currently avail-
able and designed with extended outdoor life-
REFERENCES AND NOTES
1. S. Catalanotti et al., Sol. Energy 17, 83–89 (1975).
2. C. G. Granqvist, A. Hjortsberg, J. Appl. Phys. 52, 4205–4220
3. B. Orel, M. Klanjšek Gunde, A. Krainer, Sol. Energy 50, 477–482
4. A. R. Gentle, G. B. Smith, Adv. Sci. (Weinh.) 2, 1500119 (2015).
5. M. M. Hossain, M. Gu, Adv. Sci. (Weinh.) 3, 1500360
6. E. Rephaeli, A. Raman, S. Fan, Nano Lett. 13, 1457–1461
7. A. P. Raman, M. A. Anoma, L. Zhu, E. Rephaeli, S. Fan, Nature
515, 540–544 (2014).
8. M. F. Weber, C. A. Stover, L. R. Gilbert, T. J. Nevitt,
A. J. Ouderkirk, Science 287, 2451–2456 (2000).
9. S. D. Hart et al., Science 296, 510–513 (2002).
10. J. K. Gansel et al., Science 325, 1513–1515 (2009).
11. R. D. Rasberry et al., J. Mater. Chem. 21, 13902 (2011).
12. H. E. Türeci, L. Ge, S. Rotter, A. D. Stone, Science 320,
13. D. S. Wiersma, Nat. Phys. 4, 359–367 (2008).
14. S. Grésillon et al., Phys. Rev. Lett. 82, 4520–4523 (1999).
15. L. Sapienza et al., Science 327, 1352–1355 (2010).
16. M. Segev, Y. Silberberg, D. N. Christodoulides, Nat. Photonics 7,
17. E. Yablonovitch, J. Opt. Soc. Am. 72, 899 (1982).
18. H. A. Atwater, A. Polman, Nat. Mater. 9, 205–213 (2010).
19. B. J. Seo, T. Ueda, T. Itoh, H. Fetterman, Appl. Phys. Lett. 88,
20. P. Jung et al., Nat. Commun. 5, 3730 (2014).
21. X. Shen et al., Appl. Phys. Lett. 101, 154102 (2012).
22. J. Hao, É. Lheurette, L. Burgnies, É. Okada, D. Lippens,
Appl. Phys. Lett. 105, 081102 (2014).
23. E. D. Palik, Handbook of Optical Constants of Solids (Academic,
24. W. Liu et al., Opt. Express 22, 16178–16187 (2014).
25. N. Ocelic, R. Hillenbrand, Nat. Mater. 3, 606–609 (2004).
26. I. Balin, N. Dahan, V. Kleiner, E. Hasman, Appl. Phys. Lett. 94,
27. Y. Zhao, M. A. Belkin, A. Alù, Nat. Commun. 3, 870 (2012).
28. M. S. Wheeler, J. S. Aitchison, J. I. Chen, G. A. Ozin,
M. Mojahedi, Phys. Rev. B 79, 073103 (2009).
29. C. C. Katsidis, D. I. Siapkas, Appl. Opt. 41, 3978–3987 (2002).
30. L. F. Li, J. Opt. Soc. Am. A Opt. Image Sci. Vis. 13, 1870 (1996).
31. H. W. Yates, J. H. Taylor, “Infrared transmission of the
atmosphere,” no. NRL-5453 (Naval Research Lab, 1960).
32. Materials and methods are available as supplementary
33. X. Xu, K. Vignarooban, B. Xu, K. Hsu, A. M. Kannan, Renew.
Sustain. Energy Rev. 53, 1106–1131 (2016).
34. H. Price et al., J. Sol. Energy Eng. 124, 109 (2002).
SCIENCE sciencemag.org 10 MARCH 2017 • VOL 355 ISSUE 6329 1065
Ambient Temperature Surface Temperature
Fig. 4. Performance of scalable hybrid metamaterial for effective radiative cooling. (A) A
photo showing the 300-mm-wide hybrid metamaterial thin film that was produced in a roll-to-roll manner, at a speed of 5 m/min. The film is
50 mm in thickness and not yet coated with silver.
(B) A 72-hour continuous measurement of the ambient temperature (black) and the surface temperature (red) of an 8-in-diameter hybrid metamaterial
under direct thermal testing. A feedback-controlled
electric heater keeps the difference between ambient and metamaterial surface temperatures less
than 0.2°C over the consecutive 3 days. The heating
power generated by the electric heater offsets the
radiative cooling power from the hybrid metamaterial. When the metamaterial has the same temperature as the ambient air, the electric heating power
precisely measures the radiative cooling power of
the metamaterial. The shaded regions represent
nighttime hours. (C) The continuous measurement
of radiative cooling power over 3 days shows an
average cooling power of >110 W/m2 and a noontime
cooling power of 93 W/m2 between 11 a.m. and
2 p.m. The average nighttime cooling power is
higher than that of the daytime, and the cooling
power peaks after sunrise and before sunset. The
measurement error of the radiative cooling power
is well within 10 W/m2 (32).