Radiative Cooling New Opportunities & Enabling Technologies

Aaswath P. Raman, Ph.D. aaswath@stanford.edu

Research Associate, Fan Group, Ginzton Laboratory

1  ARPA-E Alternative Power Plant Cooling Workshop, May 12, 2014

An opportunity to tap an underutilized resource

Use the cold of outer space to radiatively pump heat from the ground through sky access

New: Possible at all hours of the day through photonic design of thermally emissive layers

Meaningful cooling power that scales with area: analogies to PV

Radiative Cooling Surface!

cold outer space!(-80°C ! -270°C)!

Atmosphere  Heat  Thermal    EM  Waves  

2

I. INTRODUCTION

Radiative cooling is a technique that exploits a natural transparency window for electro-

magnetic waves in the Earths atmosphere to transport heat from terrestrial objects into cold

space. As a result, objects with the appropriate radiative properties can passively cool them-

selves down to temperatures well below the ambient. The atmospheric transparency window

is found in the 8-13µm wavelength range, as shown in Fig. 1, and fortuitously overlaps with

the blackbody spectral radiance corresponding to typical terrestrial temperatures (0-50C),

thus enabling objects at these temperatures to emit more power than they absorb.

7 9 11 13 150

0.5

1

� [µm]

Atmospheric Transmission

0�C blackbody

50�C blackbody

FIG. 1. Atmopheric Transmission in the zenith direction vs. wavelength; normalized blackbody

spectral radiance of a 0�C and a 50�C blackbody emitter

Prior work in radiative cooling has almost entirely focused on nighttime cooling, where

one aims to maximize emission in the atmospheric transparency window, without having to

contend with solar radiation. In turn, nighttime cooling is of limited practical relevance since

solar radiation is by far the greatest source of daytime heating. Early work on nighttime

cooling mainly focused on choosing the right materials [3] and exploiting simple interference

e↵ects [4, 5] to maximize emissivity in the transparency window. Granqvist et. al [4, 5]

theoretically investigated and characterized the properties of the ideal nighttime radiator,

finding that such radiator could reach nighttime temperatures which were ⇡ 50oC lower

than the ambient, with a cooling power of ⇡ 100W/m2 when the radiator temperature was

2

Atmospheric transmittance

Radiative Cooling Surface!

cold outer space!(upper atmosphere)!

Blackbody spectrum of typical Earth temperature objects overlap with window

Radiative cooling is enabled by an atmospheric transparency window between 8 – 13 μm

Heat  

3

Varies with cloud cover, geographic location and ozone pollution

Power balance equation

Figure 1(a), represents a strong departure from previously published systems. Rather than designing

a cover filter which is spatially separated from a black emitter we introduce an integrated thermally

selective emitter atop a broadband mirror. Doing so enables us to exploit near-field coupling

between material layers, leading to stronger control over emission, absorption and reflection.

Using nanophotonic concepts, our photonic structure is able to strongly suppress solar absorption

while enhancing thermal emission in the atmospheric transparency window: an ultrabroadband

performance, shown in Fig. Figure 1(b), capable of achieving a net cooling power exceeding

100W/m2 at ambient temperature.

Radiative cooling devices operate at near-ambient temperatures and therefore do not suffer from

many of the difficulties associated with other thermal applications of photonic structures which

have typically involved high temperature operation. These difficulties include numerical uncertainty

associated with the temperature-dependence of optical properties, material cohesion, small-feature

evaporation, and durability that affect other thermal applications.30 Our approach and design is

thus a departure from previous work in using nano- and micro-photonic structures for thermal

applications.

To begin our analysis, we consider a photonic structure at temperature T whose radiative

properties are described by a spectral and angular emissivity e(l ,q). The structure is exposed to a

clear sky subject to solar irradiance, and also atmospheric irradiance corresponding to an ambient

temperature Tamb. The net cooling power Pnet(T ) of a structure with area A is given by

Pnet(T ) = Prad(T )�Patm(Tamb)�Psun, (1)

where

Prad(T ) = AZ

dWcosqZ •

0dl IBB(T,l )e(l ,q), (2)

3

Figure 1(a), represents a strong departure from previously published systems. Rather than designing

a cover filter which is spatially separated from a black emitter we introduce an integrated thermally

selective emitter atop a broadband mirror. Doing so enables us to exploit near-field coupling

between material layers, leading to stronger control over emission, absorption and reflection.

Using nanophotonic concepts, our photonic structure is able to strongly suppress solar absorption

while enhancing thermal emission in the atmospheric transparency window: an ultrabroadband

performance, shown in Fig. Figure 1(b), capable of achieving a net cooling power exceeding

100W/m2 at ambient temperature.

Radiative cooling devices operate at near-ambient temperatures and therefore do not suffer from

many of the difficulties associated with other thermal applications of photonic structures which

have typically involved high temperature operation. These difficulties include numerical uncertainty

associated with the temperature-dependence of optical properties, material cohesion, small-feature

evaporation, and durability that affect other thermal applications.30 Our approach and design is

thus a departure from previous work in using nano- and micro-photonic structures for thermal

applications.

To begin our analysis, we consider a photonic structure at temperature T whose radiative

properties are described by a spectral and angular emissivity e(l ,q). The structure is exposed to a

clear sky subject to solar irradiance, and also atmospheric irradiance corresponding to an ambient

temperature Tamb. The net cooling power Pnet(T ) of a structure with area A is given by

Pnet(T ) = Prad(T )�Patm(Tamb)�Psun, (1)

where

Prad(T ) = AZ

dWcosqZ •

0dl IBB(T,l )e(l ,q), (2)

3

Figure 1(a), represents a strong departure from previously published systems. Rather than designing

a cover filter which is spatially separated from a black emitter we introduce an integrated thermally

selective emitter atop a broadband mirror. Doing so enables us to exploit near-field coupling

between material layers, leading to stronger control over emission, absorption and reflection.

Using nanophotonic concepts, our photonic structure is able to strongly suppress solar absorption

while enhancing thermal emission in the atmospheric transparency window: an ultrabroadband

performance, shown in Fig. Figure 1(b), capable of achieving a net cooling power exceeding

100W/m2 at ambient temperature.

Radiative cooling devices operate at near-ambient temperatures and therefore do not suffer from

many of the difficulties associated with other thermal applications of photonic structures which

have typically involved high temperature operation. These difficulties include numerical uncertainty

associated with the temperature-dependence of optical properties, material cohesion, small-feature

evaporation, and durability that affect other thermal applications.30 Our approach and design is

thus a departure from previous work in using nano- and micro-photonic structures for thermal

applications.

To begin our analysis, we consider a photonic structure at temperature T whose radiative

properties are described by a spectral and angular emissivity e(l ,q). The structure is exposed to a

clear sky subject to solar irradiance, and also atmospheric irradiance corresponding to an ambient

temperature Tamb. The net cooling power Pnet(T ) of a structure with area A is given by

Pnet(T ) = Prad(T )�Patm(Tamb)�Psun, (1)

where

Prad(T ) = AZ

dWcosqZ •

0dl IBB(T,l )e(l ,q), (2)

3

Radiative Cooling Surface!

Figure 1(a), represents a strong departure from previously published systems. Rather than designing

a cover filter which is spatially separated from a black emitter we introduce an integrated thermally

selective emitter atop a broadband mirror. Doing so enables us to exploit near-field coupling

between material layers, leading to stronger control over emission, absorption and reflection.

Using nanophotonic concepts, our photonic structure is able to strongly suppress solar absorption

while enhancing thermal emission in the atmospheric transparency window: an ultrabroadband

performance, shown in Fig. Figure 1(b), capable of achieving a net cooling power exceeding

100W/m2 at ambient temperature.

Radiative cooling devices operate at near-ambient temperatures and therefore do not suffer from

many of the difficulties associated with other thermal applications of photonic structures which

have typically involved high temperature operation. These difficulties include numerical uncertainty

associated with the temperature-dependence of optical properties, material cohesion, small-feature

evaporation, and durability that affect other thermal applications.30 Our approach and design is

thus a departure from previous work in using nano- and micro-photonic structures for thermal

applications.

To begin our analysis, we consider a photonic structure at temperature T whose radiative

properties are described by a spectral and angular emissivity e(l ,q). The structure is exposed to a

clear sky subject to solar irradiance, and also atmospheric irradiance corresponding to an ambient

temperature Tamb. The net cooling power Pnet(T ) of a structure with area A is given by

Pnet(T ) = Prad(T )�Patm(Tamb)�Psun, (1)

where

Prad(T ) = AZ

dWcosqZ •

0dl IBB(T,l )e(l ,q), (2)

3

Cooling Power

Pump Heat in at Pnet to maintain at T 4

The need for selective thermal emission

Figure 1(a), represents a strong departure from previously published systems. Rather than designing

a cover filter which is spatially separated from a black emitter we introduce an integrated thermally

selective emitter atop a broadband mirror. Doing so enables us to exploit near-field coupling

between material layers, leading to stronger control over emission, absorption and reflection.

Using nanophotonic concepts, our photonic structure is able to strongly suppress solar absorption

while enhancing thermal emission in the atmospheric transparency window: an ultrabroadband

performance, shown in Fig. Figure 1(b), capable of achieving a net cooling power exceeding

100W/m2 at ambient temperature.

Radiative cooling devices operate at near-ambient temperatures and therefore do not suffer from

many of the difficulties associated with other thermal applications of photonic structures which

have typically involved high temperature operation. These difficulties include numerical uncertainty

associated with the temperature-dependence of optical properties, material cohesion, small-feature

evaporation, and durability that affect other thermal applications.30 Our approach and design is

thus a departure from previous work in using nano- and micro-photonic structures for thermal

applications.

To begin our analysis, we consider a photonic structure at temperature T whose radiative

properties are described by a spectral and angular emissivity e(l ,q). The structure is exposed to a

clear sky subject to solar irradiance, and also atmospheric irradiance corresponding to an ambient

temperature Tamb. The net cooling power Pnet(T ) of a structure with area A is given by

Pnet(T ) = Prad(T )�Patm(Tamb)�Psun, (1)

where

Prad(T ) = AZ

dWcosqZ •

0dl IBB(T,l )e(l ,q), (2)

3

Emissivity of radiating surface Emissivity of atmosphere

Thermal emissivity is not a fixed number:

It can be engineered by photonic design

5

Hemispherical sky access for maximal cooling power

Hemispherical integrals:

Potential of an ideal selective emitter

I. INTRODUCTION

Radiative cooling is a technique that exploits a natural transparency window for electro-

magnetic waves in the Earths atmosphere to transport heat from terrestrial objects into cold

space. As a result, objects with the appropriate radiative properties can passively cool them-

selves down to temperatures well below the ambient. The atmospheric transparency window

is found in the 8-13µm wavelength range, as shown in Fig. 1, and fortuitously overlaps with

the blackbody spectral radiance corresponding to typical terrestrial temperatures (0-50C),

thus enabling objects at these temperatures to emit more power than they absorb.

7 9 11 13 150

0.5

1

� [µm]

Atmospheric Transmission

0�C blackbody

50�C blackbody

FIG. 1. Atmopheric Transmission in the zenith direction vs. wavelength; normalized blackbody

spectral radiance of a 0�C and a 50�C blackbody emitter

Prior work in radiative cooling has almost entirely focused on nighttime cooling, where

one aims to maximize emission in the atmospheric transparency window, without having to

contend with solar radiation. In turn, nighttime cooling is of limited practical relevance since

solar radiation is by far the greatest source of daytime heating. Early work on nighttime

cooling mainly focused on choosing the right materials [3] and exploiting simple interference

e↵ects [4, 5] to maximize emissivity in the transparency window. Granqvist et. al [4, 5]

theoretically investigated and characterized the properties of the ideal nighttime radiator,

finding that such radiator could reach nighttime temperatures which were ⇡ 50oC lower

than the ambient, with a cooling power of ⇡ 100W/m2 when the radiator temperature was

2

0.3 0.5 1 4 8 13 20 300

0.5

1

l [µm]

Em

issivity/A

bso

rp

tivity

Solar Spectrum

Atmospheric

window

Selective + 3%

Selective + 5%

Selective + 10%

(a)

0.3 0.5 1 4 8 13 20 300

0.5

1

l [µm]

Em

issivity/A

bso

rp

tivity

Solar Spectrum

Atmospheric

window

Selective + 3%

Selective + 5%

Selective + 10%

(a)

0.3 0.5 1 4 8 13 20 300

0.5

1

l [µm]

Em

issivity/A

bso

rp

tivity

Solar Spectrum

Atmospheric

window

Selective + 3%

Selective + 5%

Selective + 10%

(a)

−70 −60 −40 −20 0 20 400

40

80

120

160

Temperature (°C)

Coo

ling

Pow

er (W

/m2 )

Tambient  =  27°C  

6

I. INTRODUCTION

Radiative cooling is a technique that exploits a natural transparency window for electro-

magnetic waves in the Earths atmosphere to transport heat from terrestrial objects into cold

space. As a result, objects with the appropriate radiative properties can passively cool them-

selves down to temperatures well below the ambient. The atmospheric transparency window

is found in the 8-13µm wavelength range, as shown in Fig. 1, and fortuitously overlaps with

the blackbody spectral radiance corresponding to typical terrestrial temperatures (0-50C),

thus enabling objects at these temperatures to emit more power than they absorb.

7 9 11 13 150

0.5

1

� [µm]

Atmospheric Transmission

0�C blackbody

50�C blackbody

FIG. 1. Atmopheric Transmission in the zenith direction vs. wavelength; normalized blackbody

spectral radiance of a 0�C and a 50�C blackbody emitter

Prior work in radiative cooling has almost entirely focused on nighttime cooling, where

one aims to maximize emission in the atmospheric transparency window, without having to

contend with solar radiation. In turn, nighttime cooling is of limited practical relevance since

solar radiation is by far the greatest source of daytime heating. Early work on nighttime

cooling mainly focused on choosing the right materials [3] and exploiting simple interference

e↵ects [4, 5] to maximize emissivity in the transparency window. Granqvist et. al [4, 5]

theoretically investigated and characterized the properties of the ideal nighttime radiator,

finding that such radiator could reach nighttime temperatures which were ⇡ 50oC lower

than the ambient, with a cooling power of ⇡ 100W/m2 when the radiator temperature was

2

−70 −60 −40 −20 0 20 400

40

80

120

160

Temperature (°C)

Coo

ling

Pow

er (W

/m2 )

Potential of an ideal selective emitter

0.3 0.5 1 4 8 13 20 300

0.5

1

l [µm]

Em

issivity/A

bso

rp

tivity

Solar Spectrum

Atmospheric

window

Selective + 3%

Selective + 5%

Selective + 10%

(a)

0.3 0.5 1 4 8 13 20 300

0.5

1

l [µm]

Em

issivity/A

bso

rp

tivity

Solar Spectrum

Atmospheric

window

Selective + 3%

Selective + 5%

Selective + 10%

(a)

0.3 0.5 1 4 8 13 20 300

0.5

1

l [µm]

Em

issivity/A

bso

rp

tivity

Solar Spectrum

Atmospheric

window

Selective + 3%

Selective + 5%

Selective + 10%

(a)

Tambient  =  27°C  

Effective of parasitic ambient heating when below air temperature:

Pnonrad = hcΔT hc = 1 W/m2K  

hc = 4 W/m2K  

7

10:00 pm 11:00 pm 12:00 am 1:00 am

0

5

10

15

20

Time of Day

Tem

pera

ture

(°C

)

Radiative Cooling at Night: Experimental Data

Air  Temperature  

Al  +  SiO2  

8

Sky-facing

1 μm Low$density,Polyethylene,Cover,

SiO2%

Al,Polystyrene,packaging,

Point Po!(seI) [w/m2) Po!(bb) [W/m2) A 9.9 94

010 .lL 8 18 17

C 28 27 ::J D 37 36

5 E

0

II ! ! I I I I

14 R.H. _____ 1 <fi

? 8 23 24 01 I I 02 03

Time [hours)

FIG. 20. Upper part shows temperatures of the selective Isell and blackbody Ibbl cooling plates as a function of time during 25-26, 1980. At the points denoted A-D we fed in the indicated electriC powers P<i' Lower part shows ambient temperature To and relative humidity IR.H·1

17 .c. Radiative cooling now brought down the temperature at a rate of - 1 ·C/min. As expected, the initial decrease was most rapid for the blackbody surface. The panels were then left for about 2 h to reach their minimum temperature. The ambient temperature changed by less than 1 ·C during this period. The cooling rate was seen to diminish faster for the blackbody surface than for the infrared selective one, and after about 1 h the selective surface began to show the lowest temperature. This cooling run was interrupted at 004°, when the temperature difference Ll T =Ta - Ts was 13.8 ·K for the infrared selective plate and 13.4·C for the blackbody plate. This small difference between the two types of ings may seem discouraging, but it is important to bear In

mind that the performance of the infrared selective surface was by no means optimal, as will be discussed below.

Cooling power versus temperature difference was stud-ied by successively heating the plates and noting the stagna-tion temperatures pertaining to the different electrical power inputs (Pel)' Figure 20 shows the result of increasing Pel in four steps. We allowed at least 30 min to reach approximate thermal equilibrium in each case. It is seen that a certain input yields a larger temperature rise for the infrared selec-tive surface than for the blackbody surface, which is expect-ed since the radiated power is largest in the latter case. We continued the heating sequence after 03°° in still more steps up to plate temperatures approaching or exceeding Ta. These final experiments were somewhat troubled by falling ambient temperature accompanied with increasing relative humidity which gave some condensation of water vapor on the top polyethylene covers. However, this effect was not sufficiently serious so as to make measurements impossible.

C. Comparsion with theoretical predictions The raw data for radiative cooling shown in the preced-

ing section are plotted in Fig. 21 in a way which facilitates

4217 J. Appl. Phys., Vol. 52, No.6, June 1981

100

80

60 a..<.>

o

, , , , ,

Blackbody • / surface

(OJ "'" (4

Infrared 0 •

selective '" surface

Temperature difference, 15

FIG. 21. Measured cooling power vs temperature difference for two types of surface. The data are based on measurements presented in Fig. 20. The symbols within parentheses are uncertain since the polyethylene covers were in these cases covered with thin layers of condensed water vapor.

comparison with theory. The stagnation temperatures reached after sufficient periods of constant electrical heating give certain values of Ll T. The corresponding cooling powers are governed by

Pc =Pel = Prad - Ploss' (28) The relations between these two quantities are found to be approximately linear. As expected, the infrared selective surface displays a smaller slope and a lower intercept at LlT= O.

We first consider the values at Ll T = 0, since only radia-tive exchange with the atmosphere takes place in this case. The approximate blackbody radiator, having = 0.9, yields Pc = 84 W 1m2

• Comparison with the model calcula-tions in Sec. III C leads us to estimate that €a, :S 0.1 which seems unlikely with regard to the high ambient temperature and relative humidity. The most probable explanation is that some radiation comes also from parts of the polystyrene block which are in close vicinity of the cooling plate. The infrared selective surface shows Pc = 61 W 1m2 which is also higher than expected from the experimental value

= 0.45 for the SiO coating. This, again, points at the im-of stray radiation. The best relieffor such obscur-

ing "edge effects" would be to work with cooling panels much larger than the present ones.

When the range Ll T> 0 is considered it is essential to include also nonradiative losses. The upper dashed curve in Fig. 22 indicates Ploss vs Ll T for K = 2.5 Wm - 2K - I. This magnitude of the heat transfer coefficient applies to the two test panels. The solid lines denote Prad =Pc + Ploss, with Pc taken from Fig. 21. The intersections of the solid lines and the upper dashed one clearly represent the lowest obtainable temperatures. If nonradiative losses were absent, Fig. 22 pre-dicts that Ll T = 32 ·C could be reached for the SiO-coated surface and Ll T = 22.5 ·C could be reached for the painted surface. A comparison with the results shown in Figs. 6 and

C. G. Granqvist and A. Hjortsberg 4217

Downloaded 09 Jun 2013 to 171.67.34.205. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jap.aip.org/about/rights_and_permissions

Catalanotti, S. et al. Solar Energy 17, 83 – 89 (1975). Granqvist, C. G. & Hjortsberg, A. Applied Physics Letters 36, 139–141 (1980). Granqvist, C. G. & Hjortsberg, A. Journal of Applied Physics 52, 4205–4220 (1981). Berdahl, P., Martin, M. & Sakkal, F. International Journal of Heat and Mass Transfer 26, 871 – 880 (1983). Orel, B., Gunde, M. & Krainer, A. Solar Energy 50, 477 – 482 (1993). Suryawanshi, C. N. & Lin, C.-T. ACS Applied Materials and Interfaces 1, 1334–1338 (2009). Gentle, A. R. & Smith, G. B. Nano Letters 10, 373–379 (2010). Granqvist & Hjortsberg (1980)

Other night-sky radiators: Tedlar, paints

The challenge during the day

Daytime Radiative Cooling Surface!

cold outer space!sun!

Send    Heat  

Reflect  Sunlight  

T4 intuition:

Sun: 6000 K Earth objects: 300 K

0.3 0.5 1 4 8 13 20 300

0.5

1

l [µm]

Em

issivity/A

bsorptivity

Solar Spectrum

Atmospheric

window

Selective + 3%

Selective + 5%

Selective + 10%

(a)

9

Cooling demand peaks during the day:

Can we achieve meaningful cooling then?

240 260 280 300 320− 100

− 50

0

50

100

150

Temperature [K] ( T ambient = 300K)

Pcooling[W

/m2 ]

Ideal selective emitterSelective + 3%Selective + 5%

Selective + 10%

Simultaneously emit thermal radiation and reflect sunlight strongly

10

solar absorption!

Photonic design enables daytime radiative cooling

Achieve desired thermal emissivity by photonic design Overcome material limitations

240 260 280 300 320�100

�50

0

50

100

150

Temperature [K] (Tambient = 300K)

Pco

olin

g[W

/m2 ]

Ideal selective emitter

Optimized Design

Selective + 3%Selective + 5%

Selective + 10%

(b)

0.3 0.5 1 4 8 13 20 25 30

0

0.5

1

� [µm]

Emissivity/Absorptivity

Solar Spectrum

Atmospheric

window

(b)

Rephaeli*, Raman* and Fan, Nano Letters (2013) 11

Benefits & challenges of radiative cooling

Passive (electricity-free) cooling powers of ~30-90 W/m2 to drop temperatures 5-10°C below ambient

With advanced photonic design, available at all hours of the day

Operating costs limited to maintenance and pumping energy

Can work in the hottest (and even humid) days

Can be additive to other cooling mechanisms

12

Inherently large-area

Geometrical constraints: flat surface with access to full sky hemisphere ideal

Atmospheric constraints: performance can drop by 1/2 if cloudy/rainy

Efficient packaging needed to ensure minimal parasitic losses to the ambient

Maximal benefit available at night

Acknowledgements Marc Anoma Eli Goldstein Prof. Shanhui Fan

Advanced  Research  Projects  Agency  -­‐  Energy  

13

Linxiao Zhu Dr. Eden Rephaeli

OPEN 2012 Awardee