You are on page 1of 9

Radiative Heat Transfer Between Two Closely-Spaced Plates

Reh-Lin Chen*
Rockwell Scientific Company LLC., Thousand Oaks, CA 91360

Enhanced radiative heat transfer by proximity effect has been confirmed both
theoretically and experimentally since the early 1970’s. The proximity effect indicates that
heat transfer increases rapidly with spacing much smaller than dominated wavelength. This
topic was recently investigated on dielectric materials for a thermophotovoltaic application,
in which photovoltaic efficiency may be enhanced through proximity effect. We report a
numerical result on the radiative heat transfer between a pair of parallel electrodes in a new
thermotunneling application, in which the spacing between the electrodes is in the
nanometers range. Two electron models along with optical data were used for the
calculation. Although the calculation is mainly for absorptive materials, review for
homogeneous media is also included. The motivation of our work is to compute the heat loss
through thermal radiation as part of energy budget in a thermotunneling, direct thermal to
electric conversion (DTEC) device. This device may offer a much higher theoretical
efficiency than thermoelectric devices since it has a vacuum gap so thermal backflow and
Joule loss are minimal in the gap. We have found that the radiative heat loss is greater than
Planck blackbody radiation by one to two orders of magnitude.

Nomenclature
E = electric field, N/C
H = magnetic field, N/(A*m)
Q& = radiative heat flux, W/m2
S = Poynting vector, W/m2
T = absolute temperature, Kelvin
W = differential heat flux, W/(m2*K)
c = velocity of light in vacuum, 2.998*108 m/s
e = electron charge, 1.602*10-19C
h = Planck’s constant, 6.6256*10-34 J*s
j = current density, A/m2
k = wave number, 1/m
kB = Boltzmann constant, 1.3805*10-23 J/K
n = real part of refractive index
ni = index of refraction in medium i
rij = reflection amplitude at interface i and j
s = spacing between medium I and III, m
tij = transmission amplitude at interface i and j
β = phase angle between successively reflected waves, rad
Γ = transmissivity, 0< Γ<1
ε = dielectric constant, F/m
ε0 = permittivity of free space, 8.854*10-12 F/m
φij = phase angle of media i and j in evanescent wave transmission amplitude, rad
θ = angle measured between incident wave and normal direction of interface, rad
σ = Stefan-Boltzmann constant, 5.67*10-8 W/(m2*K4)
σ0 = electrical conductivity, 1/(Ω*m)
σ' = normalized electrical conductivity, σ ' = σ 0 /(4πε 0 ) , 1/s

*
Research Scientist, Dept. Applied Computational Physics, A24, AIAA Member.

American Institute of Aeronautics and Astronautics


λ = wavelength of electromagnetic wave, m
ϕ = azimuthal angle, rad
κ = imaginary part of refractive index
ω = angular frequency of electromagnetic wave, 1/s
ωc = electron collision frequency, 1/s

I. Introduction

T HE topic of radiative heat transfer between two closely spaced homogeneous or non-homogeneous materials
was investigated in the 1970’s by several authors; e.g., Hargreaves1 and Domoto, Boehm, and Tien2
(experimental), Cravalho, Tien, and Caren,3 Domoto and Tien,4 Boehm and Tien5, and Polder and Van Hove6
(theoretical) .The enhancement of radiative heat transfer due to proximity was confirmed. The experimental work
done by both Tien et al. and Hargreaves was at a rather large spacing of 10 µm and 1 µm, respectively, much greater
than that in a thermotunneling device of interest to us. The size of the gap is determined mostly by work function of
the electrodes. The most relevant work is by Polder and Van Hove6, in which the heat flux of a pair of parallel metal
plates at different temperatures was calculated. Polder and Van Hove chose Cr as the heat sources and compared
with measurement by Hargreaves. Chromium turned out to be an unsuitable candidate for the study because of its
magnetic properties and complicated optical properties within the mid-infrared range7. As a result, the comparison
between experiment and theory, although within the same order of magnitude, is not satisfactory in the far field,
where Planck’s law was applied for verification. Their calculation used approximated numerical integration. The
goal of our work was to find the radiative heat loss of a pair of metal plates at two different temperatures and
separated by a vacuum gap in the nanometers range. This heat loss calculation helps to estimate thermodynamic
efficiency. The scope of this paper is to review the radiative heat transfer of homogeneous media that has a
negligible imaginary part index of refraction and to compute for absorptive media as metals, which are non-
homogeneous. In the latter case, surface modes were found at a much shorter wavelength than our operating range
of 6-10 µm and were not taken into account (see Ref. 8 for surface modes, for example). Cavity QED effect is also
not considered.9 Although the calculation in this paper was done for Au and Cr, it can be extended to any suitable
non-magnetic material.
The calculation for heat transfer of homogeneous media includes the normal Planck black body radiation integral
incorporated with a transmissivity term based on the Fresnel formula, and takes into account phasing due to multiple
reflections and shifts at the interface. Our calculation of this part followed Tien et al.1 and Born,10 and will be
described in Sec. II. Both p- and s-polarizations are assumed to be equally probable since they are induced
thermally. Therefore, the one-way heat flux calculation contains four terms; i.e., two polarizations of both
propagating and evanescent waves. The net heat flux is then the difference of heat fluxes in both directions. The
calculation of radiative heat transfer in metals is more complex. Metals are nonhomogenuous media that their
dielectric constants depend on the incident angle. The aforementioned optical multi-layer approach for
homogeneous media would have led to extremely awkward expressions. Our approach in this part followed the work
by Polder and Van Hove and will be discussed in Sec. III.

Figure 1. Reflection and transmission between two Figure 2. Schematic of two metal bodies separated
dielectric materials separated by a vacuum gap. by a vacuum gap for radiative heat transfer
calculation, as described in Ref. 6.

2
American Institute of Aeronautics and Astronautics
II. Radiative Heat Transfer in Homogeneous Dielectric Media
As depicted in Fig. 1, two plates of temperatures T1 at the hot side and T3 at the cold side are separated by a
vacuum gap (n2 = 1) of spacing s. Medium I and medium III are assumed nonmagnetic and both have an index of
refraction (n1 and n3) larger than one but not necessarily equal. According to Snell’s law there will be a critical angle
such that the electromagnetic wave incident at the interface of medium one and two is totally reflected; i.e.,
n1 ⋅ sin θ c = n2 , θc being the critical angle. Waves past the critical angle will only propagate along the interface and
exponentially decay in the positive z-direction. Normally this evanescent wave is very insignificant since the gap
between the two bodies is much larger than the dominated wavelength in Wien’s law, that is, λ = 0.002898/(T*n) in
meters. The dominated wavelength is around 10 µm at room temperature. As the gap approaches the dominated
wavelength, the evanescent wave can in fact penetrate to medium III and contributes to a heat flux greater than that
in black body radiation. In the extreme at zero spacing, the radiative heat flux is n12 times the normal black body
radiation, when n1 = n3. There may also be another critical angle at the interface of medium II and III when n1 is
different from n3 and the calculation is similar. The unmatched index of refraction yields a smaller radiative heat
flux than the matched case. This near-field evanescent wave effect was demonstrated by Feynman11 using
microwaves of a few centimeters in wavelength. The formulation of radiative heat transfer including both
evanescent and propagating waves is sketched as follows.
The equations for heat flux from medium I to III for both propagating and evanescent waves are
∞ 2π θ c

∫ ∫ ∫Γ
sin θ 1 cos θ 1 d θ 1 d ϕ d λ
Q& 1prop
→3 =
2 hc 2
n1 2
prop
1→ 3
hc − (1 )
0 0 0 λ 5 ( e λ n1 k B T − 1 )
∞ 2 π θ max

∫ ∫ θ∫ Γ
sin θ 1 cos θ 1 d θ 1 d ϕ d λ
Q& 1evas
→3 =
2 hc 2
n1 2
evas
1→ 3
hc − (2)
0 0 λ 5 ( e λ n1 k B T − 1 )
c
where the transmissivity terms

n 3 cos θ 3 t12 2 t 23 2
Γ 1 prop = n 1 cos θ 1 1 + r r 2 2 − (3)
→3 12 23 + 2 r12 r23 cos( 2 β )
2 2
n 3 cos θ 3 τ 12 τ 23
Γ 1→ 3 =
evas
n 1 cos θ 1 2 cos( φ 12 + φ 23 ) + 2 cosh( 2 b ) − (4)
In Eq. (3-4), β = 2π / λ ⋅ ( sn2 cosθ 2 ) and b = −2π / λ ⋅ n12 sin θ12 − 1 . The equations are similar for heat flux
going the opposite direction. The expressions of reflection and transmission coefficients are attached in the
Appendix. Note that the 3rd integral in Eq. (2) has an upper limit that depends on the relative value of n1 and n3. The
upper angle limit, θc, is π/2 when n3 > n1 and Sin-1(n3/n1), otherwise. The transmissivity term, Γ, acts as a weighting
coefficient. Its expression depends on whether the waves are propagating or evanescent in a specific polarization and
need to be calculated accordingly. In the special case when spacing approaches zero, the sum of all heat fluxes from
both propagating and evanescent waves of either polarization adds to n12 * σT4, that is,
π
∞ 2π 2

∫ ∫∫
sin θ 1 cos θ 1 d θ 1 d ϕ d λ 2 k B 4π 5
Q& = = n2( = n 2σ T − (5 )
2 4 4
2 hc
n1 2 hc
15 c 2 h 3
)T
5 λ n1 k B T
0 0 0 λ (e −1)
Here the Stefan-Boltzmann constant was re-derived in Eq. (5). The explanation for the n2 increase is this:12 As
wavelengths depend on medium dielectric constant, the larger the dielectric constant, the smaller the wavelength.
Therefore the radiation modes that can be contained in a medium are proportional to n13. The heat flux in the
Poynting vector is proportional to n*E2. The extra modes are normally inaccessible for spacing much larger than
dominated wavelength since they exist past the critical angle. Our result for homogeneous media agrees with that in
Ref. 12 very well, including the zero-spacing heat flux. For calculation at low temperatures comparing with Ref. 1,
our magnitude is also close but no “dip” was found in the heat flux curve at wavelengths in the near-field transition
region.

3
American Institute of Aeronautics and Astronautics
III. Radiative Heat Transfer in Metals

Our calculation of radiative heat flux in metals follows the approach theorized by Polder and Van Hove.6 A
schematic of such a device is depicted in Fig. 2. The assumptions are: macroscopic Maxwell’s equations are valid,
the metals are nonmagnetic, there is zero correlation length for currents, and the quantities are stationary in time
with local thermodynamic equilibrium. Naturally the approach will not be valid when the gap spacing is on the order
of atomic size. The approach of Polder and Van Hove is similar to that of Rytov13 on electric fluctuation and thermal
radiation but Polder and Van Hove used fluctuating current instead of field. Hargreaves conducted measurements
down to 1 µm and qualitatively demonstrated the near field proximity effect on heat flux. The relevant equations for
calculation are shown in this section and the Appendix for reference. The radiative heat flux calculated by Polder
and Van Hove is temperature differential heat flux. For two bodies having a temperature difference ∆T, the total heat
flux between the two is the product of differential heat flux and ∆T.
The Polder and Van Hove approach is sketched
as follows. Two Maxwell equations for magnetic and
electric fields with a current source term due to thermal
fluctuations are shown in Eq. (6-7) The Poynting vector in
Eq.(8) in the z-direction gives the heat flux density in units
of watts per unit area. The temperature differential heat
flux is denoted as W and is divided into both propagating
(denoted sin in Polder and Van Hove paper) wave and
evanescent (denoted exp) wave contributions, as expressed
in Eq. (9).
r r r r r r r
∇ × H ( x , ω ) = iωc ε ( x , ω ) E ( x , ω ) + 4π
c j ( x , ω ) − ( 6)
r r r r Critical angle
rad
∇ × E ( x , ω ) = − iωc H ( x , ω ) − (7)
r r r r r r Figure 3. Evanescent wave snap shot from Eq.
S ( x , t ) = 4cπ E ( x , t ) × H ( x , t ) − (8) (10). Waves past critical angle decay in the z-
direction exponentially.
W = W sin + W exp − (9)

Polder and Van Hove6 argued that there would be no net energy flux when both plates are at the same temperature.
Therefore W is the temperature derivative of the Poynting vector in units of W/cm2/K. A mean temperature and a
temperature differential are used to estimate the heat flux in the temperature range of interest. It is therefore more
accurate when the temperature differential, ∆T, is small.
The electromagnetic field in the medium is expressed as
E (t ) = E0 exp(iωt ) exp( −ik x x) exp( ±ik z z ) − (10)
with a dispersion relation, k x + k z = εω / c . A visual aid for evanescent waves is shown in Fig. 3. At incident
2 2 2 2

angles smaller than the critical angle, the waves are sinusoidal with same amplitude and exponentially decay in the
z-direction when past the critical angle. The propagating constants kz (in medium) and kzv (in vacuum) are therefore
respectively, k z = εω 2 / c 2 − k x 2 , k zv = ω 2 / c 2 − k x 2 . The net differential heat flux is
∞ ω /c ∞

∫ }{ ∫ k x (t|| ∫ k (t + t ⊥ }dk x }dω − (11)


sin sin exp exp
W =W sin
+W exp
= 1
( 2π ) 2

∂T { hω
exp( hω / k BT ) −1 + t ⊥ }dk x + x ||
0 0 ω/c
The expressions of the transmission coefficients in Eq. (11) are included in the Appendix.

A. Electron Models
The convention for dielectric constants is ε = ε 1 − iε 2 = (n − iκ ) 2 = n~ 2 ,14 which gives
ε 1 = n 2 − κ 2 and ε 2 = 2nκ and implied the use of + iωt . The relative dielectric constant in bulk metals is roughly

4
American Institute of Aeronautics and Astronautics
estimated as ε ≈ −iε 2 = −i 4πσ ' / ω in the first model, where σ ' = 7.0*1016 (1/s) was used for Cr in Polder and Van
Hove paper6. Here σ ' = σ 0 /(4πε 0 ) , σ0 is the DC electrical conductivity and equals Ne 2 / mωc 2 , ωc ( = 1 / τ c ) being the
collision frequency of electron. The conductivity, σ0, of Cr and Au at 295 K are 7.8*106 and 4.55*107 1/(Ωm).15 In
the 2nd model using Drude theory, the real and imaginary parts of dielectric constant are,
ε 1 = 1 − σ 0τ c /(ε 0 + ε 0ω 2τ c 2 ) and ε 2 = σ 0 /(ε 0ω + ε 0ω 3τ c 2 ) . Drude model is in excellent agreement with
measurements between 3-30 µm16 for noble metals. The anomalous skin effect was not considered in the calculation.
This effect states that at high frequency, the electron traversing time between collisions is much greater than that in
radiation, so there will be varying electric field during the traverse.

B. Palik Optical Data


The literature of optical data for Au is extensive but most of it was done in the visible and ultraviolet range. The
optical data we used comes from the 1985 Palik optical handbook.17 In fact, the Au data were omitted in later
editions of 1991 and 1998. The optical data were used to curve fit dielectric constants and index of refraction to be
used in the heat flux calculation. The resulting real and imaginary parts of dielectric constants, ε1 and ε2, are
ε 1 = − 55 .647 + 2 .763 * 10 −16 (1856 .954 − 4 .845 * 10 −13 ( − 1 .800 * 10 15 + ω ))( − 1 .800 * 10 15 + ω ) − (12 ) ,
ε 2 = −4.395 + 2.762 * 10 −16 (−636.111 + 1.745 * 10 −13 (−1.800 * 1015 + ω ))(−1.800 * 1015 + ω ) − (13)
Figure 4 shows a plot of dielectric constant as a function of wavelength from .3 to 10 µm. The real part of dielectric
constant exceeds the imaginary part. The Au dielectric constant can be less than one in the visible and UV region. In
Fig. 5 is a plot of refractive indices of n and k from the same optical data. These curves are smooth in the range
between 6-10 µm and absent from significant peaks and troughs. The curve-fit expressions for dielectric constant
were used to plug into the Drude model for differential heat flux.

-ε1
Palik 1985 k
optical data -ε
2

Drude
theory
n

Figure 4. Dielectric constant of Au as function of Figure 5. Index of refraction of Au as function of


wavelength. Solid lines are ε 1 and dashed wavelength. Solid line is real part and dashed line is
imaginary part.
lines ε 2 using Palik optical data17 and Drude theory.

C. Result of Cr and Comparison with Polder-Van Hove Theory


In Fig. 6 the transmission coefficient of both polarizations are shown as a function of spacing, d, at a
vacuum wavelength of 5 µm. The cut-off wave number is 12566.4 (1/cm). The evanescent wave can be seen past the
cut-off. The dielectric constant used for this calculation is –200-i*130. Figure 7 shows the differential heat flux in
Polder and Van Hove paper using Cr and the Drude model. The σ ' and ω0 used in the calculations are 3.5*1016 (1/s)
and 1.41*1014 (1/s). Our result on the same conditions is shown in Fig. 8 and 9. The itemized contributions are as
indicated. Both plots look very similar to Polder and Van Hove though using a very different computational tool.
The outcomes of our calculations using the first electron model also compare similarly and are both greater than that
with the Drude model by one order of magnitude. As mentioned in the introduction, the integration is very
oscillatory in nature. The integration limits must be set properly to get a reasonable answer without paying too much
computational expense. If the range is too large, there will be insufficient grid points for accurate integration. On the
other hand if the range is too small, the frequency range that has a non-negligible contribution may not be covered
and the calculation will not be accurate enough. Besides the two electron models and optical data, we also used
Fragstein18 approximation for the Au heat flux calculation. In the approximation, a factor of (n12 + k12 ) / n14 was used
in the heat flux integral similar as in Eq. (1-2) but replacing the transmissivity term. As the vacuum gap approaches

5
American Institute of Aeronautics and Astronautics
zero, the incident angle can be assumed close to normal incidence. This yielded a heat flux of 3.8 W/cm2 and close
to the result using the Drude model.

Figure 6. Result of transmission coefficients in the Figure 7. Result of itemized differential radiative
Polder-Van Hove paper. The critical wave number heat flux by Polder and Van Hove6.
in both polarizations is shown.

W total

W-eva-||

W-prop

W-eva- ⊥

Figure 8. Calculated transmission coefficients to Figure 9. Calculated total differential heat flux
compare with Fig. 5. The parameters and axis labels and individual contributions. The propagating and
are the same as in Fig. 5. evanescent waves are the same as sinusoidal (sin)
and exponential (exp) waves denoted in Fig. 6.

The verification of our result with Polder and Van Hove started with Fig. 8. As compared with Fig. 6, a dielectric
2
constant with a real part ε1 = 1 − (ω4πσ+ω'ω
2
0
2 and imaginary part ε 2 = − ω 4(ωπσ +'ωω
2
0
2 along with σ of 3.5⋅1016 (1/s) and
0 ) 0 )

ω0 of 1.41⋅1014 (1/s) were used in the calculation. It can be seen that the agreement is very good between Fig. 7 and
Fig. 9.

D. Result of Au Heat Flux Using Polder-Van Hove Theory


The result of Au heat flux (in W/cm2) using the same Drude model is shown in Fig. 10. The corresponding σ '
and ω0 of 1.7⋅1017 (1/s) and 2.33⋅1014 (1/s) were used for Au. The individual contributions are indicated by arrows.
The heat flux did not appear to be very sensitive to the change in conductivity. The heat flux varies slowly at
spacing less than several hundred nanometers. The total net heat flux is about 6 W/cm2 when the hot and cold sides
are 400K and 300K. The same heat flux was also calculated using Palik optical data, in which the dielectric
functions of Au were curve fitted within the region of interest. This curve function was expressed in Eq. (12-13).
The result is shown in Fig. 11. Curves of different gray levels represent different contributions and are consistent in
both figures.

6
American Institute of Aeronautics and Astronautics
W-eva- ⊥
W total
W-eva-||
W-eva-||
W-prop

W-eva- ⊥
W-prop

Figure 10. Calculated heat flux for Au using Figure 11. Calculated heat flux using 1985
the Drude model. The greatest contribution Palik optical data. Unlike Fig. 9, there is a
comes from evanescent wave of s-polarization. peak and trough in the total heat flux but the
overall magnitude is the same.

Two calculated radiative heat flux results are shown in Fig. 10 and Fig. 11. The operating temperatures are both
at 400 K on hot side and 300 K on cold side. Figure 10 used the Drude model and Fig. 11 used 1985 edition Palik
handbook of optical data. Notice there is a peak and a trough in Figure 11. The apparent difference seems to come
from the shift in dielectric constants between the Drude theory and optical data as shown in Fig. 4.

Total Heat 1 nm 2 nm 5 nm 10 nm 20 nm 50 nm 100 nm 500 nm 1000 nm


Flux (W/cm2)
400 K 5.94 5.91 5.87 5.82 5.65 4.70 2.84 0.059 .0051
450 K 8.93 8.88 8.82 8.74 8.49 7.07 4.27 .089 .0077
Table 1. Radiative heat flux at two hot side temperatures using Au and Drude model. Cold side
temperature was held at 300 K.

IV. Conclusion
The radiative heat flux between two Au electrodes at two different temperatures separated by a vacuum gap
was calculated using Polder and Van Hove theory. Two electron models and optical data were used for the dielectric
properties. The radiative heat flux was on the order of 10 W/cm2, compared to Planck black body radiation of 0.1
W/cm2 at the same temperatures. In our calculation, surface plasmon mode was ignored since we are operating at a
much lower frequency (than plasma frequency of ω p = Ne 2 /(τ 0 m) < 0.2 µm). The result of heat flux at several
spacings and two hot side temperatures is shown in Table 1. The calculated heat flux decreases rapidly with spacing
above 100 nm and varies with temperature differential almost linearly since the calculation of the heat flux is based
on temperature differential so the total heat flux can be approximated with heat flux differential multiplied by ∆T. A
future study would improve this part of the calculation. Evanescent wave of s-polarization gave the greatest
contribution in net heat flux. There is a noticeable peak and trough in the result using Palik optical data (Fig. 11)
apparently coming from the difference in dielectric constant between Drude theory and optical data. Nonetheless,
the order of magnitude remains the same. Our result also compares favorably using an approximation by Fragstein,18
knowing the transmissivity term in the heat flux integral is nearly unity when spacing is much smaller than
dominated wavelength. For comparison with homogeneous media, it appears that the radiative heat transfer between
a pair of materials both having the same real index of refraction of, for example, (a) 10 (homogeneous dielectric
media) and (b) 10 - 50 I (metals), is much greater in the first case for giving same hot and cold side temperatures at a
sub-wavelength spacing. Our heat flux result is greater than blackbody radiation of the same material in the far field
but much smaller than that of homogeneous media with the same real part of the refractive index. The loss due to
radiation in the vacuum gap appears to be non-negligible and greater than the Planck blackbody radiation by one- to
two orders of magnitude for Au-Au. Metals are non-homogeneous so the Poynting vector also has a longitudinal
component parallel to electromagnetic field in addition to the perpendicular direction as in pure transverse waves.
Therefore the heat flux depends on the relative magnitude of the real and imaginary parts in dielectric constants.

7
American Institute of Aeronautics and Astronautics
Appendix
This appendix lists several equations that were used in the calculation for both homogeneous dielectric media1,10
and metals.6
For homogeneous dielectric media, the amplitudes for reflection and transmission are
ni cosθ i − n j cosθ j
ij r =
ni cosθ i + n j cosθ j − (i)
2 ni cosθ i
tij = ni cosθ i + n j cosθ j − (ii )
for propagating waves. For evanescent waves,
n1 cosθ1 −i n12 sin 2 θ1 − n2 2
τ 12 = − (iii )
n1 cosθ1 +i n12 sin 2 θ1 − n2 2

i n12 sin 2 θ1 − n2 2 − n3 cosθ 3


τ 23 = − (iv)
i n12 sin 2 θ1 − n2 2 + n3 cosθ 3

where τ 12 = exp(iφ12 ), τ 23 = exp(iφ23 ) .


The following equations were used to calculate the differential heat flux for metals6.
∞ ω /c ~
W sin = ∫ dω ∫
sin sin
dk x kx
( 2π ) 2
(t|| + t ⊥ ) ∂∂T ( ehωh/ ωkT −1 )
0 0
sin (1− R|| ) 2
t|| = i ( β|| − 2 k zv d ) 2
1− R||e

sin (1− R⊥ | ) 2
t⊥ = 2
1− R⊥ ei ( β ⊥ − 2 k zv d )

β || = 2 Arg[(εk zv − k z )(εk zv + k z )* ]
β ⊥ = 2 Arg[(k zv − k z )(k zv + k z )* ]
ε ⋅k zv − k z 2
R|| = ε ⋅k zv + k z

k zv − k z 2
R⊥ = k zv + k z

∞ ∞ ~
W exp = ∫ dω ∫
exp exp
dk x kx
( 2π ) 2
(t|| + t⊥ ) ∂∂T ( eh~ωh/ωkT −1 )
0 ω /c
exp 1−cos χ||
t|| = cosh[ 2κ ( d −δ || )]−cos χ||

exp 1−cos χ ⊥
t⊥ = cosh[ 2κ ( d −δ ⊥ )]−cos χ ⊥

kδ|| ikε + k z 2
e = ikε − k z

ik + k z 2
e kδ ⊥ = ik − k z

χ || = 2 Arg[(−ikε + k z )(−ikε − k z )* ]
χ ⊥ = 2 Arg[(−ik + k z )(−ik − k z )* ]

Acknowledgments
The author thanks DARPA and the Boeing Company for funding support on this short-term task, Prof. Herbert
Kroemer of UCSB and colleague Dr. Isoris Gergis for helpful discussions, and department manager, Dr. C.L. Chen
for encouragement and support.

8
American Institute of Aeronautics and Astronautics
References

Periodicals
1
C.M. Hargreaves, Phys. Letters 30A, 491 (1969)
2
G.A. Domoto, R.F. Boehm, and C.L. Tien, “Experimental Investigation of Radiative Transfer Between Metallic Surfaces at
Cryogenic Temperatures,” Journal of Heat Transfer, Trans. ASME Series C, Vol. 92, No. 3, 412-417 (1970)
3
E.G. Cravalho, C.L. Tien, and R.P. Caren, Journal of Heat Transfer, 351-358 (1967).
4
G.A. Domoto and C.L. Tien, “Thick Film Analysis of Radiative Transfer Between Parallel Metallic Surfaces,” Journal of
Heat Transfer, Trans. ASME Series. C 92, 399-404 (1970)
5
R.F. Boehm, C.L. Tien, “Small Spacing Analysis of Radiation Transfer Between Parallel Metallic Surfaces,” Journal of
Heat Transfer, Trans. ASME Series C 92, No. 3, 405-411 (1970)
6
D.Polder and M. Van Hove, Phys. Rev. B, Vol. 4, 1971, 3303-3314.
7
C.M. Hargreaves, Philips Res. Repts. Suppl., No. 3 (1973).
8
A. Narayanawamy and G. Chen, Appl. Phys. Lett. Vol 82, 2003, 3544-3546.
9
S. Haroche and D. Kleppner, Physics Today Vol. 42, No. 1, 24-30 (1989).
12
J.L. Pan, H. Choy, and C.G. Fonstad Jr., IEEE Transactions on Electron Devices Vol. 47, 2000, 241-249.
18
C. Fragstein, “Energy Transfer at the Interface Between Two Absorbing Media with an Emphasis on the Heat
Radiation in absorbing Bodies,” Ann. Physik, Vol. 7, 63 (1950)

Books
10
M. Born, Principle of Optics, 7th ed., Cambridge University Press (1999).
11
R.P. Feynman, Lectures on Physics, Vol II, Addison-Wesley Publishing Co. Inc., Reading MA, 1964, 33-13.
13
S.M. Rytov, Theory of Electric Fluctations and Thermal Radiation, Air Force Cambridge Research Center, Bedford, MA
(1959).
14
L. Ward, The Optical Constants of Bulk Materials and Films, Adam Hilger, Bristol, and Philadelphia (1988).
15
C. Kittel, Introduction to Solid State Physics, 6th ed., 144, John Wiley & Sons, Inc. (1986).
16
H.E. Bennett and J.M. Bennett, Optical Properties and Electronic Structure of Metals and Alloys, North-Holand,
Amsterdam (1966).
17
E. D. Palik, ed., Handbook of Optical Constants of Solids II, Academic Press, Inc., Florida (1985)

9
American Institute of Aeronautics and Astronautics

You might also like