2.

58, Spring 2006

Problem Set 1
Probs 1.8, 1.11, 3.4, 4.16, 5.13 and 5.18 (Radiative Heat Transfer, 2nd ed, M.F. Modest)

2.58J, HW #2, Spring 2006: Problem 5 can be handed in together with HW#3.
1. Problem 2.6
2. Problem 6.1.
3. Problem 3.31
4. Problem 5.34, add question (d):
(d) Find numerical solutions for the governing equations you derived in (a) using
numerical quadrature.
5. Write a Monte Carlo code for Problem 5.34 and compare your results with these
obtained by numerical method.

2.58J, HW# 3,
1. Problem 2.6 (problem 1 from HW#2)
2. Problem 3.31 (problem 3 from HW#2)
3. Monte Carlo code (problem 5 from HW#2)
4. Surface Emissivity. The refractive index of silicon at 0.63 m is (3.882, 0.019),
calculate the surface reflectivity, transmissivity, and emissivity of a semi-infinite
silicon wafer (a) at normal incidence, (b) at 30o angle of inidence, and (c) 60o
angle of incidence for both TE and TM waves. Also, estimate the penetration
depth for normal incidence.
5. Fresnel Formula for TE Wave. Derive the Fresnel formula for a transverse
electric wave incident onto a plane surface,
6. Tunneling of Photons. A vacuum gap of 0.2 m is formed between two glass
substrates. Plot the transmissivity of light from one glass substrate into another as
a function of angle of incidence for an incident TM wave at 0.5 m.

The

refractive index of the glass is taken as 1.46. Compare the results with the
situation if a thin film of glass of 0.2 m is sandwiched between vacuum.

2.58J HW#4
1. Optical properties of gold. The table below lists the optical constants of gold at
different wavelength. Use Drude model to fit these data.
(a) What is the plasmon frequency,
(b) What damping factor,
(c) Relate the damping factor to the electrical conductivity and check whether
the number makes sense,
(d) Plot your fitting and experimental data on a graph.
2. Temperature rise of nanoparticles in liquid under laser irradiation
Consider very dilute gold nanoparticles suspended in water. A laser beam at a
wavelength that is transparent to water is directed through the liquid. Because the
particles are very dilute, we can assume that every particle in the suspension will be
subject to the same laser intensity. We are interested in the temperature rise of a
nanoparticle above that of water, assuming water temperature does not change. Carry out
the following:
(a) Derive an expression of the nanoparticle temperature rise as a function of the
particle diameter, thermal conductivity of the liquid, the refractive indices of the
particle and the water, and the laser intensity.
(b) Determine an optimal wavelength that creates maximum temperature rise.
(c) Plot on graphs the particle temperature rise as a function of the laser intensity for
particle diameter values of 5 nm, 10 nm, and 30 nm at the optimal wavelength,
and 0.2 microns off the optimal wavelength (both sides of the wavelength). If you
can focus a laser beam to 5 m in diameter, what power do you need for a
temperature rise of 20 oC for 10 nm particles at the optimal wavelength?

2.58, Spring 2006

Problem Set 5
Problems 10.1, 10.7, 10.21, 10.29

2.58J
Homework No. 6
1. Solve Eq. (13.41) numerically to obtain temperature distribution between a parallel
plates for L=0.1, 0.4, 1.5, and 10. Make sure to compare your results with Fig. 13.3.
2.

(a)
Derive an expression for the heat flux based on the diffusion
approximation for radiation heat transfer between two concentric spheres of
radius R1 and R2, with emissivity 1 and 2, and temperature T1 and T2,
respectively. The medium in between the spheres is gray and isotropically
scattering with an extinction coefficient Ke.
(b)
Derive first-order temperature slip boundary conditions at both surfaces.
(c)
Obtain an analytical solution for the temperature distribution in the
spherical shell.
(d)
Obtain an analytical solution for the heat flux between the two spheres,
compare your results with that of Fig. 14-1.

3.

Problem 15.20.

2.58 Spring 2006

Midterm 1 Solutions
Question 1.
The peak wavelength of the solar radiation is around 500 nm. For a particle with a
diameter d less than 50 nm, Rayleigh scattering is a good approximation. The absorption
efficiency of the small particle is given by
Qa = 4x

m
Im 2
m

( 2 + 0.1i ) 1 0.133 d
1 4 d
Im
=
=
( 2 + 0.1i ) 2 + 2
+2

(1)

The solar irradiance that reaches the particle is

R
G = s
d se

eb (Ts )

R
= s
d se

C1

(2)

C2
exp
1
Ts

where Rs is the radius of the sun, dse the distance between the sun and the earth, and C1
and C2 are constants.
(a) The total solar energy absorbed by the particle is

qa =

d2
4

G Qa d

(3)

Plug Eqs. (1) and (2) into Eq. (3) to yield

0.133 2 d 3 Rs
qa =

4
d se
Let y =

Ts 5

C1

( Ts )

d ( Ts )

(4)

C2
exp
1
Ts

1
. Equation (4) becomes
Ts
2

0.133 2 d 3 Rs
0.133 2 d 3 Rs
y 51
5
qa =
dy =

C1Ts 0 C2 y

4
4
e 1
d se
d se

( ( 5 ) = 4! = 24 , and for 5 ,

( ) 1 .)

C1
( 5 ) ( 5 ) Ts 5
5
C2

(5)

(b) The scattering efficiency of the small particle is

2

m 2 1
8
8d
Qs = x 4 2
=

3
3 m +2

( 2 + 0.1i ) 1
2
( 2 + 0.1i ) + 2
2

d
= 0.673

(6)

Similar to part (a), the scattered energy is given by

2

0.673 5 d 6 Rs
8
qa =

C1Ts 0
4
d se

( Ts )

C2
exp
1
Ts

d ( Ts )

0.673 5 d 6 Rs
C1
=
( 8 ) ( 8 ) Ts
8

8
4
d se C2

Question 2.
(a)

1
glass

The reflectivity at the interfaces is given by

2

= 12 = 23 = r12

N 1
0.5 + i 2
= 2
=
N 2 +1
2.5 + i 2

(i) For 8 m , 2 = 0
12 = 0.04
The peak wavelength of the radiation from the oven is around 1.5
m, which is much less
than the thickness of the glass. We can use ray tracing to calculate the reflectance of the
glass slab.
Rslab =

1 +

(1 )
1

1
= 0.923
1+

Aslab = 0

= 0

2
= 0.077
1+

Tslab =

(ii) For > 8 m , 2 << 1

0.04

=e

2 d4 0

=e

100 0

= 0.285

Rslab =

1 +

2
(1 ) 2 = 0.043

1 2 2

2
1 )
(
=

= 0.262
1 2 2
= 1 Rslab Tslab = 0.695

Tslab

Aslab
= 0.695
(b)

Qe

Qconv

Qe

Qa

The glass view port absorbs irradiance from the oven and loses heat by radiation from
both surfaces as well as convection to the air. Since the area of the oven surface is much
larger than that of the glass, the oven can be treated as a blackbody source. The radiation
absorbed by the bottom surface of the glass is then given by:
4
Qa = A Toven

where A is the area of the glass surface (one side), and is the total hemispherical
absorptance of the glass, which is given by

eb d

4
Toven

8 m eb d
4
Toven

= 0.695 (1 f ( 8 m

oven

) ) = 0.018

where f ( T ) can be found in the table of blackbody emissive power.

Similarly, the total hemispherical emissivity of the glass is given by

= 0.695 1 f (8 m

glass

))

Apply energy balance to the glass view port:

4
4
Toven
= 2 Tglass
+ h(

glass

T ) = 1.39 1 f ( 8 m

glass

))T

4
glass

+ h (Tglass T )

where we neglect the absorbed radiation from the low temperature environment. Plug in
numbers and the above equation becomes

4
1.633 104 = 1.39 1 f ( 8 m Tglass ) Tglass
+ 10 ( Tglass 293)

(c) Iteration with the table value for f, we get, Tglass 930K

2.58J MIDTERM EXAM No.2

Due May 18, 2006, 11:00 am
Instruction:
This is a take home exam. You should complete the problems independently. The only person
you may consult is the TA. The last question is a bonus.

Question 1. Consider radiative heat transfer between two parallel plates (infinite wide and deep)
maintained at two temperatures T1 and T2, as shown in the following figure. The medium in
between the plates is gray, isotropically scattering, absorbing, and emitting with an extinction
coefficient Ke. The emission of both plates is diffuse-gray with an emissivity . The reflection
of both surfaces is gray but specular with a bidirectional reflectivity that is independent of
angle. The medium is in radiative equilibrium, i.e., neglect molecular heat conduction.
(a) Derive a relation between and .
(b) Derive an integral equation that determines the temperature distribution in the medium.
You do not need to solve this equation, but the integral equation should be expressed in
terms of the local blackbody emissivity power (and hence the local medium temperature)
and other given boundary conditions.
(c) Assuming that the distribution of the local blackbody emissivity power is known, derive
an expression for the heat flux between the two parallel plates.

,T1
,

Absorbing, Emitting,
Isotropically Scattering
Extinction Coefficient Ke

,T2
,

Question 2 (this is similar to problem 1 but seeks an approximate solution): Consider radiative
heat transfer between two parallel plates (infinitely wide and deep) maintained at two different
temperatures T1 and T2, as shown in the following figure. The medium in between the plates is
gray, isotropically scattering, absorbing, and emitting with an extinction coefficient Ke. The
emission of both plates is diffuse-gray with an emissivity . The reflection of the surface is gray
but specular a bidirectional reflectivity independent of angle. The medium is in radiative
equilibrium, i.e., neglect molecular heat conduction.
(a) Apply the idea of diffusion approximation, derive appropriate temperature jump
boundary conditions for the given properties of the surfaces.
(b) Combine the diffusion approximation and the boundary conditions you obtained above,
derive an analytical solution for the temperature distribution inside the medium and the
heat flux between the plates.

,T1
,

Absorbing, Emitting,
Isotropically Scattering
Extinction Coefficient Ke

,T2
,

Question 3 Consider a colloidal solution consisting of nanoparticles of diameter d=10 nm,

suspended in water. A laser beam with a power density of 106 W/m2 and a wavelength
=1.064 m is directed through the water. The volume fraction of the particles is 0.1%. The
water is sandwiched between two transparent parallel plates that have a refractive index equal to
that of water. You can neglect the reflection between water and glass plates.
(a) Determine the rate of energy deposition (per unit area of incidence) into the colloidal
system due to the absorption of the laser beam.
(b) If the nanoparticle diameter is 5 nm while the volume fraction remains 0.1%, what is the
rate of energy deposition?

Laser
Wavelength
1.064 m

1 cm
Particle diameter 10 nm
nm

Optical constant (3,0.01)

(3,0.01)

Water and hold optical constants: (1.3,0)

(1.3,0)

Question 4 (Bonus). A very long narrow channel along the x-direction is formed by two
parallel plates (infinitely deep perpendicular to the paper) separated by a distance D apart, as
shown in the following figure. The channel is filled with a gray, isotropically scattering,
absorbing and emitting medium. The effective extinction coefficient of the medium is Ke. A
radiation heat flow Q [in Watt] is established in the x-direction due to the internal temperature
gradient along the x-direction. There is no net heat flux in other directions. Both plates are
black surfaces with a surface temperature that is equal to the local temperature of the medium.
We know that KeD is of the order of one, and thus you cannot neglect the boundary effects in the
y-direction. However, the channel is long and you can neglect the boundary effect along the xdirection. Neglect heat conduction in the medium. Derive an expression for the local heat flux
inside the channel in terms of the local temperature and its local gradient.
y

Black Surface
Absorbing, emitting, and sotropically scattering
D

x
Black Surface

Chapter 6 OPTICAL PROPERTIES OF SOLIDS

We will investigate how to calculate the dielectric constants of solids. For this purpose, we
will introduce classical models. They have the advantage of easy to understand.
6.1 Lorentz Model
When bounded electrons or lattice interact with electromagnetic field, they generally
oscillate around their equilibrium position. In a quantum mechanical picture, these changes of
energy mean that system transits between one energy state to another. Classically, we could
approximate the oscillation by a damped oscillator. A damping force exists because various
collision processes (electron-electron interaction, electron lattice interaction, etc.) extract energy.
We can write done various force acting on the oscillator as
Driving force = -eEx=-eEoe-it
Spring force=-ksx
Damping force=-v=-x'
where Ex is the local electrical field, v the oscillator speed, and a negative sign has been added in
front of the electron charge.
The Newton's second law gives us
mx"=-x'-ksx-eEx

(1.1)

Since the electrical field is harmonic, Ex=Eoexp(-it), we can rewrite the above equation as
x"+x'+o2x=-eEoe-it/m

(1.2)

where =m, and ks=o2m. o is the natural frequency, and is the damping factor. To find the
solution, we assume
x(t)=Ae-it
Substituting (1.3) into (1.2), and solve for A, we get

(1.3)

A=

eEo
1
2
m o 2 i

(1.4)

and

e i t
eEo
x(t) =
m o2 2 i

(1.5)

Recall that ex is the polarization of one oscillator. The total polarization per unit volume is thus

p
N e2 E
e it
= 2
oE
P=Noex = o o 2
2
m o i o 2 i

(1.6)

N o e2
=
m o

(1.7)

where
2
p

We will see the origin for this name later.

The polarization is related to the electrical field through the electrical permittivity,
P=xoE

(1.8)

2p
x= 2
o 2 i

(1.9)

So we have

The relative dielectric constant (=or) is

r = 1 + = 1 +

2p
= 'r + i "r
2
2
o i

where the real and the imaginary parts are

(1.10)

 2p ( o2 2 )
r = 1 + 2
( o 2 )2 + 2 2

(1.11)

2p
i = 2
( o 2 )2 + 2 2

(1.12)

The complex refractive index is calculated from r as

N = r

(1.13)

n2-k2=r'

(1.14)

2nk=r"

(1.15)

or

Classically, o is the resonance frequency of the simple harmonic oscillator. Quantum

mechanically, the is the energy difference between the final and the initial states. If we have
multiple oscillator, the Lorentz model can be written as

r = 1+

2
j oj

2pj
2 i j

(1.16)

Drude Model

For metals, there is no spring to connect free electrons to ions, so o=0. From the Lorentz
model, we get

r = 1

2p
2 + i

(1.17)

The real and the imaginary parts are

r ' = 1

2p
2 + 2

(1.18)

and

r"

2p

2 + 2

(1.19)

These results are called Drude model. Generally, p>. If >p, we can see that r"-->0, which
means that -->0. At high frequency, there is no absorption. A metal becomes transparent! The
reason is that at this frequency, the electrons in the metal cannot react fast to the incident
electrical field.

Example: Self-Extinction of a Laser-Induced Plasma

In laser material processing, the ablated and/or evaporated materials are often ionized. An
ionized gas is called plasma. If the laser frequency is smaller than the plasma frequency, there
will be high reflection of the laser beam and the laser energy cannot reach the target. We always
want to operate in the transparent region,
> p =

N o e2
m o

(1.20)

This leaves an upper limit for the total number of the ionized particles in the plume
No <

2 m o
e2

(1.21)

Hagen-Rubens Equation

We can further relate the metal dielectric constant to it electrical conductivity. From our
derivation or the Lorents model, we have the electron location as a function of the time. We can
calculate the electron speed as
eEo (i )e it
v=x=
m 2 i

Let's consider the DC limit, -->0, then

(1.22)

v=

eE

(1.23)

The current flux density can be calculated from the electron velocity
J=Noev=dcE

(1.24)

From Eq. (1.23) and (1.24), we can get calculate the damping factor from its dc conductivity
2

dc

Ne 2 p o
=
=
m

(1.25)

We can use the above relation to calculate and substitute into the Drude model to calculate n
and . A further approximation is obtained observing the Drude expression, when -->0, we
have

2p
r ' = n = 1 2
1 2
+ 2

r"

2p

2p
2p
= 2n =

( 2 + 2 )

(1.26)

(1.27)

From this, we can see that

2p
0.003
dc
nk=
=
=
2
2 o
dc

(1.28)

The above relation is called Hagen-Rubens relation.

Induced Field Effects

In the derivations of the Lorentz model and the Drude model, we assume that the local electrical
field acting on an atom equals the externally applied electrical field. By doing so, we assumed
that the neighboring atoms have no effect on the atom we are considering. We can then calculate
an the electrical dipole momentum of the atom caused by the external field. But such a dipole

will generate electrical field itself. The generated electrical field is superimposed with the
external field and applied to other atoms. Similarly, the atom we are considering also experience
the electrical field generated by other atoms in the surrounding. So, strictly speaking, the
electrical field acting on the atom, Elocal we are considering consists of two parts, one is the
external electrical field Eex, the other is the field induced by the polarization of other atoms.
Elocal=Eex+Eind

(1.29)

It can be proven that the induced field is related to the polarization as

Eind=P/3o

(1.30)

( 0 )Eex = P

(1.31)

We also know that,

From the above three equations, we obtain,

Elocal =

+ 2 o
Eex
3 o

(1.32)

Now let's revisit the Lorentz model, we have obtained that

p2
P= 2
E = o Elocal
o 2 i o local

(1.33)

We also have, from definition,

D = Eex = 0 Eex + P

(1.34)

Thus,

o =
or

+ 2 o
3

(1.35)

 2pj
r 1 ( n + ik )2 1 1
=
=
r + 2 ( n + ik )2 + 2 3 j oj2 2 i j

(1.36)

where the summation is a generalization to several oscillators as we generalization the Lorentz

model. The above relation is called Clausius-Mossotti relation. Its significance lies in that it
relates the dielectric constant, which is a macroscopic quantity, to the microscopic quantity of
the oscillators. It can be further shown that the above equation can also be written as

r = 1+

2
j oj

2pj
2 i j

(1.37)

This expression is identical to our derivation for the Lorentz model except that we changed o to
o, their relation is
2oj

2oj

2pj
3

(1.38)

But this does not matter for most experiment since the value of o is often obtained by matching
the experimental results with the models. When <<o, we can neglect the terms in the
denominator. In this case, the right hand side of Eq. (1.37) is independent of frequency and
linearly proportional to the square of the plasma frequency. But since square of the plasma
frequency is proportional to the number of oscillators, which is proportional to the density, we
have

r 1
= cons tan t
r + 2

(1.39)

In this case, the frequency is far from the resonant frequency, the absorption is zero, we can
rewrite is as
n2 1 1
= constan t
n2 + 2

(1.40)

Lorenz-Lorentz equation. This equation gives us a relation between density and the refractive
index. It is particularly useful for some experiment which use the change of the refractive index
caused by density change such as the shadow graph, the interferometer methods to measure

temperature distribution. For gases, we can even approximate the expression future because
n~1. We have
n 1

= constan t

(1.41)

Debye Relaxation
There is another mechanism for polarizing matter containing permanent, as opposed to induced,
electric dipoles. Such permant dipoles already exist in the materials such as the water molecules.
Only these dipoles are randomly oriented. When an electromagnetic field is applied, these
dipoles will be realigned towards the direction of the electromagnetic field. Unlike the Lorentz
model, in which a spring will drag a polarized oscillator back to it equilibrium position, the
restoring force is random thermal motion and it only tries to randomize the dipole orientation.
To further understand the difference, imagine that we have an oscillation electromagnetic field
that applied to a Lorents oscillator and a polarized molecule. When the electromagnetic field is
turned off, the oscillator in die down in an oscillating fashion, but for the permanent dipole, the
change is no external force to force its dipole into a specific direction, it will try to randomize its
direction. Such a randomization is called relaxation. We could express the relaxation process as

Pd(t)=Pd(0)e-t/

(1.42)

-relaxation time or time constant. For fluids, this relaxation time can be calculated from a
simple model, which consider the relaxation time of a sphere of radius d in a fluid of viscosity ,

d 3
2 BT

(1.43)

Based on the described process, we can also relate the polarization with the external field and get
an expression for the dielectric constant, which is

= 1 + v +

D v
1 i

(1.44)

where v is the electrical susceptibility at high frequency side (due to lattice vibration) and cD is
the electric susceptibility at low frequency side. We can further write the expression in terms of
n and k as

n k =

nv2

n2D nv2
+
1 + 2 2

(1.45)

(n2D nv2 )
2 n =
1 + 2 2

(1.46)

The Debye relaxation generally occurs at very long wavelength (microwave frequency) because
an order of estimation of the relaxation time shows that is in the 10-10-10-12 s, which
corresponds to microwave frequency. At higher frequency, the molecules cannot follow the fast
oscillation of the electromagnetic fields. Now we should have a full picture of the dielectric
constant of materials. Please look at the extinction coefficient of water.
Kramer-Kronig Relations

Our discussion on the optical properties of solids will be complete with one more topic, this is
the Kramer-Kronig relation. This relation is a mathematical result that relative the real and the
imaginary parts of any complex functions, including the dielectric constant, the refractive index,
the complex reflection coefficient. In fact, from our previous discussions, we can see that once
dielectric function is determined, the real and the imaginary parts are automatically determined.
A more fundamental result is that as long as the function is an analytical function, their real and
imaginary parts are interrelated. The relation between the real and the imaginary parts for
dielectric constants are called Kramer-Kronig relations. We can write such a relation for the n
and as
n( ) = 1 +

k( ) =

'

k( ' )
2 d
0 '

n( ' )
2
2 d
0 '

(1.47)

(1.48)

where P signifies the Cauchy principal value of the integral because the denominator in the
integrand experience a zero,
' k( '

0 '

2 d

= lim

' k( ' )
' k( ' )
2
2 d + lim
2
2 d
'
+a '

(1.49)

We can also write a similar relation for the reflection coefficient. From our discussion on the
Fresnel reflectivity, as know that
R = | r12 |2

(1.50)

Or we could write the reflection coefficient as

r12 =

R( )ei ( )

(1.51)

In practice, we can measure the reflectivity easily but not the phase. Since r12 is an analytical
function, its amplitude and phase also obeys the Kramer-Kronig relation. We can use the
relation to calculate the phase from measured reflectivity

( ) =

R( ' )
d '
' 2

ln

(1.52)

Once we calculate the phase, we know the reflection coefficient, from which we can inverse the
Fresnel relation to get n and k. The advantage of the Kramer-Kronig relation is that it does not
depend on any mathematical model for the dielectric constants. The disadvantage of the relation
is that the integration is always over the whole spectrum. In rarely cases, we can measure one
property over the whole spectrum. What is common in practice is that people measure
reflectivity in certain frequency ranges, and for the reflectivity out side this frequency range,
they use some model based on our previous discussions. Then they use the Kramer-Kronig
relations for the consider range to derive n and k in this range. The idea is that outside the
measurement region, the accuracy of the guessed function does not affect the accuracy of the
measured range because the denominator becomes vary large and the contribution from their
integration is small.