1

S o cEo

(16)

<S> is the average energy per square meter per unit

time passing through a surface normal to the
direction of propagation.

1
2
o cEo <S>
2
Eo

E Eo e j t

is also called the intensity I

is the amplitude of the electric
field of the incident wave

Fig. 3.10 Incident plane-wave radiation of frequency

When an atom characterized by a resonance frequency o, is placed

in a region where there is a bath of electromagnetic radiation, the

j t
radiations electric field E Eo e
will drive the atoms charge qe up
and down; that is, it will accelerate the charge thus causing the atom
to re-emit electromagnetic radiation. This process, which occurs at
any frequency , is called scattering. That is,

scattering is the process by which energy is

absorbed by an atom from the incident radiation
field and re-emitted in all directions.

E Eo e j t

x [ xo e j ] e j t

Fig. 3.11 Incident light is absorbed and (re-emitted) scattered by an atom in

all directions.

14

(17)

Lets calculate the energy absorbed (hence re-emitted) by the charge

Actually the energy emitted by the accelerated charge has already been calculated
in expression (14) above, except that we need to find out the amplitude of oscillation

xo ; the latter will depend on the electric field amplitude Eo, as well as the frequency
of the incident radiation. In other words, lets calculate the relationship between Eo, ,
and

xo .
Scattered (re-emitted)
2
2
light
qe xo
P
4
3
12 o c

qe, me ,o

The spring constant is

2
chosen according to k

Incident
radiation
()

Atom model as an
2
oscillator of natural me o
resonance frequency o.

Fig. 3.12 Atom modeled as an oscillator of natural frequency o. The ability of the
oscillator to absorb energy from the incident radiation depends on .

To find xo , lets model the atom as a damped harmonic oscillator. Accordingly the
equation of motion for the charge qe is given by,

me

d 2x
dt 2

me

dx
j t
kx qe Eo e
dt

Here the term me

(18)

dx
accounts for the presence of a dissipation energy
dt

source, which, in our case, can be identified in the loss of energy due to the
electromagnetic radiation by the accelerated charge.
A stationary solution of (18) is given by,

x [ xo e j ] e j t

(19)

where

15

xo xo ()

(qe / me ) Eo

)
2 2

1/ 2

Amplitude of oscillation
as a function of frequency

(20)

and

tan 1

o 2 2

(20)

Expression (20) indicates that the amplitude of oscillation xo (and hence the
acceleration) of the charge depends on the incident radiations frequency .
Lets proceed now to calculate the total power radiate by the accelerated charge under
the influence of an electric filed of amplitude E0 and frequency . Replacing the value of
qe 2 xo 2 4

xo given in (20) for xo into the expression for the radiation power P
12o c 3
given in (14), one obtains, P

qe 2
12o c 3

(qe / me ) 2 Eo 2
(o 2 2 ) 2 ( ) 2

. Rearranging

terms,
qe 2
1
8
4
2
P o cE o 2 (
)
= <P>()
2
3 4o me c 2
(o 2 2 ) 2 ( ) 2

(21)

Expression (21) gives the total average energy emitted by the charge qe when subjected
to a harmonic electric field (given in expression (15) ) of amplitude E0 and frequency .
Notice the expression

1
2

ocEo 2 (incident energy per unit area per second, i.e. the

incident intensity Io) has been factored out in expression (21). This is convenient, for it
allows to interpret (21) the following way: Out of the incident intensity Io present in the
2
8
qe
4
2
(
)
cavity, a fraction of it equal to
is present in the
2
2
3 4 o mec
(o 2 )2 ( )2
form of scattered power. We say fraction because the units of that last expression is
area (not a simple fraction number). Hence, it is better to interpret (21) in terms of
scattering cross section).
Note (dated 09-2012).
Expression (21) quantifies the amount of energy that the the atom is able to
re-radiate (due to the fact that is is a charge) upon the incidence of an
harmonic electric field of amplitude Eo and frequency .

16

It has nothing to do with the ability of the atom to capture energy from
radiation in the cavity (as the concept of scattering cross section may
erroneously suggest). Hence, be careful with the proper interpretation of
the scattering cross section concept.) Andres

The concept of scattering cross section

If we considered a hypothetical cross section of area intersecting the incident
radiation, the amount of energy per second hitting that area would be

1
2
P [ o cEo ] I
2

(22)
cross section
of area

1
2
S o cEo
2
<S> is called the
light intensity I

Fig. 3.13 Pictorial representation of scattering cross section. Notice,

this has nothing to do with the size of the atoms nor the spatial
distribution of atoms inside the cavity. It is just a measurement of the
ability of the atom (once radiation impinges on it) to radiate energy
in all directions.

One can use the analogy of an affective area being intercepted by the incident radiation
to define how effectively the radiation is absorbed and scattered (i. e. re-emitted) by an
atom. In effect, comparing expressions (21) and (22), the total power scattered by an
atom is numerically equal to the energy per second incident on a surface of crosssection area

scattering ,

1
2

P scattering [ o cE o 2 ] scattering

(23)

where

scattering

scattering

qe 2
8
4
2
(
)
3 4o me c 2 (o 2 2 ) 2 ( ) 2

has units of area.

17

(24)

scattering

Fig.3.14 Sketch of the atoms scattering cross section.

Note (Dated 09-2012)

scattering

is an indicator of the ability of the atom scatter

light once it is excited by an harmonic electric field of amplitude Eo and

frequency w. We cannot ask, we cannot expect, the atom to scatter more
(or less) than Io (where Io.is the intensity associated to the filed incident
on the atom).
Expression (24) indicates that the closer is to o, the higher the reemitted energy.

3.1.B.c Electromagnetic Radiation Damping: What is the value of ?)

We address here the fact that
the term me

dx
in Eq. (18) (the term in the differential equation that takes into
dt

account the energy dissipation,)

should be compatible with
expression (21) (that gives the electromagnetic energy dissipated by the oscillator.)
We should require then that these two expressions be consistent with each other.
Indeed,

On one hand, the power dissipate by a oscillator is given by

[force]x(velocity) = [ me
use the expression for

dx dx
](
) = [ me ( jx) ] (
dt dt
j

jx ) = = me 2 x 2 . Here we

x(t ) given in (19) x [ xo e ] e j t .

The average value of the dissipated power will be,

(1 / 2)me 2 xo2 .

18

On the other hand, according to (14), the emitted electromagnetic power is,
q 2 xo 2

12o c 3

The last two expressions should be equal.

(1 / 2)m

2
o

q 2 xo 2
4
3
12 o c

q2
2 . Rearranging terms,
This allows to identify
3
6 o mc
qe 2
2
3 c 4o me c 2

electromagnetic
radiation damping
constant

(25)

For practical purposes, however (given the very narrow bandwidth of the cross section
() shown in Fig 3.10 above,) will be typically end up being evaluated at =0,(i.e.
the narrow bandwidth of () tell us that most of the physics happens around =0.)
Rate at which the oscillator looses energy
(A more detailed description of this section is given in the supplementary Appendix-3 of
this chapter.)
Let

W W (t )

be the average energy of an oscillating

(26)

charge at a given time t .

If the oscillating charge is left alone to oscillate, its amplitude of vibration will die out
progressively as the oscillator looses its mechanical energy by emitting electromagnetic
radiation.
If the motion of the oscillator is alternatively modeled by a mechanics equation of
motion me

d 2x
dt

me

dx
j t
kx qe Eo e
, it can be calculated that the rate at which
dt

the oscillating charge looses energy is given by,

dW
W
dt

(27)

with its corresponding solution

W (t ) Wo e t

(28)

19

As an example, an atom that has a resonance frequency corresponding to = 600 nm,

would have a damping constant of ~ 108 s-1. That is, the radiation will effectively dye
out after ~10-8 s (or after ~ 107 oscillations.)

3.1.C Radiation and thermal equilibrium

Lets consider an atom enclosed in a box made of mirror walls which contains
electromagnetic radiation. Radiation re-emitted by the atom remains inside the box
undergoing multiple bounces on the mirror walls. Lets further assume that the
temperature of the whole system is T.

Box at temperature T

Scattered (re-emitted)
light

q
Incident
radiation
Atom
(modeled as an oscillator)
Fig. 3.15 Schematic representation of an atom as an harmonic oscillator that radiates
energy. The atoms absorb energy from the electromagnetic radiation existent inside the
box (the latter assumed to be made of perfectly reflecting walls.)

How to make the temperature T intervene in an expression like (14) that gives
the power scattered by an atom in the form

P scattering

q e 2 xo

12o c

4 ?

It is plausible to assume that the equilibrium temperature should correspond

to proper value of the amplitude of the electric field,

Eo , since the higher the

value of Eo , the higher the charges amplitude of vibration xo, the greater
temperature to be associated with the atom (i.e. the amplitude of the oscillator
should increase with temperature.)
If our assumption were correct, how to find then the proper value of
corresponding to a given temperature T?

Eo

Aiming to find a proper answer lets outline some considerations:

- If an atomic oscillator had no charge, it would oscillate forever. It would have
an average energy W compatible with the temperature in the box; that is

20

W W ( T ) . In other words, it would oscillate forever with an amplitude

settled by the temperature T in the box. k xo2 = kT or mo2xo2 = kT.
-

But our atomic oscillator is charged. If it were left alone, its amplitude of
vibration xo would die out progressively, as the oscillator looses its energy by
emitting electromagnetic radiation.

If our atomic charged oscillator were in physical contact with other atoms,
energy in the proper amount will be supplied by their mutual collisions as to
keep the same temperature among themselves.
Here, however, we will consider that the energy is supplied via
electromagnetic interaction: the atom draws energy from the radiation existent
in the cavity to compensate the energy being lost by radiation (accelerated
charged particles emit radiation.) When this compensation of energy
matches, then we are at an equilibrium situation, which inherently should
occur at a given temperature (the latter settling the charges amplitude of
oscillation at a corresponding value.) We will retake the subject of
temperature dependence of x0 in Section 3.1.c.b below.

3.1.C.a Light intensity spectral density I() at equilibrium

To formalize the equilibrium situation we have to keep in mind that radiation
of different frequencies might be present in the cavity. It is convenient, then, to
introduce the concept of spectral density I():
I()
I() d

= light intensity spectral density

= light intensity of frequency within a range (,
+d inside the box
= contribution to the average electromagnetic energy per
square meter per unit time passing through a surface normal
to the direction of propagation from radiation components of
frequency within a range d

[ Intensity ] J /(m 2 s)
J
Units of [I() ] =
=
2
[ ]
1/ s
m
Before establishing the condition of equilibrium, lets make two pertinent
observations:

According to (21) and (22), the total power that the atom is able to emit
(scatter) is given by,

P ()d
0

I ( ) ( w)d ;

21

It is an integral because the atom is exposed to all the frequencies

existent in the cavity. However, the maximum emission of power
occurs when is close to to (because ( ) has a sharp peak at
.)
On the other hand, this same amount of emitted power (i. e.

I ( ) ( w)d ), can be seen from the perspective of a system loosing

d [W ( )]
( )[W ( )]
dt
characterized by a damping constant . On the other hand, the
requirement of compatibility between i) power re-radiated by the atom,
and
ii)
a
simple
damping
harmonic
oscillator
model
energy

me

d 2x
dt

due

me

to

damping

dx
j t ,
kx qe Eo e
dt

process

lead

to

expression

(25)

qe
2
( )
2 . But since all the dynamics occurs at , that
2
3 c 4 o me c
dW
(o ) W , with the
is W()~0 for , we can use
dt
interpretation that W is the total energy of the atom.

Formalization of the thermal equilibrium condition:

How much light intensity spectral density I() there must be inside the box at
temperature T for,
the electromagnetic energy re-emitted by the oscillator (which should
come from the radiation bath in the cavity) per unit time

dW
P ( )d I ( ) ( )d
0
0
dt

(29)

to be equal to

the energy lost by the oscillator per unit time

dW
W
dt

(30)

Average energy of the

oscillator at temperature T

22

Box at temperature T

qe

Fig. 3.16 Atom of natural frequency 0 in

a bath of electromagnetic radiation of
spectral density I().
Lets evaluate the integral that appears in (29).
Since the expression for () peaks at o then extending the integral
down to does not cause any significant change (this is done just to
facilitate the calculation )

I ()()d I ()()d

qe
8
4
2

(
)
2
2
3 4 o me c
(o 2 ) 2 ( ) 2
2

where

scattering

For the same reason that only the values of very close to
significantly contribute to the integral we can picture in our mind that

I ()()d

I ()()d

A(k)

Fig.3.17 Sketch of the atoms scattering cross section and the spectral
density light intensity present in side the cavity

23

o will

Accordingly the following approximations can be considered appropriate,

2 o ( o )( o ) (2o )( o )

o
4

2
2
(o 2 ) 2 ( ) 2 [o 2 ]2 ( o ) 2

[(2o )( o )]2 ( o ) 2

I ( ) I (o )

4( o ) 2 2

All these approximations lead to

I ()()d I ()()d

2
8

qe
4
2
I ( ) (
)
d
2
2
2
2
2

(o ) ( )
3 4 o me c

2
2
8
qe
o
2
I ( ) (
)
2

4( o ) 2 2
3 4 o me c

qe
2
1
2
2
I (o ) (
)

o
( o ) 2 ( / 2) 2 d
3 4 o me c 2
dx
1
x
arctan )
(using 2
2
x a
a
a

I (o )

qe
2
2 1
(
) 2 o
( ( )
2
3 4 o me c
/2 2
2

I (o )

qe
2
2 2
(
) 2 o
2
3 4 o me c

2
I ()()d I ( )
(
o

24

qe
)2 o 2
2
4 o me c

(31)

At equilibrium we should have,

I (o )

22 (
3

dW
dW
I ()( w)d =
W , which leads to
0
dt
dt

qe
) 2 o 2 W
2
4 o me c

4 o me c 2 2 2
3
I (o )
(
) 2W
2
22
qe
o
2
q2
2 evaluated at
2
3 c 4 o m c

4 o me c 2 2 2
3
q2
2
(
)
(
) 2 o W ,
2
2
2
3 c 4 o m c
2
qe

or

Using expression (25) for the value of

one
I (o )

I (o )

obtains,
3
2
2
( ) 2 o W ,
2
2 3 c

I (o )

1
2

o W
32c 2

(32)

Average energy
of the oscillator

I() is the light spectral density at =.

Here is the natural frequency of the oscillator we were focusing in. Had we used an
oscillator of a different natural frequency, lets say , we would have obtained a
similar expression (32) but with instead of . Hence, in general,

I ()

1
2 W
2 2
3 c

(32)
Average energy of the
oscillator at temperature T

light intensity
spectral density
I( )
Required light intensity spectral density I() inside the
box at temperature T , in order to maintain equilibrium
inside the cavity.

25

I( ) =
I( )

Notice
Units of [I() d] = [Intensity ] =
Units of [I() ] =

J
m2s

[ Intensity ] J /(m 2 s)
J
=
2
[ ]
1/ s
m

It is worth to highlight that,

Expression (32) has remained undisputed. That is, it is still considered correct even
when the new quantum mechanics concepts are introduced.

It is in the calculation of the average energy W where the classical and quantum
approaches fundamentally diverge.

3.1.C.b Classical calculation of the atoms average energy W.

In classical statistical mechanics there exists a very general result so called
equipartition theorem, which states that the mean value of a quadratic term in the
energy is equal to kBT. Here kB is the Boltzmanns constant and T is the absolute
temperature.
The Boltzmann distribution
The equipartition theorem can be obtained from the Boltzmanns probability
distribution for a small system A in equilibrium with a (huge) reservoir at temperature
T. The Boltzmann distribution states that the probability that the system S be found in a
state of energy

is proportional to
E / kBT

E / kBT

; that is ,

P( E ) e

Ce E / kBT

probability to find the system

A in a state of energy E

26

(33)

Energy
exchange

P(E)
Boltzmann
distribution

Reservoir at
temperature
T

Small
system A

Energy E

Fig 3.18 Left: A system interacting with a thermal reservoir. Right: Boltzmanns
distribution to find the system in a state of energy E.

The values of E could go from 0 to infinity (the reservoir being in charge of keeping the
temperature constant); but, as the expression above indicates, the states of lower
energy have a higher probability.
Since for a given energy there may be several states characterized by the same energy,
it is usual to define,
g ( E )dE
number of states with energy
(34)
E , within an interval dE ,
thus giving

Ce

E / kBT

g ( E )dE

probability to find the system A in a state

of energy between E and E+ dE,

which suggests to rather identify a probability-density

P ( E)dE Ce E / kBT g ( E )dE

P (E ) defined as follows

probability to find the

system in a state of energy between
E and E+ dE,

(35)

with C being a constant to be determined.

Since the probabilities added over all the possible states should be equal to 1, we must
require,

Ce

E '/ kBT

C 1 = e
0

g ( E ' )dE ' 1 , which gives,

E '/ kBT

(36)

g ( E ' )dE '

A self-consistent expression for P(E ) is therefore given by,

e E / kBT g ( E )dE
P ( E )dE E '/ k T
0 e B g ( E ' )dE'
(Notice in the denominator we are using a dummy variable

(37)

E ' .)

From expression (37) we can formally calculate the average energy of the system,

27

 E / kBT

g ( E )dE
Ee

E '/ k T
g ( E ' )dE '
e
0

(38)

The Equipartition Theorem

It turns out, very often the energy of the system may contain a quadratic term.
Consider for example
2

p
1
E x kx2 ...
2m 2

and we would like to calculate, for example, the average value of the kinetic energy
2

px
term
alone. As we know, being the system in contact with a heat reservoir, the
2m
2
px
value of
is sometimes high, sometimes it is low because it gains or looses energy
2m
from the heat reservoir; we would like to know what would be its average value
2

px
2m

.
2

px

2m

p
2
p x 2 mx / kBT
e
dp x
2m

(39)

px / kBT
2m

dp x

Lets call

In terms of

1
k BT

(40)

expression (39) becomes,

2

px

2m

p
2
p x 2 mx
e
dp x
2m

p 2

x
2m

dp x

p
2 mx

e
dp x

px

2m

dp x

p
2

px
2 mx

ln e
dp x
2m
0

p
2 mx

e
dp x
0

px

2m

dp x

28

(41)

Defining the variable u

px ,
2m

2
px
2m u2
1
ln
e
du ln
2m eu du

0
2m
0

px
1
ln 2m eu du
ln
0
2m

Term independent of

px
1
1

ln

2m
2
2
2

px
1
k BT
2m
2

(41)

Had we chosen any other quadratic term of the energy we would have obtained
the same result. This is the equipartition theorem. It states,

If the energy W of the system has the form

2

p
1
E x kx2 ...
2m 2

(42)

the average value of each independent

quadratic term is equal to

W E

1
k BT .
2

1
1
kT kT ...
2
2

The ultraviolet catastrophe

Lets assume there are f different quadratic terms in the expression for the total
energy W (translation motion, rotational motion, , etc.). The equipartition theorem
1
1
leads to W f k BT . Using this result in expression (32) I () 2 2 2 W , one
2
3 c
1
1
obtains I () 2 2 2 f k BT , or,
3 c
2

I ()

f k BT 2

6 2 c 2

29

classical prediction

(43)

(that assumes <W> ~ kT)

I()
Classical
prediction

Experimental
results

Frequency
Fig. 3.19 The serious discrepancy between the experimental results
and the theoretical prediction is called the ultraviolet catastrophe.

3.1.D

The birthday of Quantum Physics: Plancks Hypothesis to calculate

the atoms average energy W

To bring the theoretical prediction closer to the experimental results Planck

considered the possibility of a violation of the law of equipartition of energy described
above, expression (42).

1
2 W , BUT with the
2 2
3 c
average energy of the oscillator W not being constant (as the equipartition theorem
predicts) but rather being a function of the frequency, W () , with the following
requirements,
The starting expression would still be expression (32) I ( )

2 W ()
0
0
and

(44)

2 W () 0

For the statistical calculation of

W,

Planck did not question the classical Boltzmanns

Statistics described in the section above; that
Theory would still be considered valid.

(45)

Planck realized that he could obtain the desired behavior expressed in (44) if,
rather than treating the energy of the oscillator as a continuous
variable,
the energy states of the oscillator should take only discrete step
values:

0 ,

(46)

the energy steps would be different for each frequency

30

= ()

(47)

where the specific dependence of in terms of to be determined

Incident
radiation

Planck postulated that

the energy of the
oscillator is quantized
Fig. 3.20 An atom receiving radiation of frequency , can be
excited only by discrete values of energy 0 , , 2, 3,

According

to

Planck,

Wclassical E

in

the

classical

integral

expression

(37)

E / kBT

g ( E )dE
Ee
, one would have to replace:
E '/ k T
e
g
(
E
'
)
dE
'

g(E)dE [ g(E)dE gives the number of states with energy E

within an interval dE ]

dE

dE

n 0

thus obtaining,

WPlanckl () E

E e
n 0

En / kBT

En / kBT

n 0

where En = n() ; n= 1 2, 3,
A graphic illustration can help understand why this hypothesis could indeed work:

First we show how classical physics evaluate the average energy.

classical

E [ P( E )] dE = area under the curve of E P(E) vs E

0

31

(48)

E P(E)

P(E)

Classic calculation:
<E>= continuum addition
=Area (integral)

Boltzmann
distribution

Energy E

Energy E

Fig. 3.21 Schematic representations to calculate the average energy of

the oscillator under a classical physics approach.

Using Planks hypothesis E

E P( E n )
Planck n
n 0

Case: Low frequency values of

For this case, Planck assumed should have a small value (for the reasons
explained in Fig. 3.17).
small value

(for small values of

P(E)

(49)

E P(E)

Quantum calculation:
discrete addition

Boltzmann
distribution

Energy E

<E>=
=Area

~kB T

Energy E

Fig. 3.22 Schematic representations to calculate the average energy of the

oscillator assuming the oscillator can admit only discrete values of energy,
for the case where the separation between contiguous energy levels being
of relatively low value.

Indeed, comparing Fig. 3.21 and Fig 3.22 one notices that if is small then the

value of

E P( E ) will be very close to the classical value.

n

n 0

It is indeed desirable that Plancks results agree with the classic results at low
frequencies, since the classical predictions and the experimental results agree
well at low frequencies (see Fig. 3.19 above.)
Case: High frequency values of

32