Quantization of Normal Modes

2
2.1

Wave equation

Want to review modes of electromagnetic radiation in cavity. Start with

Maxwells equations in free space(SI units)

E
B
B
E+
t
E
B 0 0
t

= 0
= 0

(1)
(2)

= 0

(3)

= 0

(4)

= i + j + k
x
y
z

(5)

where is gradient operator

Will also need subsidiary condition, charge conservation:

+j=0
t

(6)

F = q(E + v B)

(7)

and Lorentz force:

Now take curl of Eq. (3), and use identity (E = (E)2E:

0 = ( E +
1

B
)
t

(8)


B
t | {z }
00 E
t
2
E
= 2E + 00 2
t
= 2E +

(9)
(10)
(11)

i.e. a wave equation for each component of the electric field E for waves

travelling with speed c 00. Why is it a wave eqn.? Consider soln.

E = E0f (x ct), with f any function. Check! Compare field profile
f (x) at time t = 0 with profile time t later: simply translated to right
distance ct = pulse f (x) moving to right with velocity c (vel. of light!)

Check that
1. E i (direction of propagation)
2. B field i, E
2.2

Normal modes of cavity

Example: transverse waves on string.

displacement of string (x, t), obeys
2
2
2
= cs 2
t2
x

wave equation
2

(12)

where cs velocity of sound. Boundary conditions: suppose

(x = 0, t) = 0,
(x = L, t) = 0

(13)
(14)

Normal modes of vibration: solutions of form

= f (x)cos(t + )

(15)

consistent with boundary conditions. Note these are standing waves.

Substitute in (12)solution only if
2

f (x) =

2
2d f
cs 2 .
dx

(16)

(16) called Eigenvalue problem (German: proper value) since eqn has
solutions consistent with B.C. only for special values (eigenvalues) .
General soln. to (16):
f (x) = A sin

x
x
+ B cos
cs
cs

(17)

B.C. (x = 0) = 0 = B = 0, (x = L) = 0 = A = 0 or
= ncs/L,

n = 1, 2, 3...

(18)

N.B. This purely classical result has connection with quantization as used
in SHO problem.
3

Modes in cavity: wave eqn. for electric field. Assume walls perfect
conductors, so B.C. are Ek = 0, B = 0 at walls. Search for solns of
Maxwell eqns. of form
E = cos(t + )f (r)

(19)

consistent with these B.C.

2.3

Periodic boundary conditions

We can find solutions with fixed (standing wave) boundary conditions, but
since it is sometimes convenient to use other B.C., lets look at periodic
B.C., i.e. assume cavity is periodically repeated with period L along x, y,
z x, y, z is same pt. as x + L, y, z, etc. This is ok since the shape of
cavity, exact pos. of walls cant matter if we are calculating something
which depends only on the density of modes in freq. range. such that
L sizes of interest.
Consider solutions
E = Re E0ei(krt)
4

(20)

= |E0| cos(k r t)

(21)

E0 = const. = const
Re=real part of
Remarks:
Plug into wave eqn. = 2 = k 2c2
E = 0 = k E0 = 0 transverse wave

Periodic B.C. =
kxL = 2nx, nx = 0, 1, 2...
ky L = 2ny , ny = 0, 1, 2...
kz L = 2nz , nz = 0, 1, 2...

(22)

Note allowed ks separated by 2 larger interval compared to fixed

b.c., but nx,y,z can be = same # per d.
note that for, e.g., nx = 1, + and - give independent solutions
check! This was not true for fixed B.C.
Allowed frequencies are therefore
5

 = kc =

2c 2
(nx + n2y + n2z )1/2.
L

(23)

Maxwells eqns linearlin. comb. of solns. is soln. General:

E = Re

p=1,2 nx ,ny ,nz

E0(n, p)ei(kr)t

(24)

here
p is polarization of wave: 2 lin. ind. directions of E for given k
n = nx , n y , n z
k = 2n/L
Exercise: derive form for B from Eq. (24) and James Clerks equations...
2.4

Counting modes

Need # of solutions ei(krt) with freqs in range to + . Look at

k-space:

 Allowed k form cubic lattice in k-space, spacing 2/L

volume occupied by each allowed k is (2/L)3
volume in k-space between k and k + k is 4k 2k
therefore number of ks in this shell is
N = 4k 2k (L/2)3

(25)

Now relate number of ks in (k, k + k) to number of allowed solutions

with s in (, + )
(recall = 2 = kc !) :
N = |{z}
2
2 polarizations

2.5

4
|{z}

L3
8 3

, integration
= 8L3 2/c

2 3 2

c
{z
}
|
k = 2/c

(26)

Planck Law

Planck idea (1900): each mode of radiation in cavity acts like SHO
suppose it contributes mean energy h/(exph/kT 1). Mean energy in
(, + ) is then
8L3h 3
E = 3 h/kT
c (e
1)

(27)

Energy per unit volume and bandwidth:

1 E
8h 3
u 3
= 3 h/kT
L
c (e
1)

(28)

( Aside: at high temperatures, h/kT 1, expand exponential to find

u = 8kT 2/c3, Rayleigh-Jeans law. Note theres no hclassical result.)
Net energy density now
u=

8h Z
3d
u d = 3 0
c
(eh/kT 1)

(29)

Make integral dimensionless, put = h/kT , d = kT d/h:

8h kT 4 Z x3dx
u= 3
c
h | 0 e{zx 1}

(30)

4/15
So Planck also recovered :
u = aT 4 Stephan-Boltzmann law
and determined
8 5 k 4
a =
15 (hc)3

(31)
(32)

But if had let h 0 as originally intended, nonsense! Although

made no sense classically, P. brave enough to allow h to stay. Stephans
const. pretty well known exptlly in 1900 = Planck determined h =
6.55 1027 erg sec, close to modern h = 6.625 1027 erg sec!