The Equations of Macroscopic

Electromagnetism

Physics 212 2010, Electricity and Magnetism

Michael Dine
Department of Physics
University of California, Santa Cruz

November 2010

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
We have, so far, guessed the equations of macroscopic
electrodynamics:

~
~ ·B
∇ ~ =0 ∇ ~ ×E ~ + 1 ∂ B = 0. (1)
c ∂t
~
~ ·D
∇ ~ = 4πρ ∇ ~ ×H ~ − 1 ∂ D = 4π ~J. (2)
c ∂t c
We had the relations (“constitutive relations")
~ =E
D ~ + 4π P
~ ~ =B
H ~ − 4π M.
~ (3)

Goal here is to understand the microscopic origins of these

equations. We will use a classical language, but this can readily
be translated to quantum mechanics.

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Microscopic equations:

~
∇ ~ × ~e + 1 ∂ b = 0.
~ · ~b = 0 ∇ (4)
c ∂t
~
∇ ∇~ × ~b − 1 ∂ e = 4π~j.
~ · ~e = 4πη (5)
c ∂t c
where for this lecture, the lower case denotes the microscopic
quantities.

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Some relevant length scales

:
1 Size of nuclei 10−13 cm
2 Size of atoms 10−8 cm
3 Wavelength of visible light 10−6 cm
So for visible light, for example, wave is roughly constant in
space over a volume containing a million atoms. For, e.g.,
x-rays, wavelengths are smaller than typical atomic dimensions,
so macroscopic description is inappropriate; x-rays resolve
individual atoms.

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Averaging

So it makes sense to average over distances large compared to

atomic sizes. It is not appropriate to average in time, since, e.g.
for light, frequencies are comparable to frequencies of atomic
motion.
Introduce a test function, f (~x ), and define, for some quantity
F (~x , t) (e.g. one of the fields):
Z
F(~x , t) ≡ hF (~x , t)i = d 3 x 0 f (~x 0 )F (~x − ~x 0 , t). (6)

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Choice of the test function

Would like isotropic, smooth (so, e.g. in momentum space, it

doesn’t have discontinuities or other irregularities):
Normalize to unity.
2 /R 2
f (~x ) = (πR 2 )−3/2 e−r . (7)
It is clear what this does in coordinate space. What sort of
averaging does it do in momentum space?
Good practice with Fourier transforms.

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Fourier transform of smoothed quantities

We’d like to see that high wave number (short wavelengths) are
smoothed out.
Z Z
~ −i ~k ·~x
3
F(k , t) = d x e d 3 x 0 f (~x 0 )F (~x − ~x 0 , t). (8)
Z Z
~
= d 3 xe−i k ·~x d 3 x 0 f (~x 0 ) (9)
Z Z
~ 0 ·~x 0 ~ 00 ·(~x −~x 0 )
× d 3k 0 d 3 k 00 f (~k 0 )ei k F (~k 00 , t)ei k .

Now the ~x and ~x 0 integrations give δ functions:

(2π)6 δ(~k − ~k 00 )δ(~k 0 − ~k 00 ). (10)

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
So the result is
f (~k )F(~k ). (11)
It is a useful exercise to calculate the Fourier transform of f .
More generally, we will several times be interested in the
integral: Z
~ 2 2
I(a, k ) = d 3 x ei k ·~x e−a x .
~ (12)

This is a standard integral which is done by completing the

squares in the exponent:
2
Z ~
~ −a2 (~x −i k2 )2 − 41 k 2
I(a, k ) = d 3 x e 2a a (13)

k2
− 41
= π 3/2 a−3 e a2 .

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
So
3 1 2 2
f (~k ) = √ e− 4 k R . (14)
4 π
This means that for wave lengths large compared to R,

F(~k , t) = F (~k , t) (15)

while for λ  R,
F(~k , t) → 0. (16)

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Note some useful features of this averaging:
Most important, differentiation and averaging commute:
Z
∂ ∂
F(~x , t) = d 3 x 0 f (~x 0 )F (~x − ~x 0 , t) (17)
∂xi ∂xi
Z
∂F
= d 3 x 0 f (~x 0 ) (~x − ~x 0 , t)
∂xi
∂F
=h i.
∂xi
∂
Clearly also ∂t commutes with averaging.

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
So consider, first, the homogeneous equations. Because
differentiation and averaging commute, calling
~ = h~ei B
E ~ = h~bi (18)

we have:
~
∇ ~ =0
~ ·B ~ = − 1 ∂B .
~ ×E
∇ (19)
c ∂t

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
For the inhomogeneous equations, we have to be more careful.
Consider, first,
∇ ~ = 4πhη(~x , t)i.
~ ·E (20)
We need to convert the right hand side into something
recognizable.
First, we divide the charge density into a “free" piece and a
bound piece, corresponding to electrons bound in molecules:

η = ηfree + ηbd (21)

where
X X
ηfree (~x , t) = qj δ(~x − ~xj (t)) ηbd (~x , t) = ηn (~x , t). (22)
free n

Here ηn is the microscopic charge density of the n’th molecule:

X
ηn (~x , t) = qj δ(~x − ~xj (t)). (23)
j∈n

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
For the bound part, we first write:

~xj = ~xn + ~xjn . (24)

Then Z
hηn (~x , t)i = d 3 x 0 f (~x 0 )ηn (~x − ~x 0 , t) (25)
X Z
= qj d 3 x 0 f (~x 0 )δ(~x − ~x 0 − ~xjn − ~xn )
X
= qj f (~x − ~xjn − ~xn ).
j∈n

Now we see the virtue of a smooth choice of f . For

R  angstrom, we can Taylor expand f :

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
 Xh
hηn (~x , t)i = ~ (~x − ~xn )(26)
f (~x − ~xn ) − ~xjn · ∇f
j∈n

1X ∂2
+ (xjn )α (xjn )β f (~x − ~xn ) + · · · .
2 ∂xα ∂xβ
αβ

Now X X
qj = qn ; qj ~xjn = ~pn (27)
jn

and the last sum in the brackets above is related to the quadrupole
moment. Dropping this term as small in most circumstances, we have

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
 ~ (~x − ~xn ) + . . . ...
hηn (~x , t)i = qn f (~x − ~xn ) − ~pn · ∇f (28)
This is what we would have obtained from
~ · h~pn δ(~x − ~xn )i + . . . .
hηn (~x , t)i = hqn δ(~x − ~xn )i − ∇ (29)

(Check!)

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
So summing over the molecules

hη(~x , t)i = ρ(~x , t) − ∇ ~ ~x , t) + . . .

~ · P( (30)

where
X X
ρ(~x , t) = h qj δ(~x − ~xj ) + qn δ(~x − ~xn )i (31)
j(free) n

X
~ ~x , t) = h
P( ~pn δ(~x − ~xn )i. (32)

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
So we have
∇ ~ ~x , t) = ρfree (~x , t)
~ · D( (33)
where
~ ~x , t) = E
D( ~ + 4π P(
~ ~x , t). (34)
It is straightforward to carry out this expansion to higher orders
in a/R. This is described in G. Rusasakoff, American Journal of
Physics, 38 (1970) 1188 (a rather pretty article). In this case,
~ e.g.
there are further corrections to D,
0
X ∂Qαβ
Dα = Eα + 4πPα − 4π . (35)
∂β
β

Note that there is no assumption here that the system is static!

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Now for the last of the Maxwell equations. Here, we need
~ ~ as defined
somehow to get ∂∂tD on the right hand side, with D
above (including higher order terms! We’ll content ourselves
with the leading term).
Start, as before, breaking up
~j = ~jfree + ~jbd . (36)
X
~jbd = ~jn (~x , t). (37)
n

Again X
~jn (~x , t) = qj ~vj δ(~x − ~xj ). (38)
j∈n

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Again, take
~xj = ~xjn + ~xn ~vj = ~vjn + ~vj . (39)
So X
hjn (~x , t)i = qj (~vjn + ~vn )f (~x − ~xn − ~xjn ). (40)
j∈n

Again, Taylor expand f about ~xn ; we’ll stop with the second
term:
X X
hjn (~x , t)i = qj (~vjn + ~vn )f (~x − ~xn )− ~ (~x − ~xn )
qj (~vjn + ~vn )~xjn · ∇f
(41)

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
We want to write these quantities in a more familiar form.
X d X
qj ~vjn = qj ~xjn (42)
dt
d
= ~pn .
dt

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
Now consider X
− qj (vjn )α (xjn )β (43)
1X
=− qj ((vjn )α (xjn )β − (vjn )β (xjn )α − symmetric term
2
The first term can be rewritten in terms of the magnetic moment
of the n’th molecule, in a way which is now familiar:

1 X
~n =
m qj (~xjn × ~vjn ). (44)
2c
j

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
So we have
d
h~jα i = h~jnα δ(~x −~xn )i+h ~pnα δ(~x −~xn )i+cαβγ ∂β hm ~ nγ δ(~x −~xn )i+. . . .
dt
(45)
Defining the macroscopic current density:
X X
~J(~x , t) = h qj ~vj δ(~x − ~xj )i + h qn ~vn δ(~x − ~xn )i (46)
j n

and X
~ ~x , t) = h
M( ~ n δ(~x − ~xn )i
m (47)
n

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism
 (∇ ~ α = 4π ~Jα + 4π(∇
~ × B) ~ α + 1 ∂Dα .
~ × M) (48)
c c ∂t
~ =B
So calling H ~ − 4π M,
~

~
∇ ~ = 4π ~J + 1 ∂ D .
~ ×H (49)
c c ∂t

Physics 212 2010, Electricity and Magnetism The Equations of Macroscopic Electromagnetism