A Revisit on the Derivation of

Maxwells Equations
Shan-Yung Yang, Chao-Lun Mai, and Jean-Fu Kiang
Department of Electrical Engineering and
Graduate Institute of Communication Engineering
National Taiwan University
Taipei, Taiwan, R.O.C.
Tel: 886-2-2363-5251 ext.208
Fax: 886-2-2368-3824
E-mail: [email protected]
Sept. 1, 2005
1 Introduction
The Maxwells equations can be summarized in dierential form as

E =

B Faradays law

H =

t

D +

J Amperes law

D = electric Gauss law

B = 0 magnetic Gauss law
In the early 1820s, Ampere attempted to give a combined theory of electricity and mag-
netism after hearing about the experimental results by the Danish physicist Hans Christian
Orsted. Ampere conducted various magnetic/electrical experiments, discovered electrody-
namical forces between linear wires, and formulated all the results. However, his formula
was incomplete.
In 1831, British scientist Michael Faraday discovered that when the magnetic eld of
an electromagnet was generated or extinguished by closing or opening the electric circuit
switch, an electric current can be detected in a separate conductor nearby.
In 1864, Scottish physicist James Clerk Maxwell proposed twenty scalar equations to
describe how electric charges induce electric elds (Gausss law, 1835), the experimental
absence of magnetic charges, how currents generates magnetic elds (Amperes law, 1823),
and how changing magnetic elds produce electric elds (Faradays law of induction, 1831).
Note that a correction was made to the original Amperes law. Maxwell added the time-
varying term, based primarily on theoretical reasoning, so that changing electric elds act
like electric currents to induce magnetic elds. Maxwells additional term resolved a serious
internal inconsistency in electromagnetic theory. In this aticle, his derivations are revisited
in vector format.
2 Coulombs Law and Electric Gauss Law
As shown in Fig.1, the coulomb force between the two charged particles at r
1
and r
2
is

F =
q
1
q
2
4
0
r
1
r
2
| r
1
r
2
|
3
1
where q
1
and q
2
are the charges carried by particles 1 and 2, respectively, and
0
is the
permittivity of free space.
Figure 1: Coulombs force between two charges.
Since a particle with charge q immersed in an electric eld

E( r) will experience a
coulombs force

F = q
1

E( r). Thus, the force imposed upon q can be viewed as generated by
an electric eld due to charge q
2

E ( r) =
q
2
4
0
r r
2
| r r
2
|
3
By superposition, the electric eld intensity due to N charged particles will be

E ( r) =
N

i=1
q
i
4
0
r r
i
| r r
i
|
3
where particle i is located at r
i
, carrying charge q
i
.
When the number of charge particles is large enough, the charge distribution can be
approximated as a continuum, and a charge density can be dened such that the amount
of charge in d r

is ( r

) d r

. Thus, the total electric eld can be expressed as

E ( r) =
1
4
0
___
V
( r

)
r r

| r r

|
3
d r

Note that the electric eld becomes undened at r = r

.
The divergence of any vector

A( r) is dened as

A( r) lim
V0
_
_
1
V

__

A( r) nda
_
_
(1)
where V

is an innitesimal volume around r and S

is the surface of V

.
2
Figure 2: Electric ux passing through a closed surface.
Fig.2 shows the electric ux originating from charge ( r

) d r

passes through an arbitrary

closed surface where the ux can be decomposed into many beams emanating from ( r

) d r

.
The net ux owing out of the surface can be expressed as a surface integral of the electric
eld as
__

E nda
The electric eld intensity is inversely proportional to the square of distance, and the area
normal to the electric eld is proportional to the square of distance. Thus,

E
1
n
1
da
1
=

E
2
n
2
da
2
, the net ux out of surface S sums to zero. It implies that the
charges outside of a closed surface will contribute no net ux out of that surface. Substitute

A =

E in (1), the surface integral in the numerator is zero if S

encloses no charge. Hence,

E ( r) = 0
Consider the electric eld at an arbitrary point on an innitesimal surface S

which
encloses the space V

as shown in Fig.3

E ( r) =
1
4
0
___
V

( r

)
r r

| r r

|
3
d r

where V

is the space V

excluding S

.
The total electric ux over surface S

is
__

E ( r) nda =
1
4
0
__

S
___
V

( r

)
r r

| r r

|
3
d r

nda
3
Figure 3: Small volume V

with surface S

centered at r
o
.
Note that V

is mall enough so that ( r

) can be considered as a constant and ( r r

) nda =
| r r

|
3
d, where is the solid angle of S

centered at r
o
. Thus,
__

E ( r) nda =
( r
o
)
4
0
__

S
___
V

d r

d =
( r
o
)

0
V

(2)
Taking the limit of V

, (2) is reduced to

E ( r
o
) =
( r
o
)

0
which is the electric Gauss law.
3 Charge Conservation Law
The current is dened as the rate of charge ow. In metallic conductors such as wires,
current is a ow of electrons in nature. Mathematically, current can be dened as the net
ux through a specic cross section, namely,
I =
_
S

J nda
where I is the current (in amp),

J is the current density (in amp/m
2
) and S is the cross
section through which the current ows and n is the unit normal vector of surface da.
The current density can be expressed as

J = Nq v
where N is the particle density, q is the charge per particle, and v is the average velocity of
the particles.
4
Figure 4: Current owing out of space V .
The total charge of an isolated system remains constant. Consider a closed space V
enclosed by S in free space as shown in Fig.4. The total current left the volume is
I =
__
S

J ( r) nda
The total charge q inside the volume V is
q =
___
V
( r) d r
By charge conservation, the net current I
net
out of volume V is zero, namely,
I
net
=
d
dt
___
V
( r) d r +
__

J ( r) nda = 0 (3)
Apply the divergence theorem to the second term of (3) gives
d
dt
___
V
( r) d r +
___
V

J ( r) d r = 0
This equation is valid for any arbitrary V , implying

t
( r) +

J ( r) = 0
4 Biot-Savarts Law
Magnetic eld is a vector eld generated by moving electric charges or currents. The direction
of the eld is the same as the equilibrium direction of a compass needle placed in the eld.
5
The magnitude of the eld is inversely proportional to the square of the distance between
source and observation.
By convention,

B is called the magnetic ux density (in tesla, T) or the magnetic in-
duction, and

H =

B/ is called the magnetic eld intensity (in amp/m), where is the
permeability of the medium. In CGS units,

B and

H are in gauss (G) and oersted (Oe),
respectively.
As shown in Fig.5, the magnetic eld at r induced by an innitesimal current segment
I

at r

is
d

B ( r) =

0
I

4
d

r r

| r r

|
3
which is the Biot-Savarts law. Note that the magnetic eld is undened at r = r

.
Figure 5: Magnetic eld induced by a moving charge.
The magnetic force exerted upon a charge q moving at velocity v is

F = q v

B
which is called the Lorentz force law.
Figure 6: Magnetic force between two moving charges.
As shown in Fig.6, the force exerted at r by the moving charge (current) at r

is
d

F ( r) = dq v d

B ( r) = Id

d

B ( r)
= Id

0
4
I

r r

| r r

|
3
_
6
Note that Id

=

J ( r) d r and I

=

J ( r

) d r

. Thus, the force

F ( r) is
d

F ( r) =

J ( r) d r
_

0
4

J ( r

) d r

r r

| r r

|
3
_
5 Magnetic Gauss Law
The net magnetic ux owing out of a closed surface S can be expressed as
__

B nda
As shown in Fig.7, the direction of magnetic ux density associated with an innitesimal
current element

J ( r

) d r

is on the plane normal to

J ( r

) d r

. The points on that plane

having the same distance from r

experience the same magnitude of magnetic ux density.

Thus,

B n
1
da
1
=

B n
2
da
2
, the net ux out of surface S is zero. This conclusion also
applies to any innitesimal surface S

. By denition of divergence in (1), we have

B ( r) = 0
In summary, the currents outside of a closed surface will contribute no net ux to that
surface.
Figure 7: Magnetic ux passing through a closed surface.
By the Biot-Savarts law, the magnetic ux density on S

can be expressed as

B( r) =

0
4
___
V

J ( r

)
r r

| r r

|
3
d r

(4)
7
where V

is the space enclosed by S

and V

is the space V

excluding S

as shown in Fig.3.
Assume V

is a sphere cenered at r
o
and V

is small enough so that

J ( r

) is uniform inside
the sphere. The total magnetic ux over surface S

can be calculated as
__

B ( r) nda =

0
4
__

S
___
V

J ( r

)
r r

| r r

|
3
d r

nda
=

0
4
___
V

__

J ( r
o
)
r r

| r r

|
3

r r
o
| r r
o
|
da d r

=

0
4
___
V

__

J ( r
o
) ( r r

) ( r r
o
)
| r r

|
3
| r r
o
|
da d r

(5)
Figure 8: Surface integral of magnetic ux density over S

where the current is perpendicular

to the z direction.
Choose the coordinate system such that r
o
= 0 and the current is perpendicular to
the z direction as shown in Fig.8, namely,

J ( r
o
) = xJ
x
+ yJ
y
. The integrand at point
r
1
= xx + yy + zz contributed by

J ( r

1
) at r

1
= xx

+ yy

+ zz

is

J ( r
o
) ( r
1
r

1
) ( r
1
r
o
)
| r
1
r

1
|
3
| r
1
r
o
|
=

x y z
J
x
J
y
0
x x

y y

z z

| r
1
r

1
|
3
| r
1
|
8
The integrand at point r
2
= xx + yy zz contributed by

J ( r

2
) at r

2
= xx

+ yy

zz

is

J ( r
o
) ( r
2
r

2
) ( r
2
r
o
)
| r
2
r

2
|
3
| r
2
r
o
|
=

x y z
J
x
J
y
0
x x

y y

z + z

| r
1
r

1
|
3
| r
1
|
Thus, we have

J ( r
o
) ( r
1
r

1
) ( r
1
r
o
)
| r
1
r

1
|
3
| r
1
r
o
|
+

J ( r
o
) ( r
2
r

2
) ( r
2
r
o
)
| r
2
r

2
|
3
| r
2
r
o
|
= 0
By symmetry, any contribution from r

1
to r
1
will always be canceled by another contribution
from r

2
to r
2
so that the integral over S

in (5) can be reduced to

__

B ( r) nda = 0 (6)
Applying the denition of divergence in (1) with

A =

B, (6) implies

B ( r) = 0
which is the magnetic Gauss law in dierential form.
6 Faradays Law
Moving a permanent magnet into and out of a wire coil will induce a current on that wire.
Also, moving a wire coil near a stationary permanent magnet will induce a current on the
wire. These observations can be summarized as
_
C

E ( r) d =

t
__
S

B ( r) nda (7)
for an arbitrary surface S encircled by C.
The Stokes theorem states that
_
C

E ( r) d =
__
S

E ( r) nda
for any well-behaved function

E ( r). Thus, (7) implizes that

E ( r) =

B( r)
which is the Faradays Law of induction in dierential form.
9
7 Amperes Law
For an arbitrary function f ( r), we have
___
V
d r

f ( r

) ( r r

) = f ( r)
where ( r r

) is the Dirac-delta function. By Greens theorem,

2
1
| r r

|
= 4 ( r r

)
Since

1
| r r

|
=
r r

| r r

|
3
(4) can be rephrased as

B ( r) =

0
4
___
V

1
| r r

|


J ( r

) d r

(8)
By using

A =
_

A
_

A, the integrand in (8) can be reduced to

J ( r

)
| r r

|

1
| r r

J ( r

)
Since the del operator operates on r but

J ( r

) is a function of r

. Thus,

J ( r

) = 0, and
(8) is reduced to

B ( r) =

0
4

___
V

J ( r

)
| r r

|
d r

Taking the curl of

B ( r), we have

B ( r) =

0
4

___
V

J ( r

)
| r r

|
d r

(9)
Using the identity
_

A
_
=
_

A
_

2

A, the right-hand side of (9) becomes

0
4
_
_

___
V
d r

J ( r

)
| r r

|

2
___
V
d r

J ( r

)
| r r

|
_
_
(10)
10
The rst term of (10) can be simplied as

___
V
d r

J ( r

)
| r r

|
=
___
V
d r

J ( r

)
1
| r r

|
=
___
V
d r

J ( r

1
| r r

|
(11)
where
1
| r r

|
=

1
| r r

|
.
By using
_

A
_
=

A +

A, (11) can be further reduced to

___
V
d r

J ( r

)
| r r

|

1
| r r

J ( r

)
_
=
___
V
d r

J ( r

)
| r r

|
+
___
V
d r

J ( r

)
| r r

|
(12)
Applying the charge conservation law,

J ( r

) =

t
( r

), to the rst term with

steady-state condition, and applying the divergence theorem to the second term, (12) is
reduced to

__

J ( r

)
| r r

|
nd a (13)
Choose a surface S that encloses the whole current distribution to have

J ( r) = 0 on S,
then (13) is equal to zero, and (9) can be reduced to

B ( r) =

0
4

2
___
V
d r

J ( r

)
| r r

|
=

0
4
___
V
d r

J ( r

)
| r r

|
(14)
Applying the Greens theorem to (14), we have

B ( r) =
0

J ( r)
which is the original form of Amperes law.
11
8 Modied Amperes Law
The original Amperes law

H ( r) =

J ( r) leads to

J ( r) =

H ( r) = 0, which
is valid only at the steady state.
Substituting the electric Gauss law

D( r) = ( r) into the charge conservation law

t
( r) +

J ( r) = 0, we have

t
_

D( r)

+

J ( r) = 0
or

_

t

D( r) +

J ( r)
_
= 0
which suggests that

t

D( r) +

J ( r) should be viewed as an eective current in the original
Amperes law. Hence,

t

D( r) is identied as the displacement current. The Amperes law

is thus modidied as

H ( r) =

t

D( r) +

J ( r)
References
[1] James Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed., Oxford: Claren-
don Press, 1904.
12