Maxwell–Ampere Law

The original Ampere Law

∇ × H = Jfree , (1)
I
H · d~ℓ = Ifree
net
[through loop L] (2)
L

applies to the magnetic fields of steady currents, but it does not work for the time-dependent
currents J(t) with ∇·J 6= 0. Indeed, consider an AC current I(t) flowing through a capacitor

S1
S2

(3)

loop L

Of course, the current does not really flow through the capacitor but only through the wires
connected to each of the capacitor’s plates. Inside the capacitor, there is no electric current;
instead, the charges temporarily accumulate on the plates according to

dQ
= I(t). (4)
dt

Now, let’s try to apply the Ampere Law (2) to the loop L on the diagram (3). The current
through L is defined as the net current through any surface spanning the loop, but for a non-
steady current different surfaces may give different answers. For example, the net current
through the surface S1 — which crosses one of the wires connected to the capacitor — is
the AC current I(t) in the wire, while the net current through the surface S2 — which goes

1
between the two plates — is zero. So what should we use on the RHS of eq, (2)? The I(t)?
Zero? Something else?

In 1861, James Clerk Maxwell resolved this issue by adding the displacement current

∂
Jd (x, y, z; t) = D(x, y, z; t) (5)
∂t

to the free current Jf in the Ampere Law:

∇ × H = Jf + Jd . (6)

The (x, y, z; t) in eq. (5) is the electric displacement field

D = ǫ0 E + P, (7)

so the displacement current (5) involves the time derivatives of both the electric field E and
of the dielectric polarization P,

∂E ∂P
Jd = ǫ0 + , (8)
∂t ∂t

except in a vacuum where only the first term is present.

The divergences of the free current Jf and the displacement current Jd always cancel
each other,

∇ · Jf (x, y, z; t) + ∇ · Jd (x, y, z; t) = 0, (9)

and that’s what makes the Maxwell–Ampere equation (6) mathematically consistent. Like-
wise, in the global form
I ZZ
H · d~ℓ = Jf + Jd ) · d 2 a (10)
L S

for any surface S spanning the Ampere loop L.

2
 The divergence cancellation (9) stems from the continuity equation for the free current
and charges,
∂
∇ · Jf (r, t) = − ρf (r, t), (11)
∂t

and the Gauss Law for the electric displacement field,

∇ · D(r, t) = ρf (r, t). (12)

which works without any modifications for the time-dependent fields and charges. Taking
the time derivative of both sides of the Gauss Law equation (12), we obtain

∂ ∂
∂D(r, t)
ρf (r, t) = ∇ · D(r, t) = ∇ · = ∇ · Jd (r, t), (13)
∂t ∂t ∂t

hence in light of the continuity equation (11),

∇ · Jf (r, t) = −∇ · Jd (r, t). (14)

Therefore, regardless of how the currents and the charges change with time and place, the
combined free + displacement current Jf + Jd always have zero divergence, which makes the
Maxwell–Ampere Law (6) mathematically consistent. And indeed, the Maxwell–Ampere Law
(6) happens to work for any time-dependent fields and currents.

For an example of the displacement current, consider a parallel-plate capacitor (such as

shown on diagram (3)), perhaps with a uniform dielectric between the plates. The displace-
ment field between the plates is uniform (except near the edges of the plates), and its value
follows from the plate charges ±Q via the Gauss Law,

Q
D = ẑ . (15)
plate area A

Consequently, when the capacitor’s charge Q(t) is time-dependent due to free current If (t)
in the wires connected to the capacitor, the displacement field (15) inside the capacitor

3
becomes time-dependent, which gives rise to the displacement current

dQ ẑ
Jd = . (16)
dt A

This displacement current density flows only between the plates, so the net displacement
current through the capacitor is

dQ
Id = Jd · (A ẑ) = , (17)
dt

which is exactly equal to the current If (t) in the wires connected to the capacitor,

dQ
If (t) = = Id (t). (18)
dt

Physically, this means that the combined free+displacement current If (t)+Id (t) flows without
interruption through the wires and through the capacitor. It is this un-interrupted current
I(t) which gives rise to the magnetic field surrounding the capacitor and the wires.

Maxwell Equations
And God said,

∇ · D = ρfree , (19.a)
∂B
∇×E = − , (19.b)
∂t
∇ · B = 0, (19.c)
∂D
∇ × H = Jfree + , (19.d)
∂t

and there was light.

The four equations (19.a–d) governing the macroscopic electric and magnetic fields are
usually called the Maxwell equations. Historically, 3 12 of these equations were known before
James Clerk Maxwell, but he was the first physicist to put them together and discover
that they lead to the electromagnetic waves. Maxwell has also identified the light as an
electromagnetic wave, hence the above biblical quote.

4
 Actually, Maxwell’s 1865 article A Dynamical Theory of the Electromagnetic Field had
about 20 equations, and his 1983 two-volume book A Treatise on Electricity and Magnetism
had a few more equations. Part of this proliferation was due to not using the vector notations
— it was invented a few years after Maxwell’s death, so he had to write the vector equations
in components. Also, the paper and the book included some extra equations such as the
Ohm’s law J = σE, the linear dielectric equation D = ǫǫ0 E, etc., etc. In any case, Maxwell’s
Treatise contained some tremendously important ideas, but it was poorly organised and
very hard to read. It took Oliver Heaviside, Josiah Willard Gibbs, and Heinrich Hertz
to eventually (1884?) make Maxwell’s work accessible to an average physicist. In fact, it
were Hertz and Heaviside who had grouped equations (19.a–d) together, wrote them in the
modern form, and called them the Maxwell’s equations.

In the integral form, the Maxwell equations (19.a–d) become

ZZ
D · d2 a = Qfree [inside S], (20)
S
ZZ
B · d2 a = 0, (21)
S
d
I ZZ
E · d~ℓ = − B · d2 a, (22)
dt
L SL
d
I ZZ ZZ
~
H · dℓ = 2
Jfree · d a + D · d2 a, (23)
dt
L SL SL

where S is any closed surface, L is any closed loop, and SL is any surface spanning the loop
L. These integral equations give rise to the boundary conditions for the fields at a boundary
between two dielectric and/or magnetic media. In particular, using a brick-shaped Gaussian
surface S in eqs. (20) and (21) — half the brick in one medium, half the brick in the other
— we obtain

(1) (2)
D⊥ − D⊥ = σfree , (24)
(1) (2)
B⊥ − B⊥ = 0. (25)

We have seen these equations before for the static D and B fields, but now we know they
apply without changes to the time-dependent fields.

5
 The other boundary conditions follow from eqs. (22) and (23) for a narrow rectangular
Ampere loop ∆x × ∆z, with small ∆x and much smaller ∆z ≪ ∆x. In the ∆z → 0 limit,
eq. (22) becomes

∆x · E(1) − E(2)


= 0 (26)

since the magnetic flux through the loop vanishes with loop’s area as ∆z → 0. Consequently,
for static or dynamical electric fields.

(1) (2)
Ek − Ek = 0 (27)

Likewise, in eq. (23) for the same Ampere loop, the D flux through the loop vanishes in
the ∆z → 0 limit, while the net free current through the loop is reduced to the surface free
current only, hence
(1) (2)
Hk − Hk = Kfree × n (28)

where n is the unit vector ⊥ to the boundary.

The boundary conditions (24), (25), (27), and (28) are the basis of the theory of reflection
and refraction of the electromagnetic waves at the boundaries of transparent materials. But
that subject will have to wait until the second semester of the ElectroDynamics class.

As written, the Maxwell equations (19.a–d) govern the macroscopic electric and magnetic
fields E, D, B, H in dielectric/magnetic matter. Microscopically, there are only E and B
fields, and the Maxwell equations governing them are

1
∇·E = ρ, (29.a)
ǫ0
∂B
∇×E = − , (29.b)
∂t
∇ · B = 0, (29.c)
∂E
∇ × B = µ0 J + µ0 ǫ0 . (29.d)
∂t

6
 Electromagnetic Waves
The most revolutionary aspect of the Maxwell equations (19.a–d) or (29.a–d) is that they
allow for the electromagnetic waves propagating in the absence of any charges or currents. Or
rather, it might take some time-dependent charges or currents to create an electromagnetic
wave in the first place, but one created it keeps propagating further and further away without
any additional charges or currents.

To see how this works, consider the Maxwell equations in a vacuum, in a total absence
of any charges and currents:
∇ · E = 0,
∇ · B = 0,
∂B (30)
∇×E = − ,
∂t
∂E
∇ × B = µ0 ǫ0 .
∂t

The last two equations here are coupled first-order differential equations for the electric and
magnetic fields. We can de-couple these equations by applying an extra curl:

   
∂B ∂ ∂ ∂E
∇ × (∇ × E) = ∇× − = − ∇ × B) = − µ0 ǫ0
∂t ∂t ∂t ∂t
∂ 2
= −µ0 ǫ0 2 E, (31)
∂t
   
∂E ∂ ∂ ∂B
∇ × (∇ × B) = ∇ × µ0 ǫ0 = µ0 ǫ0 ∇ × E) = µ0 ǫ0 −
∂t ∂t ∂t ∂t
∂2
= −µ0 ǫ0 2 B. (32)
∂t

At the same time, a double curl of a vector field is related to its Laplacian according to

∇ × (∇ × A) = ∇(∇ · A) − ∇2 A. (33)

In particular, for the electric and magnetic fields which have zero divergences (in the vacuum),
we have

∇ × (∇ × E) = −∇2 E, ∇ × (∇ × B) = −∇2 B. (34)

7
Consequently, the decoupled second-order equations (31) and (32) become

∂2
∇2 E(x, y, z, t) = +µ0 ǫ0 E(x, y, z, t),
∂t2 (35)
∂2
∇2 B(x, y, z, t) = +µ0 ǫ0 2 B(x, y, z, t).
∂t

Physically, this means that each component Ex , Ey , Ez , Bx , By , Bz of the electric or magnetic

field in the vacuum obeys the wave equation

1 ∂2 ∂2 ∂2 ∂2
 
− − − Field(x, y, z, t) = 0, (36)
v 2 ∂t2 ∂x2 ∂y 2 ∂z 2

with wave speed

1 1
= µ0 ǫ0 =⇒ v = √ = c = 299 792 458 m/s. (37)
v2 ǫ0 µ0

Back in Maxwell’s time, the measurements of speed of light in the vacuum were a few
percent off, about 3.12 · 108 m/s. The Coulomb constant 1/(4πǫ0 ) was also a few percent off,
so plugging it into eq. (37), Maxwell got v ≈ 3.12 · 108 m/s. The sheer coincidence between
the experimental speed of light in the vacuum and the theoretical speed of the EM waves
(also in the vacuum) immediately suggested to Maxwell that light is an electromagnetic
wave. Later, with better measurements, both speeds were corrected by a few percent, but
the Maxwell’s conclusion stands: the light is an electromagnetic wave.

Let me conclude these notes with the speed of light in a linear dielectric/magnetic
medium. In the absence of free charges or currents, the macroscopic Maxwell equations
(19.a–d) become
1
∇·E = ∇ · D = 0,
ǫǫ0
∇ · B = 0,
∂B (38)
∇×E = − ,
∂t
∂D ∂E
∇ × B = µµ0 = µµ0 ǫǫ0 .
∂t ∂t

8
Consequently,

∇2 E = ∇(∇ · E) − ∇ × (∇ × E) = −∇ × (∇ × E)
   
∂B ∂
∂ ∂E
= −∇ × − = + ∇×B = + µµ0 ǫǫ0 (39)
∂t ∂t ∂t ∂t
∂ 2
= +µµ0 ǫǫ0 2 E,
∂t

and likewise

∇2 B = ∇(∇ · B) − ∇ × (∇ × B) = −∇ × (∇ × B)
   
∂E ∂
∂ ∂B
= −∇ × µµ0 ǫǫ0 = −µµ0 ǫǫ0 ∇ × E = −µµ0 ǫǫ0 − (40)
∂t ∂t ∂t ∂t
∂2
= +µµ0ǫǫ0 2 B.
∂t

Thus, each component of the electric or magnetic field obeys the wave equation

1 ∂2 ∂2 ∂2 ∂2
 
− − − Field(x, y, z, t) = 0, (41)
v 2 ∂t2 ∂x2 ∂y 2 ∂z 2

for wave speed

1 1 c
= µµ0 ǫǫ0 =⇒ v = √ √ = √ . (42)
v2 µǫ µ0 ǫ0 µǫ