Escolar Documentos
Profissional Documentos
Cultura Documentos
1
ELECTRODYNAMICS I / IFUSP / LECTURE 1
Maxwell’s Equations
@B
r⇥E+ =0
@t
1
r·E= ⇢ r·B=0
✏0
@E
r⇥B ✏ 0 µ0 = µ0 J
@t
@⇢
+r·J=0 F = q (F + v ⇥ B)
@t
ELECTRODYNAMICS I / IFUSP / LECTURE 1 2
Gauss’s Law & Electrostatics
• The most basic of Maxwell’s Equations of Electrostatics is Gauss’s Law:
∇ ⋅ E = − ∇ ϕ = ρ( x ⃗ ) , with E ⃗ = − ∇ ⃗ ϕ .
⃗ ⃗ 2 1
ϵ0
• We saw during last class (and you know this since kindergarten!) that the solution is:
1 ρ( x ′⃗ )
ϕ( x ⃗ ) =
∫
d 3x′
| x ⃗ − x ′⃗ |
.
4π ϵ0
E q⃗
• Furthermore, we can consider the charge density (a distribution), to be a sum of point charges:
ρ( x ⃗ ) = qi δ( x ⃗ − x i⃗ )
∑
x ⃗q
i
q
x⃗−
• For a point charge, the electric field is given by:
x ⃗ − x q⃗ x⃗
⃗ ⃗ ⃗
( 4π ϵ0 | x ⃗ − x q⃗ | ) 4π ϵ0 | x ⃗ − x q⃗ |
1 q q
E q = − ∇ ϕq( x ⃗ ) = − ∇ =
x q⃗
• Now, how can we be sure that the solution above is unique?
Helmholtz Theorem
• Put another way: given only the equation
∇⃗ ⋅ E ⃗ =
1
ρ( x ⃗ ) ,
ϵ0
can we find a unique solution?
• No! We can add to E ⃗ any field H ⃗ which solves the homogeneous equation, ∇ ⃗ ⋅ H ⃗ = 0 , i.e.,
⃗ ⃗ ⃗ 1
∇ ⋅ ( E + H ) = ρ( x ⃗ ) + 0 .
ϵ0
• However, if we specify not only the divergence of a field, but also its curl (“rotational”), then the equations yield
a unique solution.
∇ ⃗ ⋅ F ⃗ = Sdiv
⃗ ,
∇ ⃗ × F ⃗ = S rot
Helmholtz Theorem
• We will show that, given the pair of equations:
∇ ⃗ ⋅ F ⃗ = Sdiv and ⃗ ,
∇ ⃗ × F ⃗ = S rot
then the unique solution for the vector field F ⃗ is given by:
F ⃗ in s(v)
F ⃗ = − ∇ ⃗ Ψ + ∇ ⃗ × R ⃗ , where
1 Sdiv( x ′⃗ ) 1 F (⃗ x ′⃗ )
∫ ∮
3
Ψ= d x′ − d s′⃗ ⋅
| x ⃗ − x ′⃗ | 4π s(v) | x ⃗ − x ′⃗ |
, and
4π v
⃗ in v
Sdiv and S rot
⃗ ( x ′⃗ ) ⃗ x ′⃗ )
R⃗=
1 S 1 F (
∫ ∮
3 rot
d x′ − d s′⃗ ×
4π v | x ⃗ − x ′⃗ | 4π s(v) | x ⃗ − x ′⃗ |
• So, we have to specify not only the sources of the divergence and curl on the “bulk" (volume),
but also the behavior of the field itself ( F )⃗ on the boundary (the surface of the volume)!
x ⃗ − x ′⃗
∇⃗
1
=−
| x ⃗ − x ′⃗ | | x ⃗ − x ′⃗ |
, and that
( )
∇⃗ ⋅ ∇⃗
1
= − 4π δ( x ⃗ − x ′⃗ )
| x ⃗ − x ′⃗ |
• Now let’s rewrite the field itself as:
d 3 x′ F (⃗ x ′⃗ ) δ( x ⃗ − x ′⃗ ) = − d 3 x′ F (⃗ x ′⃗ ) ∇ 2x⃗
1 1
F( x ⃗ ) =
∫v 4π ∫v | x ⃗ − x ′⃗ |
1 2⃗ F (⃗ x ′⃗ )
∫v
3
=− ∇ x d x′
4π | x ⃗ − x ′⃗ |
⃗ x ′⃗ ) ⃗ x ′⃗ )
F (⃗ x ⃗ ) = − ∇ x⃗ ∇ x⃗ ⋅ d x′ ∇ x⃗ × ∇ x⃗ × d x′
( | x ⃗ − x ′⃗ | ) 4π ( | x ⃗ − x ′⃗ | )
1 F ( 1 F (
∫v ∫v
3 3
+
4π
① ②
⃗ x ′⃗ )
⃗ F ( ⃗ ⃗
= d x′ F ( x ′⃗ ) ⋅ ∇ x
1 ⃗ ⃗
= d x′ F ( x ′⃗ ) ⋅ (− ∇ x′)
1
∫v ∫v ∫v
3 3 3
① = d x′ ∇ x ⋅
| x ⃗ − x ′⃗ | | x ⃗ − x ′⃗ | | x ⃗ − x ′⃗ |
⃗ ⋅ F (⃗ x ′⃗ ) ⃗ ⋅ F (⃗ x ′⃗ )
[ | x ⃗ − x ′⃗ | ] ∫v
1
+ d 3 x′ [ ∇ x′ ] | x ⃗ − x ′⃗ |
∫v
3
=− d x′ ∇ x′
F (⃗ x ′⃗ )
[ | x ⃗ − x ′⃗ | ] ∫v
1
+ d 3 x′ [Sdiv( x ′⃗ )]
∮s(v)
=− d s′⃗ ⋅ = 4π Ψ
| x ⃗ − x ′⃗ |
!!!
F (⃗ x ′⃗ )
d x′ ∇ x⃗ ×
∫v
3
② =
| x ⃗ − x ′⃗ |
F (⃗ x ′⃗ ) ⃗ ( x ′⃗ ) = 4π R ⃗
[ | x ⃗ − x ′⃗ | ] ∫v
1
+ d 3 x′ [ S rot ] | x ⃗ − x ′⃗ |
∮s(v)
=− d s′⃗ × !!!
⃗ 1 ⃗ 1 ⃗
⃗
F( x ) = − ∇x ① + ∇x × ②
4π 4π
⇒ F (⃗ x ⃗ ) = − ∇ x⃗ Ψ + ∇ x⃗ × R ⃗
• Some familiar cases are the electrostatic field with boundary conditions such that E ⃗ → 0 as x → ∞. We then have that
⃗ → 0 , and Ψ → ϕ , with the result that:
Sdiv → ρ/ϵ0 , S rot
ρ( x ′⃗ )
E (⃗ x ⃗ ) = − ∇ x⃗ ϕ = − ∇ x⃗
[ | x ⃗ − x ′⃗ | ]
1
∫
3
d x′
4π ϵ0
⃗ → μ J ⃗ and
• For the magnetostatic field with boundary conditions such that B ⃗ → 0 as x → ∞ , we have that Sdiv → 0 , S rot 0
R ⃗ → A ⃗ , with the result that:
J (⃗ x ′⃗ )
B (⃗ x ⃗ ) = ∇ x⃗ × A ⃗ = ∇ x⃗ ×
[ 4π ∫ | x ⃗ − x ′⃗ | ]
μ0 3
d x′
Boundary conditions
• The Helmholtz theorem highlights the need to consider carefully the boundary conditions of our fields:
F ⃗ = − ∇ ⃗ Ψ + ∇ ⃗ × R ⃗ , with
dS ⃗ F⃗
Sdiv( x ′⃗ ) F (⃗ x ′⃗ )
F ⊥⃗
1 1
∫ ∮
3
Ψ= d x′ − d s′⃗ ⋅ ⟸
4π v | x ⃗ − x ′⃗ | 4π s(v) | x ⃗ − x ′⃗ |
s(v)
⃗ ( x ′⃗ ) ⃗ x ′⃗ ) v
R⃗= F ||⃗
1 S 1 F (
∫ ∮
rot
d 3x′ − d s′⃗ × ⟸
4π v ⃗ ⃗
| x − x ′| 4π s(v) | x ⃗ − x ′⃗ |
• Therefore, the component perpendicular to the surface (and parallel to d S )⃗ is needed to specify the divergence
component Ψ ( → ϕ);
• The component parallel to the surface (perpendicular to d S )⃗ is needed to specify the curl component R ⃗ ( → A ⃗ ) .
Boundary conditions
• The simplest situation is one where we want to
∞
find the electromagnetic fields in around
sources which are in complete isolation. We
3 ρ, J ⃗
then have v → ℝ , and the boundary is pushed ∞ ∞
⃗ ⃗
∇⋅E =
ρ
, ∇⃗ × E ⃗ = 0
ϵ0
• In the absence of non-trivial boundary conditions, the solution is then simply:
B
E (⃗ x ⃗ ) = − ∇ ⃗ ϕ ,
1 ρ( x ′⃗ )
ϕ( x ⃗ ) =
∫
3
d x′
4π ϵ0 | x ⃗ − x ′⃗ |
• In this case the electrostatic field is the gradient of a scalar function (ϕ), and one of the
properties of the gradient is that, along any path, the integration of the gradient gives back
the scalar function itself, i.e.:
B
d l ⃗ ⋅ ∇ ⃗ ϕ = ϕ(B) − ϕ(A)
∫A
, the potential difference between A and B.
F L⃗ = q ( E ⃗ + v ⃗ × B )
⃗
• Therefore, the work done by the electric field on a point charge q is:
B
B B
d l ⃗ ⋅ (q E )⃗ = − q d l ⃗ ⋅ ∇ ⃗ ϕ = = − q ΔϕAB
∫A ∫A
WAB =
WA→∞ = q ϕA
• This means that we can associate this energy at any given point with a potential energy:
WA→∞ → U( x A⃗ ) ⟹ F (⃗ x A⃗ ) = − [ ∇ ⃗ U] = − q [ ∇ ⃗ ϕ] = q E( x A⃗ )
x A⃗ x A⃗
A
⃗
2
1 r̂ ℂ
F q1 q2 = q1 q2 2 , ϵ0 = 8.85 × 10−12 , 1ℂ = 6.2 × 1018 e
4π ϵ0 r N m2
• Electric field:
F q⃗ = q E ⃗
N
⇒ [E] =
ℂ
• Electrostatic potential:
E ⃗ = − ∇ ⃗ϕ
Nm
⇒ [ϕ] = [E]m = = V (Volt)
ℂ
N m N m /s W V
1V= = = (Watts/Ampère) , and [E] =
ℂ ℂ/s A m
• Example: a spark produces fields of strength 103 − 104 V/m
Warming up…
Spherical coordinates:
∇ ⃗ϕ = r ̂
∂ϕ 1 ∂ϕ 1 ∂ϕ
+ θ̂ + θ̂
∂r r ∂θ r sin φ ∂φ
⃗ ⃗ 1 ∂ r 2 Er ∂ sin θ Eθ ∂ Eφ
r sin θ [ ∂φ ]
1
∇⋅E = 2 + +
r ∂r ∂θ
• Let’s warm up our brain muscles by working out some very, very simple examples.
• First, let’s compute the electric field of a spherically symmetric charge density ρ(r) . 1.0
0.8
⃗ ⃗ ρ(r)
0.6
2
∇ ⋅ E =−∇ ϕ = , ρ(r) → ρ
0.4
ϵ0
0.2
0.0
-0.2
0 2 4 6 8
• Using spherical symmetry we can use the integral form of this equation: 100
50
d S ⃗⋅ E ⃗ =
Q(r) Q
∮r
0
0 2 4 6 8
0.4
r
Q(r)
∫0
0.3
ϵ0 ϵ0 E 0.1
0.0
-0.1
• We can also compute the potential, using that, for spherical symmetry, we have -0.2
15 0 2 4 6 8
r
∂ 1 Q(r′) ϵ0 ϕ
10
∫
− ϕ(r) = Er(r) ⟹ ϕ(r) = − dr′ 2
∂r 4πϵ0 0 r′
5
0 2 4 6 8
r
ELECTRODYNAMICS I / IFUSP / LECTURE 1 14






Warming up…
Cylindrical coordinates:
∇ ⃗ϕ = s ̂
∂ϕ 1 ∂ϕ ∂ϕ
+ φ̂ + ẑ
∂s s ∂φ ∂z
1 ∂ s Es 1 ∂ Eφ ∂Ez
∇⃗ ⋅ E ⃗ = + +
s ∂s s ∂φ ∂z
∇ ⃗ ⋅ E ⃗ = − ∇2 ϕ =
ρ(s)
ϵ0
• Using axial symmetry we can try to write the integral form of this equation:
z
d S ⃗⋅ E ⃗ =
Q(s)
∮s
φ s
ϵ0
• However, notice that in this case the volume (and the charge) can diverge. Nevertheless, the field should not
blow up. Let’s write this for a very very long cylinder of height Z, for which we can neglect the contribution
from the top and bottom:
Z s Z 1 λ(s)
ϵ0 ∫0
(2πs Z ) Es = ds′ s′ ρ(s′) = λ(s) ⟹ Es =
ϵ0 ϵ0 2π s
• The potential can be computed in the same way from:
s
∂ 1 λ(s′)
2πϵ0 ∫0
− ϕ(s) = Es(s) ⟹ ϕ(s) = − ds′
∂s s′
• Notice that unless the linear charge density λ → 0 for s → ∞ , then | ϕ | → ∞ !! Why??…
Warming up…
Cartesian coordinates:
∇ ⃗ϕ = x̂
∂ϕ ∂ϕ ∂ϕ
+ ŷ + ẑ
∂x ∂y ∂z
∂ Ex ∂Ey ∂Ez
∇⃗ ⋅ E ⃗ = + +
∂x ∂y ∂z
⃗ ⃗ 2 ρ(z)
• Basic example #3: planar symmetry/Cartesian coordinates: ∇ ⋅ E = − ∇ ϕ =
ϵ0
• Using planar symmetry we can again try to write the integral form of this equation:
d S ⃗⋅ E ⃗ =
Q(z)
∮s ϵ0
• However, once again the volume (and the charge) can diverge, since the area in the plane x-y diverges. Nevertheless, the field should not
blow up at some position.
• In addition, we should specify what is exactly this closed. Let’s assume that we integrate from z = − ∞ to some position z . We then have, for
some area A in the plane x-y:
A z A
ϵ0 ∫−∞
A [Ez(z) − Ez(z → − ∞)] = dz′ ρ(z′) = σ (z)
ϵ0
σ (z)
⟹ Ez + Ez,∞ = , where we defined the field at z → − ∞ in terms of the field at z → + ∞ : Ez(z → − ∞) ≡ − Ez,∞
ϵ0
• If the charge is bounded to some region not too far from the plane z = 0, then we can determine the value of the field at infinity. Since there is
no difference between “up" and “down”, we can take z → + ∞ and write:
z
1 ∞
ϵ0 [ ∫−∞ ]
σ σ (z) σ∞ 1
2 ∫−∞
2 Ez,∞ = ∞ ⟹ Ez = − = dz′ρ(z′) − dz′ρ(z′)
ϵ0 ϵ0 2ϵ0
• But we can rewrite the first integral as:
1 1 z 1 z 1 ∞ z ∞
ϵ0 [ 2 ∫−∞ ] 2 ϵ0 [ ∫−∞ ]
1
2 ∫−∞ 2 ∫−∞ ∫z
Ez = dz′ρ(z′) + dz′ρ(z′) − dz′ρ(z′) ⟹ Ez = dz′ρ(z′) − dz′ρ(z′)
Warming up…
• The simplest application would be for a pair of planes of equal and opposite charge — a
capacitor . In that case we have, between the plates:
2ϵ0 ( 2ϵ0 )
σ σ σ
Ez = − − =+ ,
ϵ0
2ϵ0 ( 2ϵ0 )
σ σ −σ
Ez = + − =0 .
d
+σ
• Therefore we get the field inside the capacitor:
σ
Ez =
ϵ0
• The electric potential is, therefore:
Capacitance:
σ σ q
ϕ = − z + C , and the potential difference is Δϕ = d = d q Aϵ0
ϵ0 ϵ0 A ϵ0 C= =
Δϕ d
F ⃗= q E ⃗ ⇒ F ⃗ = − ∇ ⃗ U = q ( − ∇ ⃗ φ) U = qφ
• Therefore, the potential energy of a small charge that is placed in this capacitor is:
q
dU = dq Δϕ = dq
C
• As we add more and more charges to the capacitor (bringing them in pairs from ± ∞ to each plate), the energy
increases such that:
A
q
−σ
2
q 1q
∫0
U(q) = dq = d
C 2 C
+σ
• But E = q/(Aϵ0) , so q = E × Aϵ0 , and we can write:
1 E 2 A 2ϵ02 1 2 1 2
U(q) = = ϵ0 (A d ) E = ϵ0 V E
2 Aϵ0 2 2
d
• This, of course, is simply the statement that the energy density in the electric field is given by: Capacitance:
1 q Aϵ0
ρE = ϵ0 E 2 C= =
2
Δϕ d
Boundary conditions
Perfect
• In these examples we have only considered charge distributions which are somehow bounded in one or E ⊥⃗ = 0 conductor
more dimensions, but the space is zero everywhere else, such that the field drops to zero at infinity.
• Now let’s suppose that we introduce some non-trivial boundary conditions, which constrain the field
over some surface (or surfaces).
• There are (generically) two types of boundary conditions for a scalar field (ϕ) on a given surface:
⃗ ϕ =0
Neumann b.c. : ( ∇ ⊥ )S
• Now, how can we accommodate all of this into our formal solution in terms of the Green function?
ρ( x ⃗ ) 1 1
∇2x ϕ( x ⃗ ) = − ⇒ ∇2 G( x ,⃗ x ′⃗ ) = δ( x ⃗ − x ′⃗ ) ⇒ G( x ,⃗ x ′⃗ ) = −
ϵ0 4π | x ⃗ − x ′⃗ |
• Therefore, we obtain:
1 ρ( x ′⃗ )
ϕ( x ⃗ ) =
4π ϵ0 ∫
d 3 x′
| x ⃗ − x ′⃗ |
Green’s functions
are not
• But what about the boundary conditions?… unique!
• At issue here is the fact that the Green function is not unique — it is degenerate with respect to terms which solve the
homogeneous equation:
∇2 G( x ,⃗ x ′⃗ ) = δ( x ⃗ − x ′⃗ ) ⇒ G → G + ϕh , where
1 1
G( x ,⃗ x ′⃗ ) = − + ϕh( x ⃗ )
4π | x ⃗ − x ′⃗ |
• Now, what are these homogeneous solutions? From the equation ∇2 ϕh = 0 we get immediately that some of those solutions are:
ϕh( x ⃗ ) = ϕ0 − ℰ 0⃗ ⋅ x ⃗ ,
ELECTRODYNAMICS I / IFUSP / LECTURE 1 * There is a caveat: radiation! We will get back to this later…22
E⃗→ 0
• The harmonic functions are one way in which we can impose boundary conditions over some
surface.
• Another way uses "copies" of the Green function itself.
• A good way of visualizing the basic Green function is to regard it as the potential of a point
charge:
ρq( x ⃗ )
[ ϵ0 4π | x ⃗ − x q⃗ | ]
q 1 1 q
2
∇ ϕq = ∇ 2
× = − δ( x ⃗ − x q⃗ ) = −
ϵ0 ϵ0
• Now, suppose that our problem is such that a whole region is shielded from us by the boundary.
Anything that goes inside of that boundary is irrelevant if we want to compute the fields on the
outside region — this is guaranteed by the Helmholtz Theorem.
• This means, in particular, that we can in principle put as many “point charges” inside that shielded
region as we want — after all, in that region, ∇2 ϕq( ∈ V ) = − (q/ϵ0) δ( x ⃗ − x q⃗ ) = 0
• The only issue is that we must pick those charges, as well as their positions, carefully such that we
ry
end up with the boundary conditions that we want to impose.
da
un
bo
ELECTRODYNAMICS I / IFUSP / LECTURE 1 23
SHIELDED
REGION
• In its simplest incarnation, the method is applied to the case of a point charge q at a position
x q⃗ = {0,0,Z0}, but in the presence of an infinite plane (at z = 0) that is also grounded —
hence, ϕ(z = 0) = 0 .
∇2 ϕq′ = 0
• In the absence of the plane, the potential of the charge would be simply:
here !
q 1 1
ϕq = ×
ϵ0 4π | x ⃗ − x q⃗ |
• But the plane changes everything, of course, because the potential above does not satisfy q
ϕ(z = 0) = 0 .
• Now, notice that the charge and the observer are in fact assumed to be in the z > 0 region, so q′
anything that happens at z < 0 is quite irrelevant: all the physics is determined by what
happens in the volume z > 0 together with the conditions at the boundary z = 0. ∇2 ϕq′ ≠ 0
• This means that we could add a “homogeneous term” like the one we found in the previous here , but
page: ϕq′ , where the charge q′ is located in the region z < 0 ! who cares?
• If we think a bit about this, the “phantom" point charge in the z < 0 region will
cancel the potential of the “real" charge if it has an equal but opposite charge
⃗ = {0,0, − Z0}
(q′ = − q), and if it is placed exactly opposite the real charge: x q′
• Then, the potential of the two charges (the real and the “phantom”, or “image"
charge), is:
q 1 q 1
ϕ= −
4πϵ0 | x ⃗ − x q⃗ | 4πϵ0 | x ⃗ − x q′
⃗ |
4πϵ0 ( | x ⃗ − Z0 z ̂ | | x ⃗ + Z0 z ̂ | )
q 1 1
ϕ= − (surface charges induced)
• Notice that for z = 0 the two terms above cancel out, giving us ϕ(z = 0) = 0 .
[A question for you: suppose that we have only a very thin conducting plate at z = 0,
and that somebody is behind that plate. What would that observer measure?….]
Next class:
• Solutions to the Laplace equations with boundary conditions: Green’s functions
for Dirichlet and Neumann conditions
• Variational techniques
• Relaxation method
• Jackson, Ch. 1