Defined electric field by F = q E
Can be expressed by field lines
Defined flux of electric field
Φ =
E ⋅ d a
Note the sign convention: positive if coming out
Gauss’s Law
E
d a = 4 π q
Useful for solving E fields with symmetries
Spherical charge distribution
E
E
=
⎧
⎨
⎪
Q r 2
Qr
ˆ for r
≥
R
ˆ for r
<
R
R 3
Q
R 2
R
R
S E r r
1
Continue with Gauss’s Law
Apply to infinite sheets of charge
Discuss energy in the electric field
Empty space with E has energy?
Define electric potential ϕ
by line-integrating electric field
Closely related to energy
Vector calculus connects electric potential to electric field and vice versa
Derive differential form of Gauss’s Law
Connect electric field and charge density
More vector calculus
J.C.F. Gauss, 1777-1855
Problem: Calculate the electric field at a distance z from a positively charged infinite plane charge
Surface charge density: σ =
Use Gauss again area z
Which surface to use? E
What symmetry do we have?
Consider a cylinder
Area A and height 2 z
E field must be vertical
How do we know that?
2
Total flux Φ total
= Φ top
+ Φ side
+ Φ bottom
Side is parallel to E E
⋅ d a
=
0 No flux
Top and bottom are symmetric Same flux
Φ total
=
2
Φ top
=
2 AE ( z )
Charge inside the cylinder is q inside
=
A
σ
Using Gauss
E ( z )
=
2
πσ
Don’t forget the direction! z
The result is worth remembering:
E
=
⎩⎪
+
2
ˆ z for z
>
0
−
2
ˆ z for z
<
0
E
Infinite sheet of charge produces uniform E field of 2 πσ above and below
Place two oppositely-charged large sheets in parallel
Consider an area A of them z
E top
E fields from the two sheets overlap and add up + σ
Between the sheets: E = 4 πσ
Cancel each other outside
−σ
Two sheets also attract each other (obviously)
Top sheet feels
E bottom
= −
2
πσ ˆ
The force on area A of the top sheet is
F
= σ
A E bottom
= −
2
πσ 2 A ˆ = −
E 2
8
π
A ˆ
E bottom
3
Imagine we move the top sheet upward by a distance d
We must do work W = Fd =
E
8
π
2
Ad
The energy of the system increases + σ by W z
E top
E bottom
Q: Where exactly is this energy?
−σ
Note that the volume of the space between the sheets increased by Ad
This is also where E field exists
Space with E holds energy with a volume density u
=
E
8
π
2
Total electrostatic energy of a system is
U
=
E 2
8
π dV
Will come back to this…
Electrostatic force is conservative
I said this in Lecture 1 without proof
Given F = q E , the above statement is equivalent to
Line integral
P
2
P
1
E ⋅ d s is path independent
Thanks to the Superposition Principle, we have only to prove this for the electric field generated by a single point charge
r
E = q r 2 r ˆ d s
r
2
Line integral
r
1 r
2 q r 2 r ˆ ⋅ d s is path independent r
1
q
4
r movement, i.e. dr
ˆ ⋅ d s
r
1 r
2 r q
2 r ˆ ⋅ d s =
r
1 r
2 r q
2 dr = q
⎛
⎝⎜
1 r
1
−
1 r
2
⎞
⎠⎟
The integral depends only on r
1 i.e., is path-independent
and r
2
,
Generalize using Superposition:
Line integral ∫ P
2
P
1
E ⋅ d s for any electrostatic field E has the same value for all paths from P
1
to P
2
q
r
1
dr
r d s
E = q r 2 r ˆ
r 2
For the special case of P
1
= P
2
, the path becomes a loop
Line integral ∫ E ⋅ d s of an electrostatic field around any closed path is zero
This is equivalent to the path-independence d s
A
Consider two paths (A and B) from P
1 to P
2
Path independence means
P
2
P
1
E ⋅ d s
A
=
Corollary above means
P
2
P
1
E ⋅ d s
B
P
1 d s
B
P
2
P
1
E ⋅ d s
A
+
P
1
P
2
E ⋅ ( − d s
B
) = 0
Will use this later when we do the “curl”
P
2
5
φ
21
P
2
P
1
≡ − E ⋅ d s
∫
P
1
P
2
Electric potential difference between P
1
and P
2
Note the negative sign!
To move a charge q from P
1
to P
2
, you must do work
W =
P
2
P
1
− q E ⋅ d s = q φ
21
As a result, the energy of the system increases by q ϕ
21
We can fix P
1
to a reference point and re-interpret this as a scalar function of the position r
r
( r ) ≡ − E ⋅ d s
0
Electric potential at position r
Reference point for the electric potential is arbitrary
If you choose e.g., point B instead of point A as the reference
φ ref = B
( r ) = − ∫
B r
E ⋅ d s r
= − E ⋅ d s
A
∫ A ∫ −
B
E ⋅ d s = φ ref = A
( r )
Electric potential is defined up to an arbitrary constant
+ const .
Just like an indefinite integral
The potential difference is physical (linked to work and energy) and must be free from arbitrary constant
φ
21
= − ∫ P
2
P
1
E ⋅ d s = − ∫
0
P
2 E ⋅ d s + ∫
0
P
1
The constant cancels in the difference
E ⋅ d s = φ ( P
2
) − φ ( P
1
)
6
Dimension of electric potential is (electric field) × (length)
Since electric field is (force)/(charge), this equals to (force) × (length)/
(charge) = (energy)/(charge)
Unit of electric potential is erg/esu = statvolt
In SI , the unit of electric potential is volt = joule/coulomb
300 volt = 1 statvolt
1 coulomb = 3 × 10 9 esu
Since the rest of the world uses SI, we will convert to SI when we deal with real-world (esp. EE) problems
Recall in calculus: F ( x ) =
f ( x ) dx f ( x ) =
We can “reverse” the line integral as well d dx
F ( x )
φ = −
E ⋅ d s
E
= −∇ φ gradient
Gradient is “ how quickly the function ϕ varies in space ”
In x-y-z coordinate system:
∇ φ
NB: gradient is a vector field
= ˆ
∂ φ
∂ x
+ ˆ
∂ φ
∂ y
+ ˆ
∂ φ
∂ z
It points the direction of maximum rate of increase (= uphill)
Hence E points downhill of ϕ
7
For a potential φ ( x , y ) = sin x ⋅ sin y
∇ φ ( x , y ) = ˆ cos( x )sin( y ) + ˆ sin x cos y
∇ φ ( x , y )
φ ( x , y )
© Prof. G. Sciolla, MIT
φ ( x , y ) = sin x with lines of constant values of ϕ
⋅ sin y
In 3-d, we find surfaces of constant potential
= equipotential surfaces
For a single point charge, equipotential surfaces are concentric spheres
Electric field is perpendicular to the equipotential surfaces
This is generally true for gradient of any function
∇ φ ( x , y )
© Prof. G. Sciolla, MIT
8
Electric potential due to a charge distribution is
φ =
N
j = 1 q r j j discrete
φ = ∫
Continuous case maybe 1-d, 2-d, or 3-d dq r continuous
φ =
λ r d φ =
σ r da φ =
ρ r dv
Since potential is a scalar, it is often easier to calculate than the electric field
Once you have ϕ , you can always get E from − ∇ ϕ
An example is in order
Problem: Calculate the electric potential produced by a thin, uniformly charged disk on its axis
z
Disk radius = a , surface charge density σ
Slice the disk into rings, then into the red bits
Charge on the red bit is σ sdsd θ
The potential due to the red bit is
Integrate: d φ =
σ sdsd θ z 2 + s 2
φ = σ
0
2 π d θ
= 2 πσ
⎡
⎣⎢ z
= 2 πσ
( z 2
2
0 a sds
+ s 2
+ a 2 z 2 + s 2
⎤
⎦⎥ s = a s = 0
− z
)
NB: absolute value in case z < 0
ds s
a
d
θ
9
Can we calculate E
)
(
We need to take gradient, which means we need to know the x, y, and zdependence
We calculated ϕ only on the z axis
We are in luck — we know E is parallel to the z axis by symmetry
E = −∇ φ = − ˆ
= − 2 πσ
⎝⎜
∂
∂ z
= 2 πσ
⎛
−
(
∂ φ
∂ z z 2 z 2 z
+ a 2
+ a 2
± 1
⎞
⎠⎟
− z
) z ˆ
ˆ for z > 0 z < 0
ds s
z
a
d
θ
φ = 2 πσ
( z 2 + a 2 − z
)
E = 2 πσ
⎛
⎝⎜
− z 2 z
+ a 2
± 1
⎞
⎠⎟
ˆ for z > 0 z < 0
Dimension: [ ϕ ] = (charge)/(length), [ E ] = (charge)/(length) 2
NB: [ σ ] = (charge)/(length) 2
φ
Far away from the disk (| z | >> a )
= 2 πσ z
(
1 + ( a z ) 2
− 1
)
⎯ z a → 2 πσ
Same as a point charge with Q = π a 2 σ z
(
1 + 1
2
( a z ) 2 − 1
)
=
π a 2 σ z
Close to the disk (| z | << a )
E
⎯ ⎯⎯ ± 2 πσ ˆ
Same as the infinite sheet of charge
10
We know a system of N charges has a total energy of
U =
1
2
N
∑ j = 1
∑ k ≠ j q j r jk q k =
1
2
N
∑ j = 1 q j
∑ k ≠ j q k r jk
=
1
2
N
∑ j = 1 q j
∑ k ≠ j
Generalizing to continuous distributions
φ jk
U =
1
2
ρφ dv
Potential at the jth charge due to the
other charges
Integrate entire space, or where ρ ≠ 0
No need to avoid j = k because “individual” charge is infinitesimally small
Does the factor 1/2 make sense?
Imagine increasing ρ slowly everywhere as
The potential is proportional to ρ , i.e.,
Work to go from s to s + ds is
Integrate W = ∫
0
1 s ds ∫ ρφ dv dW
=
1
2
∫
=
ρφ
φ
(
′ d
= dv
ρ ′ s
)
φ
ρ ′ = s ρ s = 0 → 1
φ ′ dv =
( ds ρ )( s φ ) dv
Will come back to this…
Charge is distributed with a volume density ρ ( r )
Draw a surface S enclosing a volume V
Guass’s Law:
∫
S
E ⋅ d a = 4 ∫
V dv Total charge in V
Now, make V so small that ρ is constant inside V
∫
S
E ⋅ d a = 4 πρ V for very small V
As we make V smaller, the total flux out of S scales with V
Therefore: lim
V → 0
S
E ⋅ d a
V
= 4 πρ
LHS is “ how much E is flowing out per unit volume ”
Let’s call it the divergence of E
11
div E ≡ lim
V → 0
S
E ⋅ d a
V
= 4 πρ
In the smallV limit, the integral depend on volume, but not on the shape We can use a rectangular box
Consider the left ( S
1
) and right ( S
2
) walls
S
1
∫
S
2
E ⋅ d a
E ⋅ d a
Sum =
(
E
= E ( x , y , z ) ⋅ ( − ˆ ) dydz
= E ( x + dx , y , z ) ⋅ ˆ dydz x
( x + dx ) − E x
( x )
) dydz
=
∂ E x
∂ x dxdydz
Add up all walls:
S
E ⋅ d a =
⎛
⎝
∂ E x
∂ x
+
∂ E y
∂ y dy
+ dz
S
1
E ( x , y , z )
S
2 dx
E ( x + dx , y , z )
∂ E z
∂ z
⎞
⎠
V =
( )
V div E
We now have Gauss’s Law for a very small volume/surface
div E = 4 πρ
where div E ≡ ∇ ⋅ E =
⎛
⎝
⎜
∂ E
∂ x x +
∂
∂
E y y +
∂
∂
E z z
Connects local properties of E with the local charge density
⎟
⎞
⎠
Integrate over a volume and you get Gauss’s Law back
Gauss’s Divergence Theorem (this is math!)
For any vector field F,
V
div F dv =
A
F ⋅ d a
Applying this theorem to Gauss’s Law (of electromagnetism) gives us the integral and differential versions we now know
12
Let’s calculate div E for
E = r q
2 r ˆ
We can do this by expressing E in x-y-z : E = x
Or we can express div in spherical coordinates
2 q
+ y 2 + z 2
∇ ⋅ F =
Since only E r r
1
2
∂ ( r 2 F r
∂ r
)
+ r
1 sin θ
is non-zero, we get
∂ ( F
θ sin
∂ θ
θ )
+
1 r sin θ
∂ F
φ
∂ φ x ˆ x + ˆ y + ˆ z
∇ ⋅ E =
1 r 2
∂ ( r 2 E r
∂ r
)
=
1 r 2
∂ q
∂ r
= 0
This is correct — we have no charge except at r = 0 x 2 + y 2 + z 2
At r = 0, 1/ r 2 gives us an infinity
That’s OK because a “point” charge has an infinite density
Let’s give the “point” charge a small radius R
We did this in Lecture 2, and the solution was
E
=
⎧
⎨
⎪
Q r 2 r ˆ for r
≥
R
Qr r ˆ for r
<
R
R 3
This part is same as a point charge.
We know div E = 0.
Let’s work on this part
For r < R , ∇ ⋅ E =
1 r 2
∂ ( r 2 E r
∂ r
)
=
1 r 2
∂
∂ r ⎝⎜
⎛ Qr 3 ⎞
R 3 ⎠⎟
=
3 Q
R 3
The charge density of the sphere is ρ =
4 π
3
Q
R 3
It works everywhere (as long as ρ is finite)
∇ ⋅ E = 4 πρ
13
2
We found earlier and
Are they same?
U =
8 π dV U =
1
2
ρφ dv
Consider the divergence of the product E ϕ
∇ ⋅ ( E φ ) =
∂
∂ x
( E x
φ ) +
∂
∂ y
( E y
φ ) +
∂
∂ z
( E z
φ )
=
∂ E x
∂ x
φ + E x
∂ φ
∂ x
+
∂ E y
∂ y
φ + E y
∂ φ
∂ y
+
∂ E z
∂ z
φ + E z
∂ φ
∂ z
= ( ∇ ⋅ E ) φ + E ⋅ ( ∇ φ ) = 4 πρφ − E 2
Integrate LHS over very large volume and use Divergence Theorem
V
∇ ⋅ ( E φ ) dv =
S
( E φ ) ⋅ d a
Integral of RHS must be 0, too
= 0 assuming E φ → 0 at far away
4
dv −
E 2 dv = 0
1
2
ρφ dv =
E
8 π
2 dv
Used Gauss’s Law on infinite sheet of charge
Defined electric potential by line integral
Uniform electric field E = 2 πσ above and below the sheet
Electric field has energy with volume density given by
φ
21
= −
P
2
P
1
E ⋅ d s = φ ( P
2
) − φ (
Electric field is negative gradient of electric potential
Differential form of Guass’s Law:
P
1
)
φ =
N
j = 1
E u
= unit: erg/esu = statvolt q r
= j
4
E
φ =
Equivalent to the integral form via Divergence Theorem
∫
= dq r
E
8
π
2
−∇ φ
V div F dv =
A
F ⋅ d a
14