11/3/03

Poisson and Laplace's Equations

free charge!

E=0

E =

Substituting E = into Gauss's law yields

Assuming that the permittivity is a constant throughout the given region (i.e. the medium is homogeneous), then can be taken out of the divergence operation. Thus,
( ) =

v .

free charge!

Substituting for the del operator in rectangular coordinates,

we obtain Poisson's equation:

Poisson's Equation

2 2 2 in rectangular coordinates and is referred to as the + + x 2 y 2 z 2 Laplacian operator. In cylindrical coordinates, it is

The operator 2 =

 2 =

1 1 2 2 + 2 + 2 2 z

and in spherical coordinates is

Poissons equation reduces to Laplace's equation in a region in which there are no volume sources. That is, letting v = 0 in Poisson's equation yields
2 = 0

Laplace's Equation

It is important to recognize that the potential may be obtained by solving these equations with the correct sources and subject to the appropriate boundary conditions.

EXAMPLE: Parallel-Plate Capacitor = Vo

s h - s x
=0

Vo

Note a reference potential = 0 has been chosen on the bottom platethis choice is arbitrary. But then the potential on the top conductor must be = Vo to have the correct potential difference as determined by the impressed DC source. Since the region between the conductors can be considered source free (no free charge should exist between the plates, so v = 0 there), Poissons equation becomes Laplace's equation,
2 = 2 2 2 + + = 0, x 2 y 2 z 2

which can be applied to solve for the potential at any point between the conducting plates. At this point the analysis can be simplified by assuming that the conductors are infinite in and y directions and, consequently, that the potential does not vary in these both the z directions ( z = y = 0 ) . Thus, Laplaces equation becomes
d 2 ( x ) = 0. d x2 Note that the derivatives are written above as ordinary rather than partial derivatives since is assumed to depend only on the single variable x . Integrating the above differential equation twice with respect to x, the general solution for the potential is given by = C1 x + C2 . Enforcing the boundary conditions, ( x = 0 ) = 0 and ( x = d ) = Vo , the constants C1 and C2 are obtained. This results in ( x) =

Vo x d

for the final solution of the potential between the conducting plates of the parallel-plate capacitor. x Recall that the capacitance of a parallel+ + + + + + + + + + x=d plate capacitor was previously found by Vo using Gauss's law to compute the electric field between the conductors. Having - - - - - - - - - - - - - - - - - - - x=0 derived the potential function between the conducting plates, the electric field can also be determined using E = . Thus, E = x

V = o x d x

and, consequently, the electric flux density is given by

D=E=

- Vo . x d
= s ( D1 D2 ) n

Next, the surface charge density is found by applying

(with

= x ) to the top conductor (referenced at positive Vo volts). This results in D2 = 0, n

s =
and ultimately, a parallel-plate capacitance of

Vo
d

C=

Q s A A = = . V Vo d

Note that this is consistent with the capacitance expression derived previously.

EXAMPLE: Coaxial Line:

Neglecting fringing at the ends of the coaxial line, we may assume that the equipotential surfaces are the = constant surfaces. Thus, is assumed to vary only with : i.e., = ( ) . Since there is no volume charge between the conductors, Laplaces equation in cylindrical coordinates applies:

= V0 a b
=0

2 =

1 d d = 0. d d

This differential equation may be solved by integrating twice to obtain

= A ln + B .
Applying the boundary conditions shown in the figure at = a and = b yields
ln = V0 b

[V], a< <b . b ln a Hence the electric field between the conductors is

E = =

V0 d = [V/m], b d ln a

a< <b .

The (positive) surface charge density on the inner conductor is therefore D

b a and hence the total charge on an inner conductor of length is a ln Q = Sa 2 a = 2 V0 b ln a [C] .

= Sa =

V0

[C/m 2 ]

The total line capacitance is therefore C = Q 2 [F] = b V0 ln a

and the capacitance per unit length is C = C 2 = b ln a [F/m] .

Method of Images
Oftentimes practical electrostatic problems involve charges located near planar conductors. For example, consider the situation below.

z q d x PEC

According to the previous discussion, to find the potential in the upper half-space, it is necessary to solve the Poisson equation in the region z>0 and then enforce the boundary conditions = 0 at z = 0 and 0 as r . Recall the solution for the potential of a point charge in free-space:

q 4 x 2 + y 2 + ( z h )
2

Although this potential function satisfies Poisson's equation in the upper half-space, it does not satisfy the necessary boundary conditions at x = 0 . Thus, it is not a valid solution to the scenario presented above. Rather than solve Poisson's equation directly, consider the following situation using "image charges".

z q d q'
Notice that in this problem the conducting plane has been removed and a charge q' has been placed at z = d . Thus, the potential in the upper half plane can be written as q 4 o x 2 + y 2 + ( z + d )
2

( x, y , z ) =

q 4 o x 2 + y 2 + ( z d )
2

[V]

Notice that 0 as r and that if the charge q' is selected such that q = q then = 0 at z = 0 . This latter condition makes the plane z = 0 an equipotential surface, which in turn ensures that the tangential component of the electric field vanishes there. In addition, it is important to recognize that this solution, constructed from known potential functions for point charges, automatically satisfy Poisson's equation everywhere. Note that the new potential is only valid in the region z > 0 . (The actual potential for z < 0 is zero! Why?) Thus, all conditions imposed on the original problem (charge near a conducting ground plane) are satisfied by this potential function. According to the uniqueness theorem, which will not be proven here, a solution that 1) satisfies Poisson's equation and 2) satisfies the given set of boundary conditions is the

only possible solution. This holds true irrespective of the method by which the solution is obtained. Although the method of images was presently used to arrive at the correct solution for the potential function of a point charge over a ground plane, a solution obtained by using educated guesses would have been equally valid if it satisfied the criteria 1) and 2) above.

Using the potential function given above, the electric field in the region z 0 is given by
+ yy +(z d )z + yy + (z + d)z xx q xx [V/m 2 ] E = = 4 x 2 + y 2 + ( z d )2 3/ 2 x 2 + y 2 + ( z + d )2 3/ 2

and the surface charge density on the ground plane by

s = Dz

z =0

= Ez

z =0

=
z =0

2qd [C/m 2 ] 2 2 4 ( x + y + d )
2

Notice that the total charge on the ground plane (charge in Coulombs) is calculated using QGrdPlane = which, in turn, results in QGrdPlane = q . Thus, as expected, the original charge q induces equal and opposite charge on the ground plane.

qd
2 2 2 2 x + y + d 3/ 2

dx dy