5.

1 Magnetic Scalar Potential 91

In formal analogy with the electric field, the magnetic field

strength is calculated as the gradient of this scalar potential
(the negative sign is arbitrary and serves only to emphasize the
analogy, see also 4.1)

H = - grad 'Pm
(5-2)

Taking the vortex density of both sides gives

curl H =curl grad 'Pm = 0 , (5-3)

because the differential operators curl and grad sequentially ap-

plied to an arbitrary scalar field always yield zero (see 4.2).
Moreover, this implies that only such functions as H(x,y,z) can
be represented as the gradient of a scalar field which obey curl
H = O. The latter constraint is complied with only in current-
free regions of magnetostatic vortex fields, i.e. outside current
carrying conductors. According to Section 3.2.2, curl H = J ap-
plies to the interior of the conductors (see also Chapter 6).
Hence, use of the magnetic scalar potential is restricted to
vortex-Jree, or current-Jree, regions.

As in the electrostatic field, a reference potential 'Pmo can be

arbitrarily chosen. However, this potential need not guarantee
uniqueness. Evaluating the magnetic potential-difference be-
tween two pOints PerIl and P(r2) from the familiar line integral
frequently used in electrostatic fields (see 4.2),

r1

'Pm12 = - JHodr
r2
(5-4)
92 5 Potential and Potential Function or Magnetostatic Fields

the integration path must not enclose.a current because the in-
tegral

fH. dr (5-5)

is not path-independent. Whereas in electrostatic fields we al-

ways have

(5-6)

in magnetostatic fields we will find

f H· dr = NI , (5-7)

this result being dependent on the number N of encirclements.

Choosing different paths between two points may yield different
values for the potential difference. Although we will not
elaborate on the explicit calculation of the scalar potential, the
reader will get an idea of the procedure when we derive the
magnetic vector potential in section 5.3.

5.2 Potential Equation of Magnetic Scalar Potentials

In this section we will show that a similar potential equation
V2<pm = 0 exists for the magnetic scalar potential as exists for
the scalar potential of an electrostatic field.

According to Section 3.2.4, magnetic fields always obey

divB=O. (5-8)
S.2 Potential Equation of Magnetic Scalar Potentials 93

Together with H = - grad q>m and the constitutive relation

B = ~. we obtain

div ~ =0 or div grad q>m = 0 . (5-9)

For instance. for a scalar field q>m(x.y.z) in a Cartesian coordi-

nate system. the result of the gradient operation will be

grad CI>m(x.y.z) = a:; + a:;

8.x By + a:; 8z

(5-10)

Taking the divergence of this vector field according to equation

(4-43) of Section 4.4.1. we finally obtain a potential equation
V2q>m = O. having the same fonn as V2q> = O.

(5-11 )

or. more compactly.

(5-12)

Hence. calculation of magnetic fields outside current-carrying

conductors can be accomplished with the same methods em-
ployed in electrostatic source fields. Upon integration of V2q>m =
O. i.e. solving the potential equation for given boundary
conditions. one obtains the magnetic field strength H(r) by dif-
ferentiating the scalar magnetic potential q>m(r). A typical prob-
lem that can be solved with the scalar potential is the evaluation
94 5 Potential and Potential Function or Magnetostatic Fields

of the magnetostatic shielding efficiency of a ferromagnetic

shield.

5.3 Magnetic Vector Potential

As has been mentioned above, the magnetic scalar potential faUs

inside current-carrying conductors, for example in calculating
shielding efficiencies for time-varying magnetic vortex fields or
in treating wave problems in which the magnetic effects of the
displacement currents outside current-carrying conductors
cannot be neglected (see 6.1). Fortunately, in addition to the
magnetic scalar potential <Pm(r), a magnetic vector potential
A(r) can be defined which is also applicable in regions with
current flow.

Because of the absence of sources in magnetic vortex fields we

have div B = 0 and, since the differential operators div and curl
sequentially applied to an arbitrary vector field A always yield
zero, we have div curl A = 0 (see A3). Upon equating and inte-
grating both equations (undoing the div operation, {div}-l) the
flux density B can be formally represented as the vortex density
of a vortex field A (see also 4.2, 4.5 and A4). Thus,

div B = div curl A , (5-13)

followed by {div}-l, yields

B = curl A or curIA = B
(5-14)

The differential operation div curl A = 0 can be understood not

only formally but also conceptually. The vortex densities curl A
of a magnetic vector potential A are the flux-denSity lines B of
the magnetic fteid considered. These are always solenoidal (see
Section 3.2.4) Figure 5.1.
5.3 Magnetic Vector Potentials 9S

<!>,B

Figure 5.1: Illustration of the vector potential A as a vortex field with

vortex density curl A = B. A and B obey the right-hand
rule.

In Section 2.2 we have already seen that vortices and vortex

densities, are, respectively, solenoidal tubes and lines whose
divergences are clearly zero. With this in mind, the divergence
of the vortex density of the vortex field A must also be zero, i.e.
div curl A =div B = O. The magnetic vector potential A is a vor-
tex field, its vortices are flux tubes; its vortex densities are flux-
density lines B. The vortex strength of the vector potential is
calculated from

(5-15)

.and gives the magnetic flux.

Its vortex density is by definition

curl A = B
(5-16)
96 5 Potential and Potential Function of Magnetostatic Fields

In order to underscore the linkage between these equations,

recall the induction law in'integral and differential form (see
3.1.1, 3.3.1)

curl E = -
aB
at (5-17)

which result from differentiating the preceding equations with

respect to time and recognizing dA/dt = - Ev (see 6.2.1).

For the magnetic vector potential, the question of uniqueness

must again be raised. Recall that in the electrostatic field
different scalar potential functions <per) may be associated with
the same vector field E(r)

E(r) = - grad <per) = - grad (<p(r) + <po) := - grad <p*(r). (5-18)

In like manner, in magnetostatlc fields various vector potential

functions A(r) may exist for one and the same magnetic flux-
density field B(r)

B(r) = curl A(r) = curl (A(r) + Ao) := curl A*(r) , (5-19)

since here too, a spatially independent component Ao vanishes

during the curl operation.

In addition to Ao, another term does not appear, and that is the
source field component Ao(r). Its absence is due to the fact that
the curl operator applied to the source field An(r) = grad OCr)
(or to ,the gradient of any scalar potential function) will always
yield zero,

curl grad 0 (r) = curl An(r) = 0 • (5-20)

5.3 Magnetic Vector Potentials 97

Hence.

B(r) = curl A(r) = curl (A(r) + grad O(r» := curl AU (r)

' - -_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _- - - 1 . (5-21)

In general. the vector-potentlal function of a magnetic field may

indeed be composed of a vortex field Av(r) and a source field
As(r) = An(r) = grad Q(r). Assuming that the magnetlc vector
potentlal A(r) of a magnetic field is a pure vortex field as in Fig-
ure 5.1. we can specify div A = O. In other words. a vector field
An(r) with source character does not exist. and grad O(r) = 0
(Coulomb gauge). When we will be discussing time-varying mag-
netic fields we will reconsider this choice (see Chapter 6).

Finally. we will discuss the evaluation of the magnetic vector

potential from a given current distribution. As in the electro-
static field where the potential function of a line charge was
calculated by first evaluating the contribution of an infmitesimal
line element dL or its point charge dQ (see 4.3); we start with
the contribution of an infinitesimal current filament dI to find
the magnetic field of a current-carrying conductor loop. Figure
5.2.

Figure 5.2: Calculatlon of the magnetic vector potentlal of a current

loop. Integratlon variable roo
98 5 Potential and Potential Function or Magnetostatic Fields

The vector potential of the conductor element dL at the field

point P is calculated from

dA(r) = !loI dL (5-22)

41tlr-rol

Here, one is tempted to establish an analogy between the prod-

ucts IdL and PLdL . However, the comparison is not perfect.
While an infinitesimal charge dQ may exist, an independent
infinitesimal current element will not, because currents exist
only in closed circuits (div J = 0). Hence, the equation above
makes sense only upon integration about a closed integration
path

Ar.,(r) =,b !lOI dL

:r 41tlr-rol
(5-23)

Now the analogy with the scalar potential of a line charge be-
comes more reasonable

cp
(r) -
-
f 41tEPLIr-ro
dL
I
(5-24)

For currents distributed over a surface with surface-current

density K(ro) one defines infinitesimal surface-current products
K(ro)dS, and for spatial current distributions with current den-
sity J(ro) we have infinitesimal volume-current products
J(ro)dV, where the element length dL is impliCitly included in
dS and dV,
5.3 Magnetic Vector Potentials 99

As(r) = flloK(ro)dS
41tlr-rol
S
(5-25)

Av(r) = flloJ(ro)dV
41t Ir-rol
V
(5-26)

It should be noted that in the line integral. the vector quality of

the integrand is expressed by the line element dL because the
current is by nature a flux. 1.e. a scalar.

In a Cartesian coordinate system the three component vectors

of the magnetic vector potential are given by

A -
;x. -
f lloJx(ro) dV
41tlr-rol
V

Ay = f lloJy(ro) dV
41tlr-rol
V

Az = f llo1z(ro) dV
41t Ir-rol
V
(5-27)

Hence. three integrals need to be evaluated.

100 5 Potential and Potential Function or Magnetostatic Fields

5.4 Potential Equation of the Magnetic

Vector Potential

In addition to calculating the vector potential from a given cur-

rent or current-density distribution, the vector-potential func-
tion A(r) can also be understood as a solution to a vector poten-
tial-equation.

The vortex density of a magnetic vortex-field is given by (see

3.2.2)

curlH=J. (5-28)

Substituting for H the magnetic vector potential via

1
B= ~ = curl A ~ H =-curl A, (5-29)
Jl

yields
curl curl A = J.LJ • (5-30)

In order to avoid obscuring the derivation of the vector poten-

tial-equation, the reader must accept without evidence that the
repeated application of the differential operator curl to an arbi-
trary vector field is equivalent to the difference of a cascaded
div and grad operation and the Laplacian operated on the same
vector (see A3). Thus

curl curl X = grad div X - V2 X . (5-31)

With this in mind, the above equation can be modified to