On the Dependence of Charge Density on

Surface Curvature of an Isolated Conductor

arXiv:1509.09252v2 [physics.class-ph] 17 Feb 2016

Kolahal Bhattacharya
[email protected]
Tata Institute of Fundamental Research

Abstract
A study of the relation between the electrostatic charge density at
a point on a conducting surface and the curvature of the surface (at
that point) is presented. Two major papers in the scientific litera-
ture on this topic are reviewed and the apparent discrepancy between
them is resolved. Hence, a step is taken towards obtaining a general
analytic formula for relating the charge density with surface curva-
ture of conductors. The merit of this formula and its limitations are
discussed.

PACS code: 41.20.Cv, 84.37.+q

1 Introduction
It is an observational fact that on a conducting surface, electric charge density
is greater at those points where local surface curvature is also higher. For
example, if some amount of charge is given to a needle, the charge density
would be very high at the sharp tip of the needle. The surface curvature
is also very high there, compared to the smoother cylindrical portions of
the needle. Although this is empirically known for a long time, the analytic
formula connecting the two quantities is not known. A major development in
this field may have a huge impact on modern experimental research areas, like
scanning tunnelling microscopy, field emission cathode, field ion microscopy
etc.
In general, the charge density at a point on a given conductor surface de-
pends on the curvature of the surface at that point, the shape of the conductor
and also on the presence or absence of the other nearby charged bodies. It
is proportional to the magnitude of the electrostatic field just outside the

1
conductor [1]. The field can be obtained from the electrostatic potential that
is a solution to Laplace’s equation, subjected to the boundary conditions [2].
However, the second order partial differential equation cannot be solved in
general for an arbitrary configuration of charged bodies. Even for isolated
charged conducting bodies, closed-form solutions are possible usually when
the problems exhibit planar/cylindrical or spherical symmetries. However, if
it is possible to solve, the charge density at any point on the conductor can
be easily calculated.
But this does not explicitly give the dependence of the charge density on
the surface curvature. To obtain the desired dependence, one can start from
the relation between the conductor surface curvature and the electric field
near the surface. The corresponding equation was first derived by Green [3],
who showed that
 
dE 1 1
= −2κE = −E + (1)
dn R1 R2
where dEdn
is the normal derivative of the electric field across the surface of
the conductor and κ = 12 ( R11 + R12 ) is the local mean curvature (R1 and R2
are the principal radii of curvature of the surface) at a given location. Sir
J J Thomson gave recognition to this relation in 1891 [4]. Various meth-
ods were used by different authors [5], [6], over the past century, to prove
Eq.(1). But its application to find a connection between charge density and
conductor surface curvature was first performed by Luo Enze [7]. He found
the expression for charge density as a function of the mean curvature of the
surface. But that was not the end of the story. Liu [8] (1987) showed that
for conducting bodies of specific shapes, the charge density at a point on the
surface was proportional to 0.25th power to the Gaussian curvature of the
surface at that point. McAllister verified Liu’s observation [9] and argued
that the result was correct only when the potential was a function of a single
variable. Unaware of this, in his 1987 paper, Liu made the statement that
‘Results from conductors with surfaces of different shape are so consistent
that it is natural for us to expect that the quantitative relation is a universal
rule for conductors whose surfaces can be expressed by analytic functions’.
Soon, this conjecture was proved to be incorrect by many authors [10], [11].
Whereas the works by Luo and Liu enriched this field very much, there
are some issues that need further discussions. Their works on the same prob-
lem led to apparently different results. Luo’s expression for charge density
as a function of the mean curvature bears no resemblance to Liu’s result,
which is in terms of the Gaussian curvature. Zhang criticised Luo’s work by
pointing to an incorrect assumption used for solving an integral [12]. On the

2
other hand, Liu’s formula cannot be true for surface with negative Gaussian
curvature, as that will lead to imaginary value of the charge density. This is
also not true for surfaces that have points with zero Gaussian curvatures. If
it were so, then electric charge could not accumulate on any plane conducting
surface. Therefore, there is enough scope of discussion of the results, known
so far, for better understanding of the topic.
In this article, the main results of the existing literature will be reviewed
first in the next section 2. Then, in section 3, the connection between their
works will be studied. In section 4, attempts will be made to obtain a general
analytic formula to connect the charge density and the surface curvature of
conductors. Hence, this formula will be used in section 5 to solve a standard
boundary value problem in electrostatics, to prove that the formulation may
have some practical applications. Finally, the merits and the demerits of the
derived formula will be discussed in section 6.

2 Previous Attempts
2.1 Luo Enze’s work
Luo derived a relation between the charge density σ(r) on a conductor and
the local mean curvature κM (r) based on Thomson’s equation Eq.(1). He
integrated this equation along a contour coincident with an electric field line
(which is also the direction of mean curvature vector of the local equipoten-
tial surface), emerging from any point on the conducting surface. Of course,
the mean curvature of the equipotential surfaces does not remain constant as
we move a finite distance away from the conductor. However, within an in-
finitesimal distance from the given conducting surface, it may be taken as not
to be varying too rapidly. Hence, Luo calculated the electric field very near
the surface of the conductor [13], assuming that the mean curvature of the
evolving equipotential surface would not vary at all within this infinitesimal
distance. He showed that:

E = E0 exp(−2κM n) (2)
-where E0 is the field at the conductor surface. This relation becomes more
accurate in the limit n → 0. E0 is found from the potential difference ∆V
over a small length ∆n along the field line. Luo determined its value to be:
 
2κM ∆V
E0 = (3)
exp(−2κM ∆n) − 1 ∆n→0

3
Eq.(3) was used to obtain the charge density as a function of mean curvature:
 
2κM ∆V
σ = E0 =
exp(−2κM ∆n) − 1 ∆n→0
2κM ∆V
≈ (4)
exp(−2κM ∆n) − 1

Luo reported the last equation as the desired charge density - curvature
relation. Although his results are based on few approximations, they satisfy
the experimentally observed facts to good accuracy [7]. However, the relation
is not very practical as it depends on the accurate choice of ∆n and ∆V .
Zhang [12] argued that Luo’s calculation was flawed as he had taken
constant value of mean curvature. The assumption was indeed incorrect.
However, the present author finds weakness in the illustration which Zhang
used to argue the same in his comment letter. If r is taken to be very near
the conducting surface r0 , then from Eq.(3) of the letter:

−2
r02 r02

n
E = E0 2 = E0 = E0 1 +
r (r0 + n)2 r0
 
1
≈ E0 1 − 2 n
r0
≡ E0 (1 − 2κM n) (5)
r2
Zhang argued that Luo’s calculation was flawed, as E0 r02 is different from
E0 exp(−2κM n). However, one can easily see that in the limit n → 0, they
are not different - up to the first order, as ex = 1 + x + ...

2.2 Liu and McAllister

Liu reported (1987) that if the conductor is an ellipsoid or a hyperboloid of
revolution of two sheets or an elliptic paraboloid of revolution, then σ(r) ∝
κG (r)0.25 holds on each of these bodies [8] [where κG denotes the local Gaus-
sian curvature]. The result was confirmed by McAllister in 1990, who dis-
cerned that in these special cases, Laplace’s equation becomes simply sepa-
rable and the potential becomes a function of a single variable [9].
Liu parametrized Laplace’s equation in terms of a function f such that
f (x, y, z) = k characterises an equipotential surface. One may assume that
the potential Φ = Φ(f (x, y, z)) where Φ is a solution to Laplace’s equation
∇2 Φ = 0. With this assumption, one can expand Laplace’s equation to
obtain [14]:

4
 d dΦ
( )
df df ∇2 f
dΦ
=− (6)
df
(∇f )2
Eq.(6) serves as the starting point of Liu’s work and may be referred to as a
key relation. He integrated Eq.(6) and obtained:
Z R ∇2 f
− df
Φ = A e (∇f )2 df + B (7)

Hence, he calculated particular solution Φ(r) for each of the conducting bod-
ies that he chose to examine. Then, he obtained the electric field by calculat-
ing the gradient of Φ and deduced the charge density σ at the surface of each
of those conducting bodies. Also, the Gaussian curvatures of the surfaces of
these bodies were calculated by him. Then, he showed that for each of these
bodies, σ(r0 ) ∝ κG (r0 )0.25 .
McAllister worked with a general orthogonal treatment of Laplace’s equa-
tion. He showed that when the potential is a function of a single variable, it
is possible to show that the electric field, and therefore σ(r0 ), is proportional
to √1gii , where gii is the square of the scale factor hi of the ith coordinate
of the general orthogonal coordinate system. Now, in the case of the sur-
faces examined by Liu, i.e. ellipsoid, hyperboloid of two sheets and elliptic
paraboloid, gii ∝ √1κ . Thus, the overall effect is that the charge density
G
σ(r0 ) ∝ κG (r0 )0.25 for these surfaces.

3 Connection between the existing literature

3.1 Liu’s key relation from Luo’s Approach
The starting points and approaches are apparently very different between
the works of Luo and Liu, though both of them intend to reach the same
goal: finding the dependence of charge density on the local surface curvature
for an isolated conductor. Luo gives σ as a function of mean curvature κM
whereas Liu gives σ in terms of Gaussian curvature κG . There should be a
connection between these two approaches. To explore that, we parametrize
Thomson’s theorem Eq.(1) in terms of the (equipotential) surface function
f (r) and expand the L.H.S. as below:

5
    
dE ∂  ∂Φ 
  · dr  ∂Φ 
+ 2κE = + 2κ  
ds ∂r  ∂r  ds  ∂r 
   
∂  ∂Φ ∂f   ∂Φ ∂f 
= n̂ · + 2κ  
∂r  ∂f ∂r  ∂f ∂r 
       
∂ ∂Φ  ∂f  ∂Φ ∂  ∂f  ∂Φ  ∂f 
= n̂ · + + 2κ
∂r ∂f  ∂r  ∂f ∂r  ∂r  ∂f  ∂r 
∂ 2 Φ ∂f  ∂f  ∂Φ
     
 ∂f  ∂Φ  ∂f 
= n̂ · 2
+ n · ∇  +
  (∇ · n̂)  
∂f ∂r ∂r   ∂f ∂r ∂f ∂r
2
d Φ ∂f ∂f dΦ ∂f
= 2( · )+ (∇ · )
df ∂r ∂r df ∂r
d2 Φ ∂f dΦ ∂ 2 f
= 2 ( )2 + ( )
df ∂r df ∂r2
=0

where we have used ∂f ∂r

= | ∂f
∂r
|n. After some rearrangements, we obtain
Eq.(6). Liu found it by parametrizing Laplace’s equation in terms of equipo-
tential surface function f (r). Thus, we reached starting point of Liu’s work
starting from that of Luo’s work (i.e. Thomson’s theorem). Both of them
employ Laplace’s equation at the fundamental level; because, Eq.(1) is a
consequence of Laplace’s equation [15].

3.2 Luo’s formula from Liu’s key relation

Integrating Eq.(6) between conducting surface equipotential f0 and another
arbitrary equipotential f (r), we get:
" Z #
f (r)
∇2 f
 
dΦ dΦ
(f (r)) = exp − df (8)
df df f0 f0 |∇f |2
Eq.(8) may be integrated w.r.t. f to obtain Eq.(7). Expanding Eq.(8) fur-
ther, we get:

r
∇2 f
   Z 
dΦ dΦ
(f (r)) = exp − (|∇f |n · dr)
df df f0 r0 |∇f |2
r
∇2 f
   Z 
dΦ
= exp − n · dr (9)
df f0 r0 |∇f |
∇2 f
Now, the factor |∇f |
equals:

6
 ∇2 f 1
= [∇ · (|∇f |n)]
|∇f | |∇f |
1
= [∇|∇f | · n + |∇f |∇ · n]
|∇f |
∇|∇f | · n
= + 2κM (as ∇ · n = 2κM ) (10)
|∇f |

Using Eq.(10), the integral in the argument of Eq.(8) becomes:

r Z r
∇2 f

∇|∇f |
Z
n · dr = 2κM + n · n · dr
r0 |∇f | r0 |∇f |
Z r Z r
= 2κM n · dr + n · ∇(ln|∇f |) n · dr
r0 r0
Z r Z r
∂
= 2κM n · dr + (ln|∇f |) dn
r0 r0 ∂n
Z r Z r
= 2κM n · dr + d(ln|∇f |)
r0 r0
Z r  
|∇f (r)|
= 2κM n · dr + ln (11)
r0 |∇f (r0 )|

With Eq.(11), we can reduce Eq.(9) to:

   Z r  
dΦ dΦ |∇f (r)|
= exp − 2κM n · dr − ln
df dff0 r0 |∇f (r0 )|
   Z r
|∇f (r0 )| dΦ
= exp[−2 κM n · dr] (12)
|∇f (r)| df f0 r0

dΦ(f )
Now, as ∇Φ(r) = df
∇f (r),
from Eq.(12), it follows that:
Z r
|∇Φ(r)| = |∇Φ(r0 )| exp[−2 κM n · dr] (13)
r0

Eq.(13) is the general version of Eq.(2). The required charge density is

proportional to E0 ≡ |∇Φ(r0 )| and can be found by integrating Eq.(13):
Φ0
|∇Φ(r0 )| = R r0 [−2
Rr
κM n·dr]
(14)
∞
e r0
n · dr

7
In Eq. (14), it is assumed that the reference point where potential Φ = 0, is
at infinity. This need not be the case. In some boundary value problem, if a
point r1 is given, where Φ = 0, the above equation can then be written as:
Φ0
|∇Φ(r0 )| = R r0 [−2
Rr
κM n·dr]
(15)
r1
e r0
n · dr

We shall make use of Eq. (15) in section 5, where the concept will be ap-
plied to solve a boundary value problem. In rest of the discussions, we shall
continue to refer to Eq. (14).

4 Towards a general relation

The integral in the denominator of Eq.(14) cannot be evaluated easily, as
the mean curvature κM varies across the continuum of equipotentials and
the functional form of this variation is unknown. However, the problem
can be approached from another way. Let us invoke the Gauss’s law in
electrostatics which asserts that: in a charge free region, the flux of the
electrostatic field EδA ≡ |∇Φ|δA is conserved along a flux tube (a bundle
of non-intersecting field lines perpendicular to the equipotentials). Thus,
between two equipotentials at r0 and r:

|∇Φ(r)|δA(r) = |∇Φ(r0 )|δA(r0 ) (16)

Hence, from Eq.(13) and Eq.(16), we have:
Z r
δA(r0 ) = δA(r) exp[−2 κM n · dr] (17)
r0

Therefore, using Eq.(17), Eq.(14) reduces to:

Φ0
|∇Φ(r0 )| = R ro h i (18)
n·dr
δA(ro ) ∞ δA(r)

The dependence of the charge density on surface parameters must be obtained

by studying Eq.(18). We notice that the charge density may not be dependent
only on the surface curvature factor(s). It can happen that apart from some
curvature factors, the charge density (or |∇Φ(r0 )|) also depends on some
other functions of the surface coordinates. Therefore, the problem has more
than one aspects. First, how is the charge density related to the curvature of
the surface and second, does this relation completely specify the dependence

8
of the charge density on the surface coordinates? In any case, how does it
connect to the observation made by Liu and McAllister?
To answer these questions, the dependence of δA(ro ) on the curvature of
1
the conducting surface f (r0 ) needs to be studied and δA(r) must be integrated
over a continuum of equipotentials, along an electric field line, starting from
a point r0 on the conducting surface and reaching up to infinity. Evidently,
obtaining a formula for the charge density effectively means having a method
to calculate the electric field when some amount of charge is given to a
conductor of any shape. There is no doubt that it is an extremely difficult
task.
The area element can be expressed in terms of the coefficients (E, G, F )
of the first fundamental form [16, p. 201] of the conducting surface as:
√
δA(r0 ) = EG − F 2 δuδv (19)
-when the surface is parametrized in terms of (u, v). By invoking the Brioschi
formula [17, section 1.5.2], we find:

1 1
∝√
δA(r0 ) EG − F 2
 0.25
 
 
 κG 
=
 − 1 E + F − 1 G 1
  
  2 vv uv 2 uu
E
2 u
Fu − 12 Ev   0 1
E
2 v
1
G 
2 u 
1  −  1 Ev

 F v − 2
Gu E F   12 E F  
1
 G
2 v
F G   Gu
2
F G 
(20)

where the subscripts u, v denote derivatives with respect to the corresponding

parameters. Hence, using Eq.(18) and Eq.(20), we find that charge density
σ(r0 ) is proportional to:

 0.25
 
 
Φ0  κG 
R r0 n·dr  − 1 E + F − 1 G 1
   
Fu − 12 Ev   0 1 1
h i
vv uv 2 uu
E
2 u
E
2 v
Gu

∞ δA(r)   2 2
 
1  −  1 Ev
 Fv − 2 Gu E F   12 E F  
1
 G
2 v
F G   Gu
2
F G 
(21)

9
Clearly, the dependence of σ(r0 ) on the curvature of the surface comes
through the factor κ0.25G in the numerator of Eq.(21).R rApart from this depen-
0
dence, σ(r0 ) also depends on (a) the integral factor ∞ and (b) the difference
of the determinants in the denominator of Eq.(20). All three factors con-
tribute to the charge distribution on a conducting surface. The dependence
on κ0.25
G is generic to all surfaces. The dependence on the difference of the
determinants in the denominator of Eq.(20) is also generic to all surfaces.
However, for different surfaces of different shapes, this factor assumes dif-
ferent values. If the equation of the conducting surface f (r(u, v)) specified
in the problem
R r0 is known, this factor can be easily calculated. The integral
factor: ∞ concerns the equipotentials of the problem. The contributions
of these surfaces to the integral vary, depending on the presence or absence
of other charged bodies nearby. Even if an isolated conductor is considered,
the calculation of the integral seems impossible, because the variation of el-
ementary area δA(r) across the equipotentials (from r0 to ∞) is not known.
Away from the conductor surface the shape of the equipotential is modified
significantly. Although the function of the equipotential surface still obeys
Eq.(6), the variation of δA(r) remains unknown. Unless the symmetry of
the problem allows the equipotentials to be parallel to each other (as in the
case of planar, cylindrical or spherical symmetries), the exact evaluation of
the integral
R r0 seems impossible. However, even if we keep aside the integral
factor ∞ from discussion for the time being, certain interesting aspects of
the problem can be addressed from the dependence of the charge density on
δA(r0 ) which led to Eq.(20).
0.25
First of all, the dependence of the charge density on κG appeared with-
out the assumption that the potential Φ(r) must be a function of a single
variable. However, the density does not only depend on the κG , but also
depends on the function of the surface coordinates,
R r0 through the factor in the
denominator of Eq.(20) and on the integral ∞ . As McAllister illustrated
with confocal quadrics and confocal paraboloids, the dependency on all these
factors reduces to only κ0.25
G , if the potential Φ(r) (and therefore, the equipo-
tential surface f (r)) becomes a function of a single variable. In all the other
cases, σ continues to depend on all three factors. When κG is negative or
zero, the denominator also assumes negative or zero value respectively, such
that the overall factor within the big square bracket (in Eq.(21)) remains
positive and the charge density remains real. A good example of this is pro-
vided by a toroidal conducting surface. If we assume that the radius from
the centre of the hole to the centre of the torus tube is R2 , and the radius
of the tube is R1 (such that (R2 > R1 ) ), then in the Cartesian coordinates,
the equation of a torus azimuthally symmetric about the z axis is given by:

10
 p
(R2 − x2 + y 2 )2 + z 2 = R12 (22)
For calculating the factor in the denominator of Eq.(20) (call it: DDD-
Difference of the Determinants in the Denominator), any standard parametriza-
tion of the surface may be used. In this example, we use the following
parametrization:

x = (R2 + R1 cos v) cos u

y = (R2 + R1 cos v) sin u
z = R1 sin v (23)

-where u, v ∈ [0, 2π). This particular surface is interesting, because its Gaus-
sian curvature
cos v
κG = (24)
R1 (R2 + R1 cos v)
is positive for 0 ≤ v < π2 and 3π
2
≤ v < 2π, zero for v = π2 , 3 π2 and negative for
π
2
≤ v < 3π 2
. The factor DDD was evaluated using Wolfram Mathematica-
κG

0.25
10 [18] using R2 = 5 and R1 = 4 (in arbitrary units). The plot for DDD
is shown in the following figure 1 for v ∈ [0, 2π]:
κG 0.25

DDD

0.25

0.20

0.15

0.10

0.05

v
1 2 3 4 5 6

κG

0.25
Figure 1: The plot for DDD
for torus given by Eq.(23)

κG

0.25
The above plot shows that DDD always remains positive irrespective
of the sign or value of κG of a surface. The charge density on the torus [11]
is a more complicated function of v and is different from the variation shown
in figure 1. Thus, the Compensating Factor (CF) that must be multiplied
κG

0.25
to DDD to obtain the charge density, arises due to the integral factor

11
R r0
∞
. The value of this factor can be back-calculated if the charge density is
κG

0.25
already known, using DDD . In the following table 1, the summary of
all the relevant factors is presented for a few conducting surfaces.

Table 1: Analysis of charge density dependence on various factors

κ

0.25
Surface Parametrization DDD DDD
CF
x = a cos u sin v
ellipsoid κ0.25
x 2 y 2
z 2 y = b sin u sin v ∝ sin14 v sin v
sin v
a2
+ b 2 + c2 = 1
z = c cos v
x = a sinh u cos v
hyperboloid of 2 sheets κ0.25
x2 y2 z2 y = b sinh u sin v ∝ sinh14 u sinh u
sinh u
a2
+ b2 − c2 = −1
z = c cosh u
√
x = a√u cos v
elliptic paraboloid
x2 y2 y = b u sin v constant κ0.25 none
a2
+ b2
= z
z=u

5 Application to a boundary value problem

After the discussion in the preceding section and from Eq. (15), it is evident
that if in a given boundary value problem, the point r1 where potential Φ = 0
is given, then the charge density will be proportional to:
Φ0
R ro h i (25)
n·dr
δA(ro ) r1 δA(r)

Eq.h (25)iis the same as Eq. (18) except for the lower limit of the integral
n·dr
of δA(r) . Next, we shall apply Eq. (25) to a boundary value problem and
validate the power of this formulation.
We take a classic example of an infinite conducting cone at potential V0
with its apex facing down on a grounded conducting plane [19]. The cone
has half-angle α and is open upwards (see figure 2). The apex of the cone is
separated from the conducting plane by a tiny air gap (to provide insulation).
We try to find the functional form of the charge density on the cone.
The standard method of solving this boundary value problem is to solve
the zenith angle dependent part of the Laplace’s equation. This gives the
electrostatic potential as a function of θ, i.e. Φ = Φ(θ). The electric field
is the negative gradient of this potential. The value of the field close to the
given conical conducting surface is proportional to the desired charge density.

12
 Figure 2: Conducting cone just above a grounded conducting plane

However, our formulation directly gives the functional dependence of the

electric field, near the conical surface. In this case, the equipotentials are all
conical surfaces within the region α < θ < π2 . As κG = 0 on the conical sur-
face, we calculate δA(r0 ) from Eq. (19). The conical surface is parametrized
as:

r
x= cos φ
a
r
y = sin φ
a
z=r (26)
√
Then, we can easily show that EG − F 2 ∝ r. So, at a point r0 [= (R, α, φ0 )]
on the surface of the cone, we have:

δA(r0 ) ∝ R dr R sin αdφ

= R2 sin αdrdφ (27)

If we follow a field line that originates from a patch around r0 on the given
conducting conical surface, we see that it reaches the grounded conducting
plane at a point r1 [= (R, θ = π2 , φ0 )]. We notice that only a single variable
(θ) varies in this problem. Now, the integral can be evaluated as:

13
 ro   Z ro  
n · dr
Z
R dθ
=
r1 δA(r) r1 R2 sin θ dr dφ
Z θ=α
1 dθ
=
R dr dφ θ= π2 sin θ
1 α
= tan (28)
R dr dφ 2

Therefore, from Eq. (25), Eq. (27) and Eq. (28), the charge density on the
conical surface is proportional to:
V0
σ(r0 ) ∝ (29)
R sin α ln(tan α2 )
Similarly, one can also derive the charge density on the grounded conducting
V0
plane and its value is found to be proportional to R ln(tan α
)
at a distance R
2
from the origin O. It is also possible to directly calculate the electric field
in the region α < θ < π2 using Eq. (16). As stated earlier, this formulation
is more direct in calculating charge density, compared to the conventional
methods.

6 Discussions
In this paper, we reviewed the literature on charge density dependence on
conductor surface curvature and tried to establish that these works, though
very different in their approaches, are actually related to each other. A
number of authors indicated the weak points in these works but the literature
were overall silent about how to generalise these results. The connection
between the work of Luo [7] and Liu [8] was also not well understood. A
recent pedagogical article [15] also could not throw light on this topic.
In this paper, we clarified that the charge density partially depends
on 0.25th power of the Gaussian curvature of the given conducting surface
(through Eq.(21)), though the mean curvature κM of the equipotentials ap-
pears in the integral (from r0 to r) in the expression of electric field near
the conducting surface in Eq.(14). It is an important point, because the κG
is an intrinsic property of a surface whereas κM is not. However, our work
indicates that charge density also depends on other functions dependent on
surface coordinates. Thus, the dependence on κG cannot be used always to
know the distribution of charge on a given conducting surface. In some cases

14
like elliptic paraboloid, it may happen that the contributions from DDD and
CF are constants and σ ∝ κ0.25 G - but that does not happen in general.
We have shown that R r0when the reference point of the potential is at a finite
n·dr
distance, the integral r1 δA(r) can be evaluated if certain symmetries permit.
In this case, the formulation offers a direct way to find the functional depen-
dence of the density of charge.
Rr However,
Rr if the symmetries do not permit,
evaluation of the integral r10 (or ∞0 ) seems to be an impossible task, even
when the potential depends on a single variable, because the area element
δA(r), n(r) and dr gradually evolve along with the different equipotentials.
The contribution of this factor into the charge density is quantified by the
compensating factor CF. For the examples in table 1, it appears that CF
somehow depends on dz dξ
, where z is a function of ξ (in the parametric equa-
tion of the surface). But this empirical observation should not be interpreted
as a general rule.

7 Acknowledgements
I feel indebted to Dr. Debapriyo Syam for precious encouragements from
him while trying to solve the problem. The supports from from my friends
Tanmay, Manoneeta and Tamali were also of great help.

References
[1] David Jeffrey Griffiths and Reed College. Introduction to electrodynam-
ics, volume 3. prentice Hall Upper Saddle River, NJ, 1999.

[2] Walter Greiner. Classical electrodynamics. Springer Science & Business

Media, 2012.

[3] George Green. An essay on the application of mathematical analysis to

the theories of electricity and magnetism, volume 3. author, 1889.

[4] Ezzat G Bakhoum. Proof of thomson’s theorem of electrostatics. Journal

of Electrostatics, 66(11):561–563, 2008.

[5] GA Estevez and LB Bhuiyan. Power series expansion solution to a clas-

sical problem in electrostatics. American Journal of Physics, 53(2):133–
134, 1985.

[6] Richard C Pappas. Differential-geometric solution of a problem in elec-

trostatics. SIAM Review, 28(2):225–227, 1986.

15
 [7] Luo Enze. The distribution function of surface charge density with
respect to surface curvature. Journal of Physics D: Applied Physics,
19(1):1–6, 1986.

[8] Kun-Mu Liu. Relation between charge density and curvature of sur-
face of charged conductor. American Journal of Physics, 55(9):849–852,
1987.

[9] IW McAllister. Conductor curvature and surface charge density. Journal

of physics. D, Applied physics, 23(3):359–362, 1990.

[10] Myriam Dubé, Mario Morel, N Gauthier, and AJ Barrett. Comment

on“relation between charge density and curvature of surface of charged
conductor,”by kun-mu liu [am. j. phys. 55, 849-852 (1987)]. American
Journal of Physics, 57:1047–1048, 1989.

[11] M Torres, JM González, and G Pastor. Comment on“relation between

charge density and curvature of surface of charged conductor,”by kun-
mu liu [am. j. phys. 55, 849-852 (1987)]. American Journal of Physics,
57:1044–1046, 1989.

[12] Yuan Zhong Zhang. A comment on’the distribution function of surface

charge density with respect to surface curvature’. Journal of Physics D:
Applied Physics, 21(7):1235, 1988.

[13] Luo Enze. The application of a surface charge density distribution func-
tion to the solution of boundary value problems. Journal of Physics D:
Applied Physics, 20(1):1609–1615, 1987.

[14] William Ralph Smythe and William R Smythe. Static and dynamic
electricity, volume 3. McGraw-Hill New York, 1950.

[15] Mehdi Jafari Matehkolaee and Ali Naderi Asrami. The review on the
charge distribution on the conductor surface. European J Of Physics
Education, 4(3), 2013.

[16] Theodore Frankel. The geometry of physics: an introduction. Cambridge

University Press, 2011.

[17] Shlomo Sternberg. Curvature in mathematics and physics. Courier Cor-

poration, 2012.

[18] Wolfram Mathematica. Wolfram research. Inc., Champaign, Illinois,

2012.

16
[19] Matthew NO Sadiku. Elements of electromagnetics. Oxford university
press, 2007.

17