Sunteți pe pagina 1din 6

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/236597622

Equilibrium charge density on a conducting


needle
Article in American Journal of Physics September 1997
Impact Factor: 0.96 DOI: 10.1119/1.18671

CITATIONS

READS

10

80

1 author:
Mark Andrews
Australian National University
29 PUBLICATIONS 347 CITATIONS
SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,


letting you access and read them immediately.

Available from: Mark Andrews


Retrieved on: 05 July 2016

Equilibrium charge density on a conducting needle


Mark Andrews
Department of Physics, The Faculties, Australian National University, ACT 0200, Australia

~Received 16 January 1997; accepted 9 March 1997!


The concept of an equilibrium continuous charge density on a conducting line, although in some
ways unphysical, does provide useful physical insight into needlelike conductors. It is shown that
the charge density on such a line must be uniform. A simple method is developed for calculating the
charge density on long, thin conductors with circular cross section. Also, the equilibrium
distribution of N equal charges on a line is shown to approach equal spacing ~except near the ends!
as the number of charges grows. 1997 American Association of Physics Teachers.

I. INTRODUCTION
In a recent paper1 Griffiths and Li sought the equilibrium
charge density for a charged conducting line of finite length.
Finding no direct solution to this problem, they considered
the charge density on some three-dimensional long, thin conducting surfaces ~spheroids and circular cylinders!. No firm
conclusion was reached; indeed, they suggested that the
problem may be ill-posed, the answer depending on the particular model used to represent the needle. The equilibrium
distribution of positions for N equal charges confined to a
finite line was also considered. Again, no definite conclusion
was reached about the limiting distribution as N becomes
very large.
It will be argued here that there is a simple solution to
both these problems. The only reasonable interpretation of
the equilibrium density on a line is as the limit of a long, thin
conducting surface as it is made infinitesimally thin. In this
limit, the density must be uniform. In particular, it will be
shown that, for any conducting surface of revolution ~revolved about the z axis! with radius r (z)5 t f (z), the density
projected onto any part of the axis that does not include
846

Am. J. Phys. 65 ~9!, September 1997

either end can be made as uniform as we please by making t


sufficiently small. Also, the distribution of N equal charges
on a line must approach uniformity as N!` ~in the sense
that the spacing on any part of the line that does not include
either end can be made as uniform as we please by making N
sufficiently large!.
At first sight it would appear that the equilibrium distribution on a line conductor cannot be uniform because a uniform distribution would give a nonzero electric field at any
point not at the center of the line. Obviously the contribution
to the field at such a point from the charge to one side of it,
will differ in magnitude from the opposite field that comes
from the other side.
On the other hand, it will now be shown that a continuous
nonuniform charge distribution cannot be in equilibrium. If
the line runs from z5a to z5b and the charge density is
l(z), then the field at point z is
E~ z !5

1
4pe0

l~ z !
sgn~ z2 z ! d z ,
~ z2 z ! 2

~1.1!

where sgn(z2z) stands for the sign of (z2 z ). This integral


1997 American Association of Physics Teachers

846

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.56.64.29 On: Thu, 06 Nov 2014 23:56:31

has a nonintegrable singularity, so the field is infinite ~due to


the charge in the neighborhood of z!. However, the contributions from either side of z are in opposite directions, so
even though they are infinite, perhaps they can cancel. To
investigate this, integrate by parts:

l~ b ! l~ a !
1
E~ z !5
2
2
4 p e 0 b2z z2a

l 8~ z !
dz .
u z2 z u

~1.2!

This shows that the force will be infinite unless l 8 (z)50


everywhere on the conductor, i.e., unless the charge distribution is uniform.
One might conclude that the concept of an equilibrium
charge density on a line is not useful. The fact that such an
object would have infinite energy is a further indication of its
unphysical nature ~but this applies to any line of charge!.
Nevertheless I believe that the concept is consistent and useful in that it provides insight into the properties of needle
conductors. The equilibrium charge distribution on a line
must be taken to be uniform; the fact that there is an extra
finite force from one side of any noncentral test point does
not matter because this finite force will be negligible compared with the infinite forces from the immediate neighborhood. @This point has also been made by R. H. Good.#2 For a
physical needle, the forces from the nearby region dominate
those from further away and this dominance increases as the
needle is made thinner @as t !0 for a surface of revolution
with radius r (z)5 t f (z)#. It cannot be concluded, however,
that any conductor with a sufficiently large ratio of length to
maximum radius will be approximately uniform; indeed a
counter-example will be given.
If this physical reasoning is valid, then the charge density
must be uniform, even if the conducting line is curved and
even if there are external fields or other conductors nearby.
This is because at any point the force from the nearby charge
infinitely dominates that from further away and therefore it is
only the uniformity of the line density that is important; any
external fields or curvature do not matter. Of course, the
conducting line must be smooth so that locally it is essentially straight.
A simple example of the independence of external fields is
a long, thin spheroid placed in a uniform electric field so that
the axis is in the same direction as the field. This can be
solved exactly, with or without net charge on the spheroid.
The charge density ~projected onto the axis! approaches uniformity as the radius gets small even if the field is present.
Another consequence of the fact that no polarization can occur on a conducting line is that such a conductor can have no
effect on the field. Thus the distance beyond which a thin,
uncharged spheroid no longer significantly affects the uniform field gets small with the radius. The details are given in
Sec. V.

II. SPHEROIDAL CONDUCTORS


A. Potential due to a finite line
Since I am asserting that the charge on a conducting
straight line is uniformly distributed, it is of interest to consider the potential for such a line, taken to lie between 2a
and 1a on the z axis. Integrating the contributions from
each point on the line gives the potential at any external
point as
847

Am. J. Phys., Vol. 65, No. 9, September 1997

V ~ z, r ! 5

l
4pe0

2a

Ar

1 ~ z2 z ! 2

dz.

~2.1!

This integral is elementary and the result is3


V5

l
ln@~ u1a ! / ~ u2a !# ,
4pe0

~2.2!

where u5 21(r 1 1r 2 ) and r 1 and r 2 are the distances from


the point to each end of the line ~see Fig. 1!. Since V is a
function of u alone, it is constant on prolate spheroids with
the ends of the line as foci. The length of the spheroid ~along
the axis! is 2u and its radius, i.e., its maximum distance
from the z axis, is r 0 5(u 2 2a 2 ) 1/2. The equation for the
spheroid can be written as ( r / r 0 ) 2 512(z/u) 2 . For r 0 !a,
V'(l/2p e 0 )ln(a/r0), which agrees with the usual potential
for an infinitely long charged line obtained using Gauss
theorem.
B. Conducting prolate spheroids
It follows that charged conducting prolate spheroids have
equipotential surfaces that are confocal spheroids. The field
lines starting at a point on the charged line form a surface ~in
this case an hyperbola! always orthogonal to the equipotential surfaces. By Gauss theorem, the charge on the axis between any two such field-line surfaces will equal the charge
between the same two surfaces on any of the spheroids taken
to be a conductor. Near the line charge, the surfaces must be
normal to the line and therefore, as r 0 tends to zero, the
charge density on the spheroid must approach that on the line
~except possibly near the end!. Griffiths and Li have shown
that in this spheroidal case, the density projected onto the
axis is uniform for all r 0 ~but, of course, it extends beyond
the foci!.
III. NEEDLES
One way to calculate the charge distribution on an isolated
conductor is to solve Laplaces equation in the space outside
the conductor with the boundary conditions that the potential
goes to zero at large distances and is constant on the surface.
For a needle ~a thin conducting surface of revolution! there is
a much easier way.
First, an approximate expression will be found for the potential on any thin axisymmetric charged surface ~not necessarily a conductor!. This expression can be easily inverted to
give the charge density in terms of the potential and then
applied to a conductor by requiring the potential on its surface to be constant.
A. The potential on a thin surface of revolution due
to a given charge distribution over the surface
The potential at a point on the surface can be calculated by
integrating the contributions from the whole surface: V
5(1/4p e 0 ) * dq/ u r2r8 u . The surface charge density s (z)
depends only on z and therefore can be expressed as a line
density l(z)52 pr (z) s (z).
For a spheroid the integration over the surface must give
the same result as integrating the corresponding uniform line
density on the axis ~between the foci!. For a surface which is
very thin, the contributions from far away ~compared with
the local radius! are obviously almost the same as from a line
charge with the same line density.
Mark Andrews

847

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.56.64.29 On: Thu, 06 Nov 2014 23:56:31

grow without bound for small r ~because the small denominator near z 5z is canceled by the numerator!; it is therefore
valid to neglect these terms for thin enough conductors, but
some could be included to give a better approximation.
C. The line charge density of a needle approaches a
constant as it becomes thinner

Fig. 2. Equipotentials of V as given in Eq. ~4.1! for a 51, b 50.5,


4 p e 0 V52, 3, 4. ~V increases as the charged line is approached.!

Now keep the shape the same but let the surface become
thinner. Put r (z)5 t f (z) and let t !0 while keeping the
same total charge. Then Eq. ~3.2! implies that l(z) becomes
more uniform because
ln@ 2 A~ a2z !~ a1z ! / r ~ z !#

The potential can be made as large as we please by making the spheroid thin enough, i.e., making the radius r 0 small
enough. But the contribution from the far region is independent of r 0 and therefore the proportion of the potential which
comes from a certain such nearby region can be made as
large as we please by making r 0 small enough.
It is now clear that for any charged surface of revolution
such that the line density l(z) does not change appreciably
over several radii, the potential at any point z, not near the
ends, will be approximately that of a spheroid of the same
length with uniform line density l(z) over its length and
radius r (z) at point z. ~The spheroid need not be of the same
length, but the approximation will usually be better if it is.!
Close to the charged line ( r !a), but not too near its ends
(a2 u z u @ r ), it is easy to see that u'a1 21r 2 a/ @ (a2z)(a
1z) # . Thus, except near the ends of a thin axisymmetric
charged surface,
V~ z !'

1
l ~ z ! ln@ 2 A~ a2z !~ a1z ! / r ~ z !# .
2pe0

~3.1!

B. The charge distribution on a needle


A good approximation is to invert Eq. ~3.1!, with constant
potential V:
l ~ z ! '2 p e 0 V/ln@ 2 A~ a2z !~ a1z ! / r ~ z !# .

~3.2!

The total charge Q and the potential V are, of course, proportional. Hence, if V is unknown but Q is known, one simply calculates Q5 * l(z)dz and scales V to make Q correct.
There is an alternative way to derive these results. The
potential due to the line charge density l(z) is
V ~ z, r ! 5

1
4pe0

l~ z !

2a

Ar

1 ~ z2 z !

dz.

~3.3!

5ln@ 2 A~ a2z !~ a1z ! / f ~ z !# 1ln@ 1/t #

and clearly the z dependent terms contribute less as t !0.


This does not mean that thin conductors must be uniformly charged; an example will be given in Sec. IV.
D. An example: A thin conducting cylinder
In this case, r (z) is constant and then Eq. ~3.2! gives a
simple expression for the charge density. Numerical results
have been published by Waterman and Pedersen4 for six values of r /a ranging from 1022 to 10240. The charge density
was found to fit well to @ 12(z/a) 2 # 2 g with g reasonably
well approximated by 21/ @ ln(4a/r)2 37]. It will now be shown
that this empirical expression for l(z) agrees, for small
enough r /a, with ~3.2!, which is equivalent to l(z)
g ln@12(z/a)2#) with
g 5 21/ln(2a/r). Now @ 1
5l 0 /(11
2 2g
2
2(z/a) # 51/exp(g ln@12(z/a) #), and this can be well approximated by 1/(11 g ln@12(z/a)2#) provided that g ln@1
2(z/a)2#!1, i.e., not too close to the ends. @The proportional
error will be less than e if u z u ,12 21( 21r ) 2 e .# The difference
g causes only a small error if r /a is small.
between g and
Graphing the results confirms that the error in ~3.2! is of the
same order as that resulting from the empirical form for g
used by Waterman and Pedersen, and that the error decreases
with the radius.
IV. LINEAR LINE CHARGE DENSITY
A. Line with linearly increasing charge density
It is not difficult to calculate the equipotential surfaces for
the case where l5 a 1 b z. At any external point z, r,
V ~ z, r ! 5

Expressing l~z! as a power series about z then leads to


l~ z !
V ~ z, r ! 5
4pe0

1
1
4pe0

2a

Ar 2 1 ~ z2 z ! 2

(
n51

l ~ n !~ z !
n!

dz

~ z 2z ! n

2a

Ar 2 1 ~ z2 z ! 2

dz,
~3.4!

(n)

where l (z) is the nth derivative of l~z! evaluated at z


5z. The first term is just the approximation used for ~3.1!; it
is very large if r is small ~because near z 5z the denominator
in the integrand becomes very small!. The other terms do not
848

Am. J. Phys., Vol. 65, No. 9, September 1997

~3.5!

1
4pe0

a1bz

2a

Ar

1 ~ z2 z ! 2

dz

1
u1a
1 b ~ r 1 2r 2 ! . ~4.1!
~ a 1 b z ! ln
4pe0
u2a

Holding V constant then gives an equipotential surface. The


family so generated tend to spheres far from the line and
become pear-shaped closer to it ~see Fig. 2!.
If V is very large, the last term will be negligible. Also, as
in ~3.1!, close to the charged line but not too near its ends,
V ~ z, r ! '

1
~ a 1 b z ! ln@ 4 r 22 ~ a2z !~ a1z !# ,
4pe0

~4.2!

so that for large constant V, r 2 54(a2z)(a1z)


exp( 2 4pe0V/(a 1 bz)). This makes these surfaces easy to
Mark Andrews

848

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.56.64.29 On: Thu, 06 Nov 2014 23:56:31

visualize: The equipotentials are obtained ~except near the


ends! by taking the spheroid of unit radius with ends at
6a, and decreasing the radius by the factor exp
(22pe0V/(a1bz)). It is clear that the central radius r 0
5exp(22pe0V/a) can be made as small as we please by
making V/ a large enough.
For any given r 0 it is possible to have any amount of
change in the charge density along the needle by making the
radius change by a sufficiently large factor along the needle.
This shows that it is not true that a thin needle ~shape unspecified, but central radius given a very small value! must
have essentially uniform line charge density. On the other
hand, the ratio by which the radius must change to produce a
given variation in charge density increases as r 0 gets
smaller; this is consistent with our earlier result that if the
shape r (z)/ r 0 is kept fixed, then the charge density must
become more uniform as r 0 decreases.
A numerical example will serve to show that, because of
the slow logarithmic dependence on r 0 of the scale of the
density changes, the atomic limit may constrain what can be
achieved. Suppose that r /a'231028 ~i.e., about 20 atoms
radius with a length of 10 cm!, then for a density change of
10% the radius would vary by an extra factor of about 12
~i.e., the radius would come down to two atoms near one
end!.

Fig. 3. Equipotentials for an uncharged conducting spheroid in a uniform


electric field. The radius of the spheroid is 0.1 and its foci are at z5
61. One quadrant of its section is shown as the bold curve. The field
strength in Eq. ~5.3! is taken to be E51. The potential V is zero on the plane
z50 and on the spheroid. The curves shown have V50.2, 0.4, 0.6, 0.8, 1.0,
1.2. The dashed curve shows the confocal spheroid where the distortion of
the potential is 10%. ~All the curves shown are sections of surfaces of
revolution about the z axis.!

in Sec. II. If the potential corresponding to a uniform field of


strength E in the z direction is added to the second term in
~5.1!, then the sum is

V. NEEDLE IN AN ELECTRIC FIELD


A. Uncharged line conductor in an electric field
When an uncharged conductor is placed in a field, the
charge carriers move in response to the field, thus creating a
nonzero charge distribution over the conductor. For a line
conductor a nonuniform distribution is not possible and the
only uniform charge distribution with no net charge is zero
everywhere. Since no charge can move, the field has no effect on the conductor. For a thin conductor we would therefore expect the effect of the conductor on the field to be less
the thinner it is.
As an example, consider a conducting prolate spheroid
inserted into a uniform field with the axis of the spheroid
parallel to the field. This is a case which can be solved
exactly5 by separation in spheroidal coordinates, but in essence the solution has already been found in Sec. IV.

V52Ez @ 12 ~ b /2p e 0 E !@ Q ~ u/a ! 2a/u ## .

~5.3!

This potential ~a! satisfies Laplaces equation, ~b! is constant


on the spheroid u5u 0 , where
~ b /2p e 0 E !@ Q ~ u 0 /a ! 2a/u 0 # 51,

~5.4!

and ~c! at large distances becomes 2Ez @ 12( b /2p e 0 E)


3(a/u) 3 # . It is therefore the potential corresponding to the
uncharged spheroid u5u 0 placed in the uniform field ~see
Fig. 3!.
The potential for the spheroid with charge q placed in the
same uniform field will then be obtained by adding
(q/4p e 0 a)Q(u/a) to ~5.3!.
C. Thin spheroidal conductor in an electric field

B. Spheroidal conductor in an electric field


In Sec. IV the potential due to a line charge density
l(z)5 a 1 b z between z52a and z5a was given in ~4.1!.
It is easy to show from the triangle in Fig. 1 that r 22 2r 21
54az. Therefore, r 2 2r 1 52az/u and the potential in ~4.1!
can be written as
V5

1
@ 2 a Q ~ u/a ! 12 b z @ Q ~ u/a ! 2a/u ## ,
4pe0

~5.1!

where
Q ~ x ! 5 21 ln@~ x11 ! / ~ x21 !# .

~5.2!

Both the term in a and the term in b must satisfy


Laplaces equation. The first, Q(u/a), is constant on spheroids; it is the potential due to a charged spheroid considered
849

Am. J. Phys., Vol. 65, No. 9, September 1997

Here, we require u 0 to approach a, which implies that the


radius r 0 5(u 20 2 a 2 ) 1/2!0. Then Q(u/a)!ln(a/r0) and
~5.4! becomes b 52 p e 0 E/(ln(a/r0)21). This shows that, for
a given field strength E, b !0 as r 0 !0; i.e., the line charge
density approaches uniformity as the spheroid becomes thinner.
It will now be shown that the region of significant influence on the field does decrease as the radius of the thin
spheroid is decreased, but this happens very slowly. The proportional deviation of the potential from that due to the uniform field is seen from ~5.3! to be @ Q(u/a)2a/u # /
@ Q(u 0 /a)2a/u 0 # 'ln(a/r)/ln(a/r0). This is constant on each
confocal spheroid. A numerical example will show how extremely thin a conducting spheroid must be to have a thin
region of influence on the field. For a needle 10 cm long and
1029 m in radius ~about ten atoms! the distortion at radius
Mark Andrews

849

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.56.64.29 On: Thu, 06 Nov 2014 23:56:31

r 51 cm is log 10/log 108 '12%. Clearly, keeping the distortion within a region much less than about 1-cm radius is
not possible because of the atomic limit.

to first order in a and for large n. The total field at z will


then be
E ~ z ! '2

VI. N EQUAL CHARGES ON A FINITE LINE


The equilibrium distribution of positions for N equal
charges confined to a finite line, unlike a continuous charge
density on a finite line, presents no problems of principle;
there is clearly a unique answer for any N and this will have
spacing that decreases toward the ends. The interesting question raised by Griffiths and Li concerns the limiting distribution for large N. It will now be shown that the distribution on
any part of the line that does not include either end will
become more uniform as N!`. As in the continuous case,
this is essentially because the force on any charge is dominated by that from its near neighbors; and this dominance
increases with N.
If the spacing d in the near neighborhood of one of the
charges is constant, then the force from the neighbors on one
side will exactly balance the force from the same number of
neighbors on the other side. The field due to the nearest n
neighbors ~on one side! is (q/4p e 0 d 2 ) ( nk51 k 22 , where q is
the charge on each. If N is increased with the total charge
kept constant, then d }N 21 and q}N 21 and the field will
increase as N. But clearly the field from the rest of the line
will barely change with N if the overall distribution does not
change much. To achieve equilibrium the spacing must
change slowly to produce a slight imbalance in the very large
forces from the near neighbors to balance the ~relatively
small! unequal forces from the rest of the line. Because the
force from the near neighbors increases rapidly with N, the
rate at which the spacing changes with distance must decrease, i.e., the spacing must become more uniform as N
increases.
To be more precise about how the spacing changes with
N, suppose that the spacing changes linearly so that the position of charge k is z k 5z 0 1k d (12 a k d ). Thus d is the
spacing at z 0 and a can be related to l 8 (z 0 ), i.e., how fast
the density changes with z. The density at z k is q/ d k , where
q is the individual charge and d k 5 21(z k11 2z k21 )5 d (1
22 a k d ). Therefore, l(z k )5(q/ d )(112 a k d ) and l 8
52 a l.
The field at z 0 due to these 2n charges ~n on either side of
z 0 ! will be
E n5

q
4pe0

'2

(
k51

aq
p e 0d

1
1
22
~ z 0 2z 2k !
~ z 0 2z k ! 2
n

(
k51

1
1
'2
l 8 ln n
k
2pe0

~6.1!

1
1
l 8 ln n1
2pe0
4pe0
1
4pe0

l~ z !

~ z2 z ! 2
z1b

z2b

2a

l~ z !
dz
~ z2 z ! 2

dz,

~6.2!

'n d . This should be independent of n and b,


where b'b
and we can therefore expect that l 8 (z) varies as C/ln N for
large N. ~The independence of n and b will come about
through a cancellation of 2l 8 (z)ln b, arising from n
5Nb/a, with the same term arising from the integral. This
can be confirmed by integrating by parts.!
The slow logarithmic dependence of the density on N explains why Griffiths and Li saw little change for the range of
values they used.
A first approximation to the distribution well away from
the ends could be obtained by putting E(z)50 in Eq. ~6.2!
and l~z! constant inside the integrals. To do better than this
would probably require some solution to the problem of what
happens near the ends.

VII. CONCLUSION
The concept of an equilibrium continuous charge density
on a conducting line makes physical sense if it is taken to be
the limit of a needle conductor as the needle becomes infinitesimally thin, providing the limit is taken in an appropriate
way. In this sense, it does not depend on the initial shape of
the needle. The charge density on such a line must be uniform. The same limit is obtained for the equilibrium distribution of N equal charges on a line: the distribution will
become uniform ~except near the ends! as the number of
charges grows.

ACKNOWLEDGMENT
Thanks are due to David Griffiths for useful comments
and for drawing my attention to Ref. 4.
1

D. J. Griffiths and Y. Li, Charge density on a conducting needle, Am.


J. Phys. 64, 706714 ~1996!.
2
R. H. Good, Comment on Ref. 1, Am. J. Phys. 65, 155156 ~1997!.
3
L. Eyges, The Classical Electromagnetic Field ~Constable, London, 1972;
Dover, New York, 1980!, Sec. 4.4.
4
P. C. Waterman and J. C. Pedersen, Scattering by finite wires, J. Appl.
Phys. 72, 349359 ~1992!.
5
P. M. Morse and H. Feschbach, Methods of Theoretical Physics ~McGrawHill, New York, 1953!, Sec. 10.3.

A THEORY LOOKING FOR A QUESTION


My difficulty, then, is that I do not know what observations complex systems dynamics is
trying to explain. It is a theory looking for a question to answer.
John Maynard Smith, Life at the Edge of Chaos?, The New York Review of Books, 2 March 1995.

850

Am. J. Phys., Vol. 65, No. 9, September 1997

Mark Andrews

850

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.56.64.29 On: Thu, 06 Nov 2014 23:56:31

S-ar putea să vă placă și