calculation of the electrostatic field of a system of spherical segments
TRANSCRIPT
ISSN 1063-7842, Technical Physics, 2008, Vol. 53, No. 8, pp. 1086–1090. © Pleiades Publishing, Ltd., 2008.Original Russian Text © E.M. Vinogradova, N.V. Egorov, K.A. Krimskaya, 2008, published in Zhurnal Tekhnichesko
œ
Fiziki, 2008, Vol. 78, No. 8, pp. 128–131.
1086
FORMULATION OF THE PROBLEM
Let us consider the axisymmetric problem of deter-mining the electrostatic potential of a system consistingof an arbitrary number of nonconcentric spherical seg-ments located between two spheres.
This axisymmetric problem will be solved in abispherical system of coordinates. The relation betweenbispherical coordinates (
α
,
β
) with cylindrical coordi-nates (
r
,
z
) has the form [1]
Coordinate surfaces
β
= const define a family ofnonintersecting nonconcentric spheres, which aredescribed by the following equation:
Coordination surfaces
α
= const define a family ofspindle-shaped surfaces of revolution, which areorthogonal to surfaces
β
= const and are defined by theequation
The parameters of the problem are as follows:
N
isthe number of segments;
α
i
≤
α
≤
π
,
β
=
β
i
are the sur-
faces of segments (
i
= );
f
i
(
α
) are the first-kindboundary conditions on corresponding segments (
i
=
); 0
≤
α
≤
π
,
β
=
β
0
,
β
=
β
N
+ 1
are the surfaces ofthe spheres; and
f
0
(
α
),
f
N
+ 1
(
α
) are the first-kind bound-ary conditions on the spheres.
z ir+ icα iβ+
2---------------,cot=
0 α π, ∞ β ∞< <– .≤ ≤
r2 z c βcoth–( )2+c
βsinh--------------⎝ ⎠
⎛ ⎞2
.=
r c αcot–( )2 z2+c
αsin-----------⎝ ⎠
⎛ ⎞2
.=
1 N,
1 N,
Without loss of generality, we can set
β
i
<
β
k
for
i
<
k
.If product
β
0
β
N
+ 1
is zero, one of the spheres opens toform a plane
β
= 0. When
β
0
β
N
+ 1
> 0, the system of seg-
Calculation of the Electrostatic Field of a System of Spherical Segments
E. M. Vinogradova, N. V. Egorov, and K. A. Krimskaya
St. Petersburg State University, St. Petersburg, 198904 Russiae-mail: [email protected]
Received November 26, 2007
Abstract
—The electrostatic potential distribution is determined for a system of axisymmetric electrodes in theform of nonconcentric spherical segments.
PACS numbers: 41.20.Cv, 41.85.Ne
DOI:
10.1134/S1063784208080185
SHORTCOMMUNICATIONS
r
c
z
α
=
α
k
β
=
β
k
β
=
β
N
β
=
β
N
+ 1
β
=
β
0
> 0
… …
Fig. 1.
r
c
z
c
……
α
=
α
k
β
=
β
k
> 0
β
=
β
N
> 0
β
=
β
N + 1
> 0
β
=
β
1
< 0
β
=
β
0
< 0
Fig. 2.
TECHNICAL PHYSICS
Vol. 53
No. 8
2008
CALCULATION OF THE ELECTROSTATIC FIELD 1087
ments is within one of the closed spheres, but outsidethe other sphere (see Fig. 1). If
β
0
β
N
+ 1
< 0, the seg-ments lie outside the spheres (Fig. 2).
The distribution of electrostatic potential
U
(
α, β)satisfies the Laplace equation and the following bound-ary conditions:
(1)
SOLUTION OF THE PROBLEM
To solve boundary-value problem (1), we divide theentire region occupied by the electron-optical systeminto N + 1 subregions: βi ≤ β ≤ βi + 1. For each subregion,we can present the potential distribution U(α, β) =
Ui(α, β) (i = ) in the form of the following expan-sion in Legendre polynomials [1, 2]:
(2)
Coefficients A0, n, AN + 1, n can be calculated fromboundary conditions (1) on spheres β0, βN + 1:
(3)
The representation of potential in form (2) satisfiesthe Laplace equation and the potential continuity con-ditions at the interfaces between the regions. Theboundary conditions and the continuity conditions forthe derivative of the potential along the normal to theinterfaces between subregions lead to the system ofpaired equations
∆U α β,( ) 0=
U α βi,( ) αi α π≤ ≤ f i α( ), i 1 N,= =
U α β0,( ) f 0 α( )=
U α βN 1+,( ) f N 1+ α( ).=⎩⎪⎪⎨⎪⎪⎧
0 N,
Ui α β,( ) βcosh αcos–=
×n
12---+⎝ ⎠
⎛ ⎞ β βi–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βi 1+ βi–( )sinh
-----------------------------------------------------Ai n,
n 0=
∞
∑
+
n12---+⎝ ⎠
⎛ ⎞ βi 1+ β–( )sinh
n12---+⎝ ⎠
⎛ ⎞sinh βi 1+ βi–( )-------------------------------------------------------Ai 1+ n, Pn cos α,( ),
βi β βi 1+ , i< < 1 N, .=
Ak n,2n 1+
2---------------
f k α( )βk αcos–cosh
----------------------------------------Pn αcos( ) αsin α,d
0
π
∫=
k 0, N 1.+=
(4)
Let us introduce the following notation:
(5)
Equations (5) form a system of linear algebraic equa-
tions in coefficients Ai, n (i = ). A similar system was
n12---+⎝ ⎠
⎛ ⎞ 1
n12---+⎝ ⎠
⎛ ⎞ βi βi 1––( )sinh
-----------------------------------------------------Ai 1– n,–n 0=
∞
∑
+ n12---+⎝ ⎠
⎛ ⎞ βi βi 1––( ) n12---+⎝ ⎠
⎛ ⎞ βi 1+ βi–( )coth+coth⎝ ⎠⎛ ⎞
× Ai n,1
n12---+⎝ ⎠
⎛ ⎞ βi 1+ βi–( )sinh
-----------------------------------------------------Ai 1+ n, Pn αcos( )– = 0,
0 α αi,≤ ≤
Ai n, Pn αcos( )n 0=
∞
∑
= f i α( )
βicosh αcos–---------------------------------------Pn αcos( ) αdαsin ,
αi α π, i≤ ≤ 1 N, .=⎩⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎧
C1 n, = n12---+⎝ ⎠
⎛ ⎞ β1 β0–( )coth n12---+⎝ ⎠
⎛ ⎞ β2 β1–( )coth+⎝ ⎠⎛ ⎞
× A1 n,1
n12---+⎝ ⎠
⎛ ⎞ β2 β1–( )sinh
--------------------------------------------------A2 n, ,–
Ci n,1
n12---+⎝ ⎠
⎛ ⎞ βi βi 1––( )sinh
-----------------------------------------------------Ai 1– n,–=
+ n12---+⎝ ⎠
⎛ ⎞ βi βi 1––( )coth n12---+⎝ ⎠
⎛ ⎞ βi 1+ βi–( )coth+⎝ ⎠⎛ ⎞
× Ai n,1
n12---+⎝ ⎠
⎛ ⎞ βi 1+ βi–( )sinh
-----------------------------------------------------Ai 1+ n, , i– 2 N 1–, ,=
CN n,1
n12---+⎝ ⎠
⎛ ⎞ βN βN 1––( )sinh
---------------------------------------------------------AN 1– n,–=
+ n12---+⎝ ⎠
⎛ ⎞ βN βN 1––( )coth⎝⎛
+ n12---+⎝ ⎠
⎛ ⎞ βN 1+ βN–( )coth ⎠⎞ AN n, .
1 N,
1088
TECHNICAL PHYSICS Vol. 53 No. 8 2008
VINOGRADOVA et al.
solved in [3] for a particular case β0 = –∞, βN + 1 = ∞. In thegeneral case [4], having solved system (5), we obtain
(6)
For convenience of subsequent calculations, weintroduce new coefficients Bk, n:
(7)
Then system (4) of paired summatory equations canbe reduced to the form
(8)
where
Ai n,
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βi–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
--------------------------------------------------------=
× n12---+⎝ ⎠
⎛ ⎞ βk β0–( )Ck n,sinhk 1=
i
∑
+
n12---+⎝ ⎠
⎛ ⎞ βi β0–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
---------------------------------------------------------
× n12---+⎝ ⎠
⎛ ⎞ βk β0–( )Ck n, ,sinhk i 1+=
N
∑i 1 N, .=
B1 n,1
n12---+⎝ ⎠
⎛ ⎞ β1 β0–( )sinh
--------------------------------------------------– C1 n, ,+=
Bi n, Ci n, , i 2 N 1–, ,= =
BN n,1
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βN–( )sinh
---------------------------------------------------------- CN n, .+–=
n12---+⎝ ⎠
⎛ ⎞ Bi n, Pn αcos( )n 0=
∞
∑ 0, 0 α αi,≤ ≤=
gikn Bk n,
k 1=
i 1–
∑ 12--- gii
n+⎝ ⎠⎛ ⎞ Bi n,+
n 0=
∞
∑
+ gkin Bk n,
k i 1+=
N
∑ Pn αcos( ) Fi α( ),=
αi α π,≤ ≤⎩⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎧
giin 1
2--- e
– n12---+⎝ ⎠
⎛ ⎞ βN 1+ βi–( )
n12---+⎝ ⎠
⎛ ⎞ βi β0–( )sinh–=
(9)
(10)
(11)
(12)
+ e– n
12---+⎝ ⎠
⎛ ⎞ βi β0–( )
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βi–( )sinh
× 1
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
---------------------------------------------------------,
gikn
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βi–( ) n12---+⎝ ⎠
⎛ ⎞ βk β0–( )sinhsinh
n12---+⎝ ⎠
⎛ ⎞sinh βN 1+ β0–( )---------------------------------------------------------------------------------------------------------,–=
k i,<
gkin
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βk–( ) n12---+⎝ ⎠
⎛ ⎞ βi β0–( )sinhsinh
n12---+⎝ ⎠
⎛ ⎞sinh βN 1+ β0–( )---------------------------------------------------------------------------------------------------------,–=
i k,<
F1 α( )f 1 α( )β1cosh αcos–
----------------------------------------=
–12---
n12---+⎝ ⎠
⎛ ⎞ β1 β0–( )en
12---+⎝ ⎠
⎛ ⎞ βN 1+ β1–( )–
sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
------------------------------------------------------------------------------------
⎝⎜⎜⎜⎜⎛
–n 0=
∞
∑
+
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β1–( )en
12---+⎝ ⎠
⎛ ⎞ β1 β0–( )–
sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
--------------------------------------------------------------------------------------
⎠⎟⎟⎟⎟⎞
A0 n,
+
n12---+⎝ ⎠
⎛ ⎞ β1 β0–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
--------------------------------------------------------AN 1+ n, Pn αcos( ),
Fi α( )f i α( )βicosh αcos–
---------------------------------------=
–
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βi–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
--------------------------------------------------------A0 n,
⎝⎜⎜⎜⎛
n 0=
∞
∑
+
n12---+⎝ ⎠
⎛ ⎞ βi β0–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
--------------------------------------------------------AN 1+ n,
⎠⎟⎟⎟⎞
Pn αcos( ),
i 1 N, ,=
TECHNICAL PHYSICS Vol. 53 No. 8 2008
CALCULATION OF THE ELECTROSTATIC FIELD 1089
To solve system (8) of paired equations, we intro-duce new unknown functions φi(t) instead of coeffi-cients Bi, n with the help of the substitution
(13)
With the help of this substitution, the second equa-tions from system (8) of paired equations are trans-formed into identity [1]. Carrying out substitution (13)in the first equations in system (8), we obtain a systemof Fredholm coupled integral equations of the secondkind in functions φi(t):
(14)
(15)
Kernels Kik(x, t) of Eqs. (14) are symmetric and canbe written in the explicit form,
(16)
Thus, to solve boundary-value problem (1), we mustsolve system (14) of second-kind Fredholm equationswith right-hand sides defined by formulas (12) and (15)and with symmetric kernels (16). Consequently, formu-
FN α( )f N α( )βNcosh αcos–
-----------------------------------------=
–
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βN–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
---------------------------------------------------------A0 n,
n 0=
∞
∑
–12---
n12---+⎝ ⎠
⎛ ⎞ βN β0–( )en
12---+⎝ ⎠
⎛ ⎞ βN 1+ βN–( )–
sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
--------------------------------------------------------------------------------------
⎝⎜⎜⎜⎜⎛
+
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βN–( )en
12---+⎝ ⎠
⎛ ⎞ βN β0–( )–
sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
----------------------------------------------------------------------------------------
⎠⎟⎟⎟⎟⎞
AN 1+ n,
× Pn αcos( ).
Bi n, φi t( ) n12---+⎝ ⎠
⎛ ⎞ tsin t.d
αi
π
∫=
12---φi x( ) 2
π--- φk T( )Kik x t,( ) td
αk
π
∫k 1=
N
∑+ Φi x( ),=
αi x π,< <
Φi x( ) 2π--- d
dx------
Fi α( ) αsin
2 xcosh αcos–( )--------------------------------------------- α,d
x
π
∫=
i 1 N, .=
Kik x t,( ) gi k,n n
12---+⎝ ⎠
⎛ ⎞ x n12---+⎝ ⎠
⎛ ⎞ t.sinsinn 0=
∞
∑=
las (2), (3), (6), (7), and (13) define an analytic solutionin the entire domain of the initial boundary-value prob-lem on electrostatic potential distribution.
THE CASE OF CONSTANT VALUESOF POTENTIAL AT THE ELECTRODES
If constant values of potentials fi(α) = fi = const (i =
) are preset at all electrodes of the system,coefficients A0, n and AN + 1, n defined by formulas (3)have the form
In this case, the right-hand sides of system (14) canbe calculated explicitly:
If one of the spheres in the initial system of electrodesis missing (this corresponds to β0 = –∞ or βN + 1 = ∞), for-mulas (15) and (16) for calculating the kernels andright-hand sides of the system of coupled second-kindFredholm integral equations (14) are simplified consid-erably.
Let us suppose that βN + 1 = ∞; in this case, fromEqs. (9)–(11), we obtain
We introduce the following notation:
0 N 1+,
Ak n, 2 f ken
12---+⎝ ⎠
⎛ ⎞ βk–
, k 0 N 1.+,= =
Φi x( ) 2π--- 1
2--- f i
x2---
βi
2----coshsin
βi
2----cosh x
2---cos–
----------------------------------–
⎝⎜⎜⎜⎛
=
+ 2
n12---+⎝ ⎠
⎛ ⎞ βN 1+ βi–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
--------------------------------------------------------- f 0en
12---+⎝ ⎠
⎛ ⎞ β0–
⎝⎜⎜⎜⎛
n 0=
∞
∑
+
n12---+⎝ ⎠
⎛ ⎞ βi β0–( )sinh
n12---+⎝ ⎠
⎛ ⎞ βN 1+ β0–( )sinh
-------------------------------------------------------- f N 1+ en
12---+⎝ ⎠
⎛ ⎞ βN 1+–
⎠⎟⎟⎟⎞
---
------× n
12---x+⎝ ⎠
⎛ ⎞sin
⎠⎟⎟⎟⎞
, i 1 N, .=
giin 1
2---e
2n 1+( ) βi β0–( )–,–=
gikn 1
2--- e
n12---+⎝ ⎠
⎛ ⎞ βi βk––
en
12---+⎝ ⎠
⎛ ⎞ βi β0–( ) βk β0–( )+( )–
–⎝ ⎠⎜ ⎟⎛ ⎞
, k i.≠–=
µik βi βk– , νik βi β0–( ) βk β0–( ).+= =
1090
TECHNICAL PHYSICS Vol. 53 No. 8 2008
VINOGRADOVA et al.
Using summation of trigonometric series [5], wecan present kernels Kik(x, t) and right-hand sides Φi(x)of system (14), which are defined by formulas (15)and (16), in the form
If both spheres are absent in the system of elec-trodes, then β0 = –∞ and βN + 1 = ∞. In this case, formu-
las (15) and (16) for calculating the kernels and right-hand sides of system (14) have the form
CONCLUSIONSWe have found the electrostatic potential distribu-
tion in a system containing an arbitrary number of axi-symmetric electrodes in the form of nonconcentricspherical segments located between two closed spheri-cal surfaces. The problem of determining the unknowncoefficients in the expansion of the potential is reducedto the solution of the system of coupled second-kindFredholm integral equations. All geometrical sizes andpotentials of the segments and spheres are parametersof the problem. Such systems can be used in modelingan electron gun with a field tip. The determination ofthe potential distribution in such systems is the mostcomplicated part of the problem since the geometricalsizes of the electrodes differ by several orders of mag-nitude, which complicates the application of numericalmethods of calculation.
REFERENCES1. Ya. S. Uflyand, Method of Paired Integral Equations in
Problems of Mathematical Physics (Nauka, Leningrad,1977) [in Russian].
2. N. V. Egorov and E. M. Vinogradova, Vacuum 72, 103(2004).
3. Yu. N. Kuz’kin, Zh. Tekh. Fiz. 40, 2276 (1970) [Sov.Phys. Tech. Phys. 15, 1777 (1970)].
4. E. M. Vinogradova, N. V. Egorov, and R. Yu. Baranov,Radiotekh. Elektron. (Moscow) 45, 638 (2007).
5. I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals,Series, and Products (Nauka, Moscow, 1971; Academic,New York, 1980).
Translated by N. Wadhwa
Kik x t,( )14---
12---µiksinh
µiksinh x t–( )cos–-----------------------------------------------
⎝⎜⎜⎛
=
–
12---νiksinh
νikcosh x t–( )cos–------------------------------------------------
⎠⎟⎟⎞
12--- x t–( )cos
–
12---µiksinh
µiksinh x t+( )cos–------------------------------------------------
⎝⎜⎜⎛
–
12---νiksinh
νikcosh x t+( )cos–-------------------------------------------------
⎠⎟⎟⎞
12--- x t+( )cos , i k≠ ,
Kii x t,( )12--- βi β0–( )sinh–=
×
12--- x t–( )cos
2 βi β0–( )cosh x t–( )cos–-----------------------------------------------------------------
–
12--- x t+( )cos
2 βi β0–( )cosh x t+( )cos–----------------------------------------------------------------- ,
Φi x( ) 2π--- 1
2--- f i
x2---
βi
2----coshsin
βi
2----cosh x
2---cos–
----------------------------------– 2 f 0x2---sin+
⎝⎜⎜⎜⎛
=
--
----×
12--- βi β0–( ) β0+( )cosh
βi β0–( ) β0+( )cosh xcos–---------------------------------------------------------------------
⎠⎟⎟⎟⎞
.
Kik x t,( ) 14--- 1
2---µiksinh=
×
12--- x t–( )cos
µikcosh x t–( )cos–------------------------------------------------
12--- x t+( )cos
µikcosh x t+( )cos–-------------------------------------------------–
⎝ ⎠⎜ ⎟⎜ ⎟⎛ ⎞
,
Φi x( ) 1π--- f i
x2---
βi
2----coshsin
βi
2---- x
2---cos–cosh
----------------------------------.–=