Home Search Collections Journals About Contact us My IOPscience

Spheroidal wave functions for the hydrogen atom

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1967 Proc. Phys. Soc. 90 887

(http://iopscience.iop.org/0370-1328/90/4/301)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 128.143.23.241
This content was downloaded on 11/11/2013 at 16:10

Please note that terms and conditions apply.

p R o ~PHYS.
. SOC., 1967, VOL. 90. PRINTED IN GREAT BRITAIN

Spheroidal wave functions for the hydrogen

atom
C. A. COULSON and A. JOSEPH
Mathematical Institute, Oxford
MS. receiced 11th November 1966

Abstract. T h e Schrddinger equation for a hydrogen atom can be solved in

spherical polar, parabolic and spheroidal coordinates. Relations between these
types of solutlon are obtained by an algebraic method dependmg on the existence
of an extra constant of the motion for the two-centre Kepler problem. Limiting
conditions and certain orthogonality relations are studied. It is shown that, 111
Fock’s method of stereographic projection of momentum space onto the surface
of a four-dimensional hypersphere, one cannot find an orthogonal coordinate
system to separate the equations of motion, corresponding to the use of spheroidal
coordinates in configuration space. This is in marked contrast with the situation
with spherical polar and parabolic coordinates.

1. Introduction
The object of this paper is to point out certain results which can be obtained from a
study of the relationship between the Runge-Lenz vector and the extra constant of the
motion for the hydrogen molecule ion in the adiabatic approximation, to which attention
has recently been drawn (Coulson and Joseph 1967) (the two-centre Kepler problem).

2. Coordinate systems
The separation of variables in the Schrodinger equation for a quantum-mechanical
system, when written in a set of orthogonal coordinates, is closely related to the existence
of one or more constants of the motion, When more than one such coordinate system
exists, there will be different representations of the solutions of the Schrodinger equa-
tions, and to each of them there corresponds a set of commuting constants of the motion,
vhich may be simultaneously assigned definite eigenvalues. The relationship between
these different representations can often be more readily understood in terms of the
algebraic properties of the constants of the motion than from an analysis of the eigen-
functions, though of course there are connections between the two. I t is with the former
approach that we shall be concerned here.

3- The Kepler problem

The constants of the motion for the one-centre Kepler problem (hydrogen-like atoms)
have the remarkable property of forming the elements of an O4 algebra (Kursunoglu
1962). Furthermore, the Schrodinger equation may be solved in a variety of different
coordinate systems (Eisenhart 1934 a), each of which demonstrates a different representa-
tion of this algebra. Knowledge of the algebra thus gives information relating the
1 887
888 C. A. Coiilson and A. Joseph

solutions in different coordinates. Notably Stone (1956) has been able to show, by this
means, that the matrix elements of the unitary transformation interconverting spherical
polar and parabolic coordinate wave functions are just the Clebsch-Gordan cOefficjents,
Here we shall be concerned with the corresponding transformations from spherical to
spheroidal wave functions. These, although no less important, have to some extent been
neglected.

4. Spheroidal wave functions

If we use atomic units, so that e = 1 = m = fi, and place a nuclear charge q at the
fixed point (0, 0, a), then the Schrodinger equation for the electron in the corresponding
hydrogen-like ion is

where
r12 = x2+y2+(z-u)2.

Equation (1) may be regarded as a special case of the Schrodinger equation for the
two-centre Kepler problem in which one of the two nuclear charges is zero. Now, on
account of the existence of an extra constant of the motion (Erikson and Hill 1949),
this latter equation is separable in spheroidal coordinates. Thus a similar result holds
for the hydrogen atom. This fact may be expressed by saying that the operator

which leads to this constant of the motion, is diagonalized in the representation obtained
by separating equation (1) in spheroidal coordinates, the foci of which are at (0, 0, &a).
Now we have shown (Coulson and Joseph 1967) that the translation

x+x, y+y, z+x+u

which takes (0, 0, a) to (0, 0, 0), transforms the operator (2) into
L2-2upoA,+2a2H (3)
where p , is the constant
p, = (-ZE)1'2 (4)
and A, is the x component of the Runge-Lenz (Born and Jordan 1930) vector A (some-
times called Hamilton's vector in the classical-mechanics form)

A = -1[ - {1( L x p ) - ( p x L ) ) -
Po 2 .I;
4y
(5)

The normalization factor p , is introduced in ( 5 ) to ensure that A, is a member of the 04

algebra.
I t is now a simple matter to obtain the matrix elements in the transformation to
spherical polar or parabolic wave functions. We illustrate by choosing the former,
since it is likely to be the more important.
 Spheroidal wave functions f o r the hydrogen atom 889
The wave functions #nlm, obtained by the separation of the Schrodinger equation
for the hydrogen-like atom in spherical polar coordinates, are characterized by the
following eigenvalue equations ( H u l t h h 1932):

where n, 1, m are integers satisfying

n 2 1 > jml 20.
Thus n+ 1 corresponds to the main (or principal, or total) quantum number, and
p o = q/(n+ 1). In addition it has been shown by Miller (1964), from the properties of
the O4 algebra, that the non-zero matrix elements of A, in this representation are, to
within an unimportant phase factor,
(n-l)(n+1+2)(1+mt l ) ( l - m + l )
(A.);;;,',""
= (Az):;iYI,m =
( (21+ 1 ) ( 2 Z + 3 )
We shall denote the spheroidal wave functions by @&,. Our previous discussion
(cf. equation (3)) has shown that they may be chosen such that

A comparison of equations (6) and (8) shows that we may expand the spheroidal
functions @nkm in terms of the set )cnlm, where n and m remain fixed, but lvaries from Iml
ton. Thus
n
@flkm = 2
l=lml
Cl(k)#nlm* (9)

The expansion coefficients Cl(k)can now be derived from the conditions which follow
when (9) is substituted in (8b) and we use (6b) and (7):

The cl(k)have non-trivial values only if the determinantal equation (11) is satisfied:

0
890 C. A. Coulson and A. Joseph

This determinant is of order n + 1- Iml. It has real distinct roots in k(k +- 1) which give
the allowed, usually non-integral, values of the k quantum number. Agreement with the
results of Coulson and Robinson (1958) can be obtained if we make the identification
k ( k + l ) = A‘ (12)
+
where A’ is written iox the quantity A - a2q2/(n 1)2,and A is the separation constant
for (1) when using spheroidal coordinates. (The separation constant A must not be
confused with the operator A in ( 5 ) : we have kept the symbol A to facilitate comparison
with Coulson and Robinson (1958).) It is easily seen that the identification (12)jis
justified by separation of equations (1)and (8b),followed by a comparison of the constants,
Coulson and Robinson obtained the set of allowed values of A by applying appropriate
boundary conditions to the wave function in spheroidal coordinates, from which a
determinantal equation equivalent to (11) followed. They were then able to give an
explicit form to the eigenfunctions Dnkm.Similarly we may obtain, for each allowedvalue
of R, and chosen m and n,the set of expansion coefficients (Cl‘k);I = lml,jml+ 1, ...,E ) .
The simplest non-trivial examples are given in (13) in which aq replaces a when q # 1:
1, m=O
(1 + 2)+ (1 + a2)1:2)3:2
{ +
2(1 2)
k,(k, + 1) = 1 - (1+ a y
a[2{(1+ 2)+ (1 + a”)””] - ll2

9) + (1 + a2/9)1‘Z
2( 1 + 2/19) a2 1/2

k2(k,+1) = 4-2(1+7)

The first set of the above equations may be regarded as representing the s, P mixing
in the second quantum shell required to diagonalize the spheroidal coordinate representa-
tion. Similarly, the second set corresponds to p, d mixing in the third quantum shell with
a x component of angular momentum equal t o 3.
 Spheroidal wave functions for the hydrogetz atant 891
I t is unfortunate that the general solution of (10) and (11) cannot readily be obtained
in closed form. This is an illustration of the general intractability of spheroidal functions.
However, numerical solutions are straightforward and, in many cases, may be sufficient.

5. Limiting conditions
The behaviour of the spheroidal functions at the limiting values a +-0 and a 3 CC
is of some interest. We should, naturally, expect them to go over into spherical polar
and parabolic wave functions respectively. That this is indeed the case follows at once
from the form of the extra constant of the motion L2- 2apoAz in the spheroidal repre-
sentation. According to (8b) this is diagonalized in these coordinates. Now as a -+ 0,
this operator -+L2, which is diagonalized in spherical polars. Moreover, as a 3 cc,
the operator is increasingly dominated by the term -2ap,A, and, as Bargmann (1936)
has shown, A, is diagonalized in parabolic coordinates.
Another limiting condition may be obtained, for small np, by expanding the eigen-
value R as a power series of the form
m

ks = c %Ls,m(adt.
t=o
(t)

Substitution of (14) in (11) gives

where s is any integer satisfying

n 2 s 2 jmj
and may therefore be identified at a = 0 with the orbital angular momentum quantum
number 1.
The coefficients mh?)ssmmay be obtained in closed form. They are

n,s8m -
E (2) - 2
(2s- 1 ) ( 2 ~ +
1)(2~+3)
{n(n+Z)+s(s+ 1)+nz2}
- -1.
3m2
s(s+ 1)
(16)

Substitution of these approximate eigenvalues in (10) yields the corresponding set of

approximate expansion coefficients C,(k).
A similar development for large values of aq can be obtained from a study of the
relationship between spheroidal and parabolic coordinates.

6. Orthogonality
The chief importance of the spheroidal wave functions for a hydrogen-like atom
results from the fact that these functions are the correct zero-order functions to use, in a
perturbation expansion, when the perturbation arises from a second coulombic centre
(at (0, 0, - a)). This is a consequence of the orthogonality relationship

except when
m’ = m, and R ( R f 1) = K’(k’+ 1) (see equation (12))
where
Y22 = x2+y2+(z+u)?
892 C. A. Coulson aizd A. Joseph

This result was first pointed out by Coulson and Robinson (1958). However, we shall
rederive it by explicit reference only to the relevant constants of the motion.
The existence of the extra constant of the motion for the two-centre problem with
charges q1 and q2 can be exhibited by the commutation relation (Coulson and Joseph
1967)

Thus the result

(@nicmt F @ n k , m , ) 0 E (19)
holds trivially. The case of m‘ # m in (17) is elementary; we may therefore restrict
ourselves tc m’ = m. Then substitution of the eigenvalues of the operators in %leads to

The term in En vanishes on account of the orthogonality of the (Dnkm, and so we reduce
to (17), as required. I n addition it may be shown that when n f n’

By a similar argument using the relevant constant of the motion (Redmond 1964)
the well-known result (Landau and Lifshitz 1958) may be established that the parabolic
solutions diagonalize the perturbation due to a uniform electric field (the Stark effect).

7. Momentum space
A global realization of the O4 symmetry of the hydrogen atom was given by Fock
(1935). He showed that, by a stereographic projection of momentum coordinates, the
Schrodinger equation (1) could be reduced to the problem of free motion on the surface
of a suitable four-dimensional hypersphere. Furthermore, the angular momentum L
and the Runge-Lenz vector A generate rotations on this hypersphere. It is not sur-
prising that the spherical polar and parabolic coordinate systems have their analogues
in this projected space. It would be very illeresting to see whether this is also true of
spheroidal coordinates. For, if it were true, it might give an alternative approach to the
eigenvalue problem in this representation. However, we can show that it is not true.
The argument is as follows.
Eisenhart (1934 b) has shown that there are at most three orthogonal coordinate
systems such that the Schrodinger equation on the hypersphere separates, and such
that one of the components of angular momentum (L, say) is diagonalized. We have
already seen that two of these characterize the spherical polar and parabolic solutions for
the hydrogen atom in ordinary configuration space. The third is given by
x1 = Y cos a cn(b, a ) cn(c, a’)
x2 = r sin a cn(b, a ) cn(c, a‘)
(21)
xs = Y dn(6, a ) sn(c, x’)

x4 = rsn(b, a ) dn(c, a’)

where x 2 + U‘’ = 1, and where sn, cn, dn are the Jacobian elliptic functions (Whittaker
 Spheroidal wave functions for the hydrogen atom 893
and Watson 1946). Now, unfortunately, separation of the equation for free motion in
these coordinates does not correspond to the use of spheroidal coordinates in ordinary
configuration space. In fact it is characterized by the eigenvalue equations

Lzynpm = mrnpm

(Pi-X'ZAa")'rnp, = p n p m (22)

These equations are not the same as those set out in @a, b, c). Once again we see the
intractable nature of spheroidal coordinates.

8. The harmonic oscillator

In conclusion it may be mentioned that the Schrodinger equation for the a-dimen-
sional isotropic harmonic oscillator (n 2 2) may be separated in several coordinate
systems, including spheroidal coordinates (Coulson and Joseph 1967). An analysis
similar to the present one may therefore be given to interrelate these different representa-
tions. Further details will be published elsewhere.

Acknowledgments
One of us (A.J.) wishes to express his gratitude to Corpus Christi College, Oxford,
for a Junior Research Fellowship, and to the Science Research Council for a maintenance
grant.

References
BARGMANN,V., 1936, Z. Phys., 99, 576-82.
BORN,M., and JORDAN,P., 1930, Elementare Qztantenmechanzk (Berlin: Springer-Verlag), 0 35,
pp. 177-83.
COULSON, C. A., and JOSEPH,A., 1967, to be published.
COULSON, C. A., and ROBINSON, P. D., 1958, Proc. Phys. SOC.,
71, 815-27.
EISENHART, L. P., 1934 a, Phys. Rev., 45, 427-8.
- 1934 b, Ann. Math., Pranceton, 35, 284-305.
ERIRSON, H. A., and HILL,E. L., 1949, Phys. Rev., 75, 29-31.
FOCK,V., 1935, Z. Phys., 98, 145-54.
HULTH~N, L., 1932, Z. Phys., 86, 21-3.
KURSUNOGLU, B., 1962, Modern Qzdanttmz Theory (San Francisco, London. Freeman Press),
pp. 362-7.
LASDAU,L. D., and LIFSHITZ, E. M., 1958, Quantum Mechanics (Oxford: Pergamon Press),
5 73, pp. 251-7.
MILLER,W., 1964, Mem. Amer. Math. Soc., No. 50, 30-9.
REDMOND, P. J., 1964, Phys. Rev., 133, B1352-3.
STOSE,A. P,, 1956, Proc. Camb. Phtl. Soc., 52,424-30.
bi7HITTAKE~,E. T., and WATSON, G. N., 1946, Modern Analyszs, 2nd edn (Cambridge Cambridge
University Press), chap. 20, pp. 429-61.