Vous êtes sur la page 1sur 28

Quantum Mechanics of Rotating Molecules

W. Happer
November 6, 2013
1 Introduction
It is hard to nd a discussion of rotating bodies in either elementary or advanced quantum
mechanics texts. Perhaps the best available is the chapter on polyatomic molecules in the
text Quantum Mechanics by Landau and Lifschitz[1]. These notes outline the basic quantum
mechanics of rotating bodies, a key part of the physics of polyatomic molecules, deformed
nuclei and some other parts of physics. Since the notes were written in a hurry over the
mid-term break, they undoubtedly contain many errors and unclear parts. Feedback on how
to improve the notes, either in person or by e-mail, would be welcome. I will post updated
versions of these notes every day or two as we discuss rotating molecules over the next two
week.
Rigid rotors. The simplest rotating system is a rigid rotator. A classical example is an
innitely thin straight rod that spins freely in space, much like a twirling baton that has
been tossed by a band majorette. A quantum mechanical example would be a rotating H
2
molecule. One can specify the orientation of the internuclear axis of the molecule with a
colatitude angle and an azimuthal angle . The rotational Hamiltonian of the molecule is
H =

2
2I

L L, (1)
where L is the angular momentum operator and I

is the moment of inertia for rotations


around any axis through the center of mass and perpendicular to the internuclear axis. The
energy eigenfunctions and quantized energies are

jm
= Y
jm
, and E
j
=

2
2I

j(j + 1), (2)


where Y
jm
= Y
jm
(, ) denotes a spherical harmonic. The angular momentum quantum
numbers j = 0, 1, 2, . . . and m = l, l 1, . . . , l are given by the eigenvalue equations
L L
jm
= j(j + 1)
jm
,
L
3

jm
= m
jm
. (3)
Here L
3
= x
3
L is the projection of the angular momentum operator along the z axis, x
3
of a
xed, laboratory coordinate system. The energy levels E
j
of (2) are (2j +1)fold degenerate
since all of the wave functions
jj
,
j,j1
, . . . ,
j,j
have the same energies. The azimuthal
quantum number m has no eect on the energy.
1
Symmetric tops. The next simplest rotating system is the symmetric top. A classical
example is an elongated American football, rotating freely in space with no air resistance.
A quantum mechanical example is a molecule like methyl chloride, ClCH
3
, with a three-fold
rotational symmetry axis through the Cl and C nuclei. Symmetric tops have three nonzero
moments of inertia, two of which are identical. Three parameters, often taken to be the
Euler angles, , are needed to specify the orientation of a symmetric top. The rst gure
of the Wikipedia article on Euler angles gives a good sketch of . We can think of
the symmetry axis as being dened by an aximuthal angle = and by a colatitude angle
= . The amount that the top has been rotated about its symmetry axis is given by the
third Eulers angle . The rotational Hamiltonian of a symmetric-top molecule is
H =

2
2I

L L +

2
2
_
1
I

1
I

_

L
2
3
(4)
Here

L
3
= x
3
L is the projection of the angular momentum operator along the symmetry
axis x
3
of the molecule. The energy eigenfunctions and quantized energies are

jml
=

2j + 1
8
2
D
j
ml
, and E
jl
=

2
2I

j(j + 1) +

2
2
_
1
I

1
I

_
l
2
. (5)
Here D
j
mk
is the complex conjugate of the Wigner D-Function, D
j
mk
= D
j
mk
(, , ). The co-
ecient

(2j + 1)/(8
2
) is for normalization. A very through discussion of the D-Functions
can be found in Quantum Theory of Angular Momentum by Varshalovich, Morskalev and
Khersonskii [2]. We will have more to say about the D-Functions later in these notes. The
moment of inertia around the symmetry axis of the symmetric top is I

and the two equal


moments about independent principal axes perpendicular to the symmetry axis are I

. If
I

> I

(e.g. ClCH
3
) the top is said to be prolate. If I

< I

(e.g. SO
3
) the top is said to
be oblate. A symmetric top with I

= I

(e.g. CH
4
) is said to be a spherical top.
The angular momentum quantum numbers j = 0, 1, 2, . . ., m = j, j 1, . . . , j, l =
j, j 1, . . . , l are given by the eigenvalue equations
L L
jml
= j(j + 1)
jml
,
L
3

jml
= m
jml
,

L
3

jml
= l
jml
. (6)
The energy levels E
jl
of (5) are (2j + 1)fold degenerate since all of the wave functions

jjl
,
j,j1,l
, . . . ,
j,j,l
have the same energies. The quantum number l is the projection
of the angular momentum on the symmetry axis. From (5) we see that E
jl
= E
j,l
so
unless l = 0 the energy states with specied values of j and m will have an additional two-
fold degeneracy for l. Various small perturbations always lift the degeneracy in l and
produce a closely-spaced energy doublet. The characteristic doublet structure of symmetric-
top energy levels plays an signicant role in many physical processes.
In the limit that the molecule is linear of nearly linear we will have I

. From (5) we
see that as I

0, E
jl

2
l
2
/(2I

if l = 0. States with quantum number l = 0 will


have so much energy that they are almost never excited for molecules in thermal equilibrium.
2
Then the only states of (5) that occur are those with l = 0, for which (5) becomes

jm0
(, , ) =

2j + 1
8
2
D
j
m0
=
1

2
Y
jm
(, ), and E
jl
=

2
2I

j(j + 1). (7)


The states and energies (7) are the same as those of rigid rotor, given by (2), aside from
the factor of 1

2 multiplying the spherical harmonic, which ensures that the wave func-
tion is normalized for integrating over dsin dd instead of sin dd. For the identity
D
j
m0

(2j + 1)/(4) = Y
jm
see 4.17 (1) of Varshalovich et al.[2].
Asymmetric tops. The most complicated rotating system is the asymmetric top. A
classical example might be an Australian boomerang, rotating freeing in space with no air
resistance. A quantum mechanical example would be a water molecule, H
2
O, for which the
three principal moments of inertia, I
1
, I
2
and I
3
are all dierent. The molecular orientation
can be still specied by Euler angles, , just as for the symmetric top, with the orientation
of one of the principal axes, say x
3
, dened by an azimuthal angle = and by a colatitude
angle = . The amount that the top has been rotated about x
3
, is given by the third
Eulers angle . The rotational Hamiltonian of an asymmetric-top molecule is
H =

2
2
_

L
2
1
I
1
+

L
2
2
I
2
+

L
2
3
I
3
_
. (8)
Here

L
k
= x
k
L for k = 1, 2, 3 is the projection of the angular momentum operator along
the kth principle axis x
k
of the molecule. The energy eigenfunctions are superpositions of
symmetric-top eigenfunctions

jmn
=

2j + 1
8
2

l
D
j
ml
v
ln
, with

l
|v
ln
|
2
= 1. (9)
For asymmetric tops there is no general formula for the energies, analogous to those of (2)
and (5). The expansion coecients v
ln
are the eigenvectors of the Hamiltonian (8).
The angular momentum quantum numbers j = 0, 1, 2, . . ., m = j, j 1, . . . , j and the
energy E
jn
of the wave function (9) are given by the eigenvalue equations
L L
jmn
= j(j + 1)
jmn
,
L
3

jmn
= m
jmn
,
H
jmn
= E
jn

jmn
. (10)
The energy levels E
jn
of (10) are (2j + 1)fold degenerate since all of the wave functions

jjn
,
j,j1,n
, . . . ,
j,j,n
have the same energies.
2 Rotations
Let a molecule contain atoms, with masses M
1
, M
2
, . . . , M

. We suppose that the center


of mass of the molecules is at rest at the origin of a xed, laboratory coordinates system,
3
and that the atomic constituents are displaced from the center of mass by the distances
r
1
, r
2
, . . . , r

. Then the masses and position vectors satisfy the constraint

= 0. (11)
We introduce Cartesian basis vectors along the axes of the laboratory coordinate system,
x = x
1
, y = x
2
, and z = x
3
. (12)
We can write the position vector of the th atom in the molecule as
r

=
3

j=1
x
j
r
j
x
j
r
j
, where r
j
= x
j
r

(13)
In (13) and elsewhere, we will often use the Einstein convention, that repeated Cartesian
indices of vector or tensor elements are to be summed over their three possible values. The
basis vectors (12) form a right-handed, orthonormal coordinate system, such that
x
i
x
j
=
ij
, and x
i
x
j
=
ijk
x
k
. (14)
The antisymmetric unit tensor
ijk
is dened by

ijk
=
_
_
_
1 if ijk = 123, 231, 312,
1, if ijk = 213, 132, 321,
0 otherwise.
(15)
We see from (14) that antisymmetric unit tensor can be written as
(x
i
x
j
) x
k
= x
i
(x
j
x
k
) =
ijk
. (16)
We will often use (16) to write the cross product of two vectors, A = x
i
A
i
and B = x
j
B
j
as
AB = A
i
B
j
x
k

ijk
. (17)
On the right side of (17) it is understood that the repeated indices i, j, and k, are summed
over.
Rigid molecules. We assume that the center of mass of the molecule at rest and that the
molecules are rigidly attached to each other so they can only move if the entire molecule
rotates with instantaneous angular velocity . The velocity of the th atom must be
v

= r

. (18)
Perturbation theory can be used to account for small vibrational contributions to the ve-
locities when the constituent atoms of the molecule move with respect to each other. The
angular velocity can be written in terms of the basis vectors (12) as
=
j
x
j
(19)
4
We can use (18) to calculate the classical angular momentum L (in units of ) of the molecule
about its center of mass,
L =

( r

)
=

(
i
r
j
x
k
)
ijk
=

r
s

i
r
j

ijk

skt
x
t
=

(r
j

i
r
j
x
i
r
i

i
r
j
x
j
). (20)
Here we used the simple identity, that can be veried by inspection,

ijk

uvk
= (x
i
x
j
) (x
u
x
v
) =
uv
ij
=
iu

jv

iv

ju
. (21)
We often write (20) as the formally simpler expression,
L = I . (22)
The moment-of-inertia dyadic is
I =

(r
j
r
j
x
i
x
i
r
j
r
i
x
j
x
i
) =

(r
2

U r

). (23)
We will use math boldface fonts to denote dyadics, direct products of two vectors, as
opposed to dot-products or a cross-products. An example is the unit dyadic used in (23)
U = x
i
x
i
= x
1
x
1
+x
2
x
2
+x
3
x
3
. (24)
The matrix elements of the inertial dyadic between laboratory basis vectors are the elements
of the inertial tensor,
I
ij
= x
i
I x
j
=

(r
2

ij
r
i
r
j
). (25)
We see that the inertial tensor (25) is real and symmetric, I
ij
= I
ji
. It therefore has three
real eigenvalues I
k
for k = 1, 2, 3 and three orthogonal eigenvectors, which we denote by x
k
.
Aside from normalization, the eigenvectors are dened by the eigenvalue equation
I x
k
= I
k
x
k
. (26)
The doubled index on the right of (26) is not summed over. We will call the eigenvectors
x
k
body-xed basis vectors, and we will enumerate them so they have orthonormality
properties completely analogous to those of (14) and (24), that is
x
i
x
j
=
ij
, x
i
x
j
=
ijk
x
k
, and x
i
x
i
= U (27)
The eigenvalues I
k
of the inertial tensor are the principal moments. From(26), (27) and
(23) we see that
I
k
= x
k
I x
k
=

(r
2

|x
k
r

|
2
) =

,j=k
M

|x
j
r

|
2
0. (28)
5
According to (28), the principal moments cannot be negative, although for linear molecules
like CO
2
, one of the moments can be nearly zero, so the molecule can be well approximated
as a rigid rotor. For the remainder of this discussion we will assume that all three principal
moments are positive and of comparable sizes. This is always true for nonlinear molecules
consisting of three or more atoms.
Multiplying both sides of (26) on the right by x
k
, summing over the repeated index k
and using the completeness property of (27), x
k
x
k
= U, we nd that the inertial dyadic can
be written as
I =

k
I
k
x
k
x
k
. (29)
Since we assume that none of the principal moments I
k
is zero, we see by inspection of (29)
that the inverse inertial dyadic is
I
1
=

k
x
k
x
k
I
k
, with I I
1
= I
1
I = U (30)
If we multiply both sides of (22) on the left by I
1
, and recall that I
1
is symmetric we nd
= I
1
L = L I
1
. (31)
Using (18) we nd that he kinetic energy H of the rotating molecule is
H =

1
2
M

1
2
M

( r

) ( r

)
=
1
2

i
r
j

u
r
v

ijk

uvk
=
1
2

i
r
j

u
r
v

uv
ij
=
1
2

(
i
r
j

i
r
j

i
r
j

j
r
i
) =
1
2

i
I
ij

j
. (32)
From (32) and (31) we see that we can write the energy as
H =
1
2
I =

2
2
L I
1
L. (33)
The Hamiltonian (33) is often taken to be the starting point for discussing rotations of rigid
bodies in both classical and quantum mechanics. The Hamiltonians (4) and (8) come from
(33) with the angular momentum L interpreted to be a quantum mechanical operator.
2.1 Euler angles.
The orientation of a rigid body is conveniently described with the Euler angles , which
parameterize the relative orientation of laboratory-xed basis vectors x
k
and body-xed basis
vectors x
j
We can use the unit dyadic (24) to write the body-xed basis vectors x
j
in terms
of the laboratory-xed basis vectors x
k
,
x
j
= U x
j
= x
k
X
kj
. (34)
6
The elements of the direction-cosine matrix are
X
kj
= x
k
x
j
. (35)
The direction-cosine dyadic is dened to be
X = x
k
x
j
X
kj
= x
j
x
j
. (36)
The direction-cosine dyadic rotates laboratory-xed unit vector to body-xed unit vectors
x
j
= X x
j
(37)
We will write X as the product of three successive rotations
X = C

A. (38)
The dyadic A rotates the laboratory basis vectors x
j
by an angle about the axis x
3
into
the primed basis vectors
x

j
= A x
j
= x
k
A
kj
. (39)
The elements of the 3 3 matrix A are
A
kj
= x
k
A x
j
, with A =
_
_
cos sin 0
sin cos 0
0 0 1
_
_
. (40)
In analogy to (36) we can write the rotation dyadic A as
A = x
k
x
j
A
kj
= x

j
x
j
. (41)
The rotation dyadic A is unitary and the elements A
kj
are real so that
A
1
= x
j
x
k
A
1
jk
= A

= x
j
x
k
A
kj
= x
j
x

j
, or A
1
jk
= A
kj
. (42)
The middle dyadic B

of (38) rotates the primed basis vectors x

j
by an angle about
the axis x

2
(the line of nodes) into the double-primed basis vectors
x

j
= B

j
= x

k
B
kj
. (43)
The elements of the 3 3 matrix B are
B
kj
= x

k
B

j
, or B =
_
_
cos 0 sin
0 1 0
sin 0 cos
_
_
. (44)
In analogy to (36) we can write the rotation dyadic B

as
B

= x

k
x

j
B
kj
= x

j
x

j
. (45)
In analogy to (42) we can write
B

1
= x

j
x

k
B
1
jk
= B

= x

j
x

k
B
kj
= x

j
x

j
, or B
1
jk
= B
kj
. (46)
7
The leftmost dyadic C

of (38) rotates the double-primed basis vectors x

j
by an angle
about the axis x

3
into the body-xed basis vectors
x
j
= C

j
= x

k
C
kj
. (47)
The elements of the 3 3 matrix C are
C
kj
= x

k
C

j
, or C =
_
_
cos sin 0
sin cos 0
0 0 1
_
_
. (48)
In analogy to (36) we can write the rotation dyadic C

as
C

= x

k
x

j
C
kj
= x
j
x

j
. (49)
In analogy to (42) we can write
C

1
= x

j
x

k
C
1
jk
= C

= x

j
x

k
C
kj
= x

j
x
j
, or C
1
jk
= C
kj
. (50)
We can use (39) and (45) to write
B

= A x
j
x
k
A
1
B
jk
= A B A
1
, (51)
where we noted from (42) that x

k
= x
k
A
1
and we dened the unprimed rotation dyadic
B as
B = x
j
x
k
B
jk
. (52)
Proceeding in a similar manner and using (51) we nd
C

= B

j
x

k
B

1
C
jk
= A B A
1
x

j
x

k
A B
1
A
1
C
jk
= A B x
j
x
k
B
1
A
1
C
jk
= A B C B
1
A
1
(53)
where we noted from (41) and (42) that A
1
x

j
= x
j
and x

j
A = x
j
, and where we dened
the unprimed rotation dyadic C as
C = x
j
x
k
C
jk
. (54)
Using (51) and (53) we nd that we can write the rotation dyadic (38) as
X = C

A = A B C, (55)
or
X = ABC. (56)
Inserting the 33 matrices for A, B and C from (40), (44) and (48) into (56) we nd matrix
direction cosines X
kj
= x
k
x
j
between laboratory-xed axes x
k
and body-xed axes x
j
.
X =
_
_
cos cos cos sin sin cos cos sin sin cos cos sin
sin cos cos + cos sin sin cos sin + cos cos sin sin
sin cos sin sin cos
_
_
. (57)
We can parameterize every possible rotation with Euler angles in the range
0 < 2, 0 < , 0 < 2. (58)
8
2.2 Spin matrices
If we replace the nite rotation angle by an innitesimal angle, || << 1 we can expand
of (40) as a power series to terms of order to nd the intesimal rotation
A = U iS
3
. (59)
Here the unit matrix is
U =
_
_
1 0 0
0 1 0
0 0 1
_
_
(60)
Here S
3
is one of the three spin matrices
S
1
=
1
i
_
_
0 0 0
0 0 1
0 1 0
_
_
, S
2
=
1
i
_
_
0 0 1
0 0 0
1 0 0
_
_
, S
3
=
1
i
_
_
0 1 0
1 0 0
0 0 0
_
_
. (61)
One can readily see that the elements of the spin matrices are given by
(S
i
)
jk
=

ijk
i
. (62)
The elements of the antisymmetric unit tensor were given by (15). Squaring the spin matrices
of (61) we nd
S
2
1
=
_
_
0 0 0
0 1 0
0 0 1
_
_
, S
2
2
=
_
_
1 0 0
0 0 0
0 0 1
_
_
, S
2
3
=
_
_
1 0 0
0 1 0
0 0 0
_
_
. (63)
Summing the squares we nd
S
i
S
i
= 2U = s(s + 1)U. (64)
It is also straightforward to demonstrate that the matrices (61) satisfy the commutation
relations
[S
i
, S
j
] = i
ijk
S
k
(65)
We conclude that the three laboratory-xed basis vectors x
k
act like the three independent
spin components of a particle of spin quantum number s = 1.
We can use (62), (63) and the 3 3 unit matrix U, with matrix elements U
jk
=
jk
, as
basis matrices to write the matrix (39) for a nite rotation angle as
A = U S
2
3
iS
3
sin + S
2
3
cos = e
iS
3
. (66)
As indicated in (66), we can also write A = e
i
3
by expanding the exponential in a power
series, e
x
= 1 + x + x
2
/2! + x
3
/3! + , with x = iS
3
, and noting that for i = 1, 2, 3,
S
i
= S
3
i
= S
5
i
= S
7
i
=
S
2
i
= S
4
i
= S
6
i
= S
8
i
= (67)
9
There is no sum on repeated indices i in (67). In like manner we can show that
C = U S
2
3
iS
3
sin + S
2
3
cos = e
iS
3
,
B = U S
2
2
iS
2
sin + S
2
2
cos = e
iS
2
, (68)
Equivalent dyadic spin matrices are readily constructed, for example,
S
3
= x
j
x
k
(S
3
)
jk
, S

2
= x

j
x

k
(S
2
)
jk
, S

3
= x

j
x

k
(S
3
)
jk
. (69)
The rotation dyadics of (42), (44) and (48) for the individual Euler angles become
A = exp(iS
3
), or A = U S
3
S
3
iS
3
sin +S
3
S
3
cos . (70)
B

= exp(iS

2
), or B

= U S

2
S

2
iS

2
sin +S

2
S

2
cos . (71)
C

= exp(iS

3
), or C

= U S

3
S

3
iS

3
sin +S

3
S

3
cos . (72)
Here it is understood that for any dyadic operator M, constructed from 3 3 matrices in
analogy to (69) the exponential function is
exp(M) = U +M +
1
2!
(M)
2
+
1
3!
(M)
3
+ (73)
The full rotation dyadic (55) becomes
X = exp(iS

3
) exp(iS

2
) exp(iS
3
)
= exp(iS
3
) exp(iS
2
) exp(iS
3
). (74)
2.3 Rotational velocity
We can write the angular velocity of (18) in terms of the rates of change of Euler angles
and their axes of rotation as
= x
3
+

x

2
+ x

3
. (75)
We can use (39) and (40) to write
x

2
= x
k
A
k2
= x
1
sin +x
2
cos . (76)
Noting that x

3
= x
3
, we use (36) and (57) to write
x

3
= x
k
X
k3
= x
1
cos sin +x
2
sin sin +x
3
cos . (77)
Substituting (76) and (77) into (75) we nd that the rotational velocity can be written as
=
j
x
j
, (78)
where the coecients
j
of the laboratory-xed unit vectors x
j
can be described by the
matrix equation
_
_

3
_
_
=
_
_
0 sin cos sin
0 cos sin sin
1 0 cos
_
_
_
_


_
_
. (79)
10
The inverse of (79) is
_
_


_
_
=
1
sin
_
_
cos cos sin cos sin
sin sin cos sin 0
cos sin 0
_
_
_
_

3
_
_
. (80)
To nd the projections

j
of the rotational velocity on the body-xed bases we use (47)
and (48) with the fact that C
1
= C

to write
x

2
= x

2
= x
j
C
1
j2
= C
2j
x
j
= x
1
sin + x
2
cos , (81)
and we use (36) and (57) with the fact that X
1
= X

to write
x
3
= X
3j
x
j
= x
1
sin cos + x
2
sin sin + x
3
cos . (82)
Substituting (81) and (82) into (75) we nd that the rotational velocity can be written as
=

j
x
j
, (83)
where the coecients

j
of the body-xed unit vectors x
j
can be described by the matrix
equation
_
_

3
_
_
=
_
_
sin cos sin 0
sin sin cos 0
cos 0 1
_
_
_
_


_
_
. (84)
The inverse of (84) is
_
_


_
_
=
1
sin
_
_
cos sin 0
sin sin sin cos 0
cos cos cos sin sin
_
_
_
_

3
_
_
. (85)
2.4 Angular Momentum Operators
Our discussion of rigid bodies so far has been entirely classical. We can use (84) with (33)
to write the Hamiltonian in terms of the coordinates and their derivatives

and
solve for the classical motion with Lagrangian or Hamiltonian classical mechanics. But in
quantum mechanics, the coordinate derivaties

cannot be taken as independent variables.
Rather, the dependence of the rigid-body orientation on time t is dened a wave function
= (, , , t). (86)
The probability,at time t, of nding the orientation of the rigid body in the angle-space vol-
ume element dsin d d centered on specied by Euler angles is |(, , , t)|
2
dsin d d.
The evolution of the wave function is described by the time-dependent Schroedinger equation
i

t
= H =

2
2
L I
1
L (87)
11
We use the classical Hamiltonian (33) but with the angular momentum L interpreted as an
operator on functions of the Euler angles.
Setting aside rigid bodies briey to discuss the simpler case of a point particle with
the wave function (r, t), we recall that we can dene wave functions that have subject to
innitesimal displacements in t time, r space or in angular orientation by
T
t
(t + t) = (t), or T
t
(t) = (t t) (88)
T
r
(r + r) = (r), or T
r
(r) = (r r) (89)
T

(r + r) = (r), or T

(r) = (r r). (90)


The values of the displaced wave functions at the displaced values of the coordinates are
the same as the values of the original wave functions at the at the original values of the
coordinates. From (88)(90) we see that
T
t
=
_
1 t

t
_
= (1 +it H/) (91)
T
r
=
_
1 r
_
= (1 ir p/) (92)
T

= (1 [ r] ) = (1 i L). (93)
The familiar expression L = ir for the orbital angular momentum (in units of ) of a
particle moving in three dimensions comes from (93).
Suppose we rotate the wave function (86) through the innitesimal angle . As a result
of the rotation, an orientation that was originally specied by the Euler angles will
be specied by the slightly incremented Euler angles + , + and + . We will
assume that the angular momentum operator L for a rigid body is determined by expressions
analogous to (90) and (93), so that
T

(, , ) = ( , , ) = (1 i L)(, , ). (94)
From (94) we see that the orbital angular momentum operator for a rigid-body wave function
must be dened by
L =
1
i
_

_
. (95)
To nd dierential operators to represent the projections L
j
= x
j
L, on the right of (95) we
let = t, =

t and = t, we use (80) to express ,

and in terms of
j
. On
the left of (95) we set L = t
j
L
j
. Equating coecients of
j
we nd the projections of
the orbital angular momentum operators on the laboratory-xed basis vectors,
L
1
=
1
i sin
_
cos cos

sin sin

+ cos

_
. (96)
L
2
=
1
i sin
_
sin cos

+ cos sin

+ sin

_
. (97)
L
3
=
1
i

. (98)
12
From (96) and (97) we nd
L
1
iL
2
=
e
i
i sin
_
cos

i sin

_
. (99)
To nd dierential operators to represent the projections

L
j
= x
j
L, on the right of (95) we
let = t, =

t and = t, we use (85) to express ,

and in terms of

j
. On
the left of (95) we set L = t

L
j
. Equating coecients of

j
we nd the projections of
the orbital angular momentum operators on the body-xed basis vectors,

L
1
=
1
i sin
_
cos

+ sin sin

+ cos cos

_
. (100)

L
2
=
1
i sin
_
sin

+ sin cos

cos sin

_
. (101)

L
3
=
1
i

. (102)
From (100) and (101) we nd

L
1
i

L
2
=
e
i
i sin
_

i sin

+ cos

_
. (103)
2.5 Commutators of L
k
and

L
k
Not surprisingly, the projections L
j
= x
j
L of L on laboratory axes x
j
satisfy the usual
commutation relations
[L
j
, L
k
] = i
jkv
L
v
, and [L
j
, L
k
L
k
] = 0. (104)
One can verify that the dierential operators (96) (98) satisfy the commutation relations
(104).
People are often surprised to nd that the commutation relations for the projections

L
k
= x
k
L on the body-xed axes x
j
have the wrong sign,
[

L
j
,

L
k
] = i
jkv

L
v
, and [

L
j
,

L
k

L
k
] = 0. (105)
Other interesting properties of the angular momentum operators are
[L
j
,

L
k
] = 0, (106)
that is, any laboratory-xed projection L
j
commutes with any body-xed projection

L
k
. We
also nd that
L
j
L
j
=

L
j

L
j
= L L. (107)
The identities (105) (107) can be proved in a straightforward but tedious way with the
dierential-operator expressions (100)(102) and (96)(98). We outline a less tedious proof
here.
13
Rotating the body-xed basis vectors x
k
by an additional innitesimal angle will
transform them to x
k
+ x
k
, and we see from (35) that the value of the direction-cosine
matrix at the incremented Euler angles will be
X
jk
( + , + , + ) = x
j
( x
k
+ x
k
) = (1 + i L)X
jk
. (108)
In writing (108) we set = X
jk
in (??) and used (??) for E
1
. From (108) we nd
LX
jk
= i (x
j
x
k
) (109)
Setting = x
r
in (109) and recalling that x
k
= x
s
X
sk
, we can equate coecients of
to nd
L
r
X
jk
= i
rjs
X
sk
= [L
r
, X
jk
]. (110)
Setting = x
r
in (109) are recalling that x
j
= X
js
x
s
, we can equate coecients of to
nd

L
r
X
jk
= i
rks
X
js
= [

L
r
, X
jk
]. (111)
According to (110) L
r
changes the left subscript of X
jk
and according to (111)

L
r
changes
the right subscript.
Writing the body-xed projections as

L
j
= x
j
L = x
j
x
k
L
k
= X
kj
L
k
, (112)
we nd the commutation relations
[

L
j
,

L
k
] = (X
rj
L
r
X
sk
L
s
X
sk
L
s
X
rj
L
r
)
= (X
rj
X
sk
L
r
L
s
X
sk
X
rj
L
s
L
r
)
+i (
rsl
X
rj
X
lk
L
s

srl
X
sk
X
lj
L
r
)
= X
rj
X
sk
[L
r
, L
s
] + i (
rsl
X
rj
X
lk
L
s

srl
X
sk
X
lj
L
r
)
= i (X
rj
X
sk

rsl
L
l
+
rsl
X
rj
X
lk
L
s

srl
X
sk
X
lj
L
r
)
= i (
jkv
X
lv
L
l

jkv
X
sv
L
s

jkv
X
rv
L
r
)
= i
jkv

L
v
. (113)
In the proof (113) we made use (110), (104) and of the identity

rst
X
ri
X
sj
=
ijk
X
tk
, (114)
which follows from
x
i
x
j
=
ijk
x
k
, with x
i
= x
r
X
ri
, etc. (115)
Any laboratory-xed component L
j
of the angular momentum operator commutes with
any body-xed component

L
k
since
[L
j
,

L
k
] = L
j
X
rk
L
r
X
rk
L
r
L
j
= i
jrv
X
vk
L
r
+ X
rk
L
j
L
r
X
rk
L
r
L
j
= i
jrv
X
vk
L
r
+ iX
rk

jrv
L
v
= i
jrv
X
vk
L
r
+ iX
rv

jvr
L
r
= 0. (116)
14
The sum of the squares of the three projections L
j
is equal to the sum of the squares of
the three projections

L
j
.
L L = L
j
L
j
= X
jr

L
r
X
js

L
s
= X
jr
X
js

L
r

L
s
iX
jr

rsv
X
jv

L
s
=
rs

L
r

L
s
i
rsr
L
s
=

L
r

L
r
. (117)
3 Spherical Basis
We dene complex spherical basis vectors

with azimuthal quantum numbers = 1, 0, 1


in the laboratory-xed system as

1
=
x
1
+ ix
2

2
,
0
= x
3
,
1
=
x
1
ix
2

2
, (118)
or in the body-xed system as

1
=
x
1
+ i x
2

2
,

0
= x
3
,

1
=
x
1
i x
2

2
. (119)
We can write (118) and (119) as

= x
i
N
i
, or

= x
j
N
j
, where N =
1

2
_
_
1 0 1
i 0 i
0

2 0
_
_
. (120)
One can verify that the transformation matrix N is unitary
N

= N
1
, (121)
so that the inverse of (120) is
x
j
=

j
, or x
j
=

j
, where N

=
1

2
_
_
1 i 0
0 0

2
1 i 0
_
_
. (122)
The spherical basis vectors are orthonormal and complete in the sense that

, and

= U
or

, and

= U. (123)
We can write the unit dyadic as
U =

= x
i
N
i

= x
j
x
j
=

j
x
j
or
U =

= x
i
N
i

= x
j
x
j
=

j
x
j
, (124)
15
From (120), (40), (42) and (48) we nd
N

AN =
_
_
e
i
0 0
0 1 0
0 0 e
i
_
_
N

BN =
_

_
1+cos
2

sin

2
1cos
2
sin

2
cos
sin

2
1cos
2
sin

2
1+cos
2
_

_
N

CN =
_
_
e
i
0 0
0 1 0
0 0 e
i
_
_
(125)
or
N

ANN

BNN

CN = N

ABCN = N

XN = D
1
(, , ) (126)
where the Wigner D-function D
1
() denotes the 3 3 matrix,
D
1
(, , ) =
_

_
1+cos
2
e
i(+)

sin

2
e
i 1cos
2
e
i()
sin

2
e
i
cos
sin

2
e
i
1cos
2
e
i() sin

2
e
i 1+cos
2
e
i(+)
_

_
(127)
The row index and column index of the matrix element D
1

take on the values =


1, 0, 1 and = 1, 0, 1. Using (124) and (126) we can write the rotation dyadic X of (51)
as
X = U X U =

i
X
ij
N
i

D
1

. (128)
3.1 Wigner D-functions
It is convenient to describe the wave functions of polyatomic molecules as superpositions
of Wigner D-functions. The most important properties of the D-functions follow from the
fact that are irreducible representations of the rotation group of three-dimensional space.
To take advantage of this connection of D-functions to rotations, a special case of which
already encountered in (128), we consider a quantum-mechanical particle with integer spin
quantum number j. We can describe the spin either in the laboratory or the body-xed basis
discussed above, with spin operator J = x
i
J
i
= x
i

J
i
.The components of the spin operators,
referred to laboratory and body-xed basis vectors, are (2j +1) (2j +1) matrices with the
usual commutation relations
[J
i
, J
j
] = i
ijk
J
k
and [

J
i
,

J
j
] = i
ijk

J
k
. (129)
There is no anomalous sign for the commutator [

J
i
,

J
j
], as there was for the commutator
[

L
i
,

L
j
] for the orbital angular momentum operators of a rigid body in (104), because unlike
16
the orbital angular momentum operators of the rigid body, L
j
or

L
j
, the spin operators J
i
and

J
i
do not operate on the Euler angles .
We introduce spin basis functions |jm in the laboratory system with the familiar prop-
erties
J J|jm = j(j + 1)|jm,
J
3
|jm = m|jm,
J

|jm =

(j m)(j m + 1)|j, m1. (130)


The D-function[2] can be dened as
D
j
ml
(, , ) = jm|R|jl, (131)
where the rotation operator is
R = e
iJ
z
e
iJ
y
e
iJ
z
. (132)
The D-functions can be written as
D
j
ml
(, , ) = e
im
d
j
ml
()e
il
, where d
j
ml
() = jm|e
iJ
y
|jl. (133)
Completeness of the D-Functions. The D-functions D
j
ml
form a complete set for ex-
panding functions of the Euler angles . As shown by Varshalovich 4.10.(5) [2], they
have the orthogonality relations

2
0
d


0
sin d

2
0
dD
j
ml
D
j

=
8
2
[j]

jj

mm
. (134)
Unless otherwise noted in (134), and in subsequent expressions, D
j
ml
= D
j
ml
(, , ).
3.2 The Products L

D
j
ml
and

L

D
j
ml
Let a D-Function D
j
ml
() be dened by (131) and consider a subsequent rotation T

=
1 i J by the innitesimal angle . If applied after the rotation R, the innitesimal
rotation T

would produce a net rotation T

R = (1i J)R, parameterized by the Euler


angles + , + and + . Taking = D
j
ml
in (94) we nd
D
j
ml
( + , + , + ) = jm|(1 i J)R|jl = (1 + i L)D
j
ml
, (135)
or
jm|JR|jl = LD
j
ml
. (136)
Multiplication by L
j
. Multiplying both sides of (136) by x
3
we nd
jm|J
3
R|jl = mD
j
ml
= L
3
D
j
ml
. (137)
17
Noting from Varshalovich 4.4(4) [2] that
D
j
ml
= (1)
ml
D
j
m,l
, (138)
and setting m m and l l we nd that (137) gives
L
3
D
j
ml
= mD
j
ml
. (139)
Multiplying both sides of (136) by x

= x
1
ix
2
we nd
jm|J

R|jl = jm|J

R|jl =

(j m)(j m + 1)D
j
m1,l
= L

D
j
ml
. (140)
Setting m m and l l in (139) and using (137) we nd
L

D
j
ml
=

(j m)(j m + 1)D
j
m1,l
. (141)
In summary, the operators L
3
and L

have no eect on the superscript j or the second


subscript l of the D-Function D
j
ml
. As shown in (141), the raising/lowering operator L

raises and lowers the rst subscript m in by the same amount and with the same coecients
as for an angular momentum eigenfunction |jm. As shown in (138), multiplying D
j
ml
by the
operator L
3
is equivalent to multiplication by m, as though L
3
were multiplying an angular
momentum eigenfunction |jm.
Multiplication by

L
j
. We rotate the spin functions |jm through the Euler angles
with the rotation operator R of (132) to get rotated spin functions
|jl} = R|jl, or jm| = {jm|R. (142)
The rotated spin operators

J
j
, already mentioned in (129), can be written in two alternate
ways as

J
j
= RJ
j
R
1
= x
j
J. (143)
In analogy to (131) we have

J|jl} = j(j + 1)|jl},

J
3
|jl} = l|jl},

|jl} =

(j l)(j l + 1)|j, l 1}. (144)


Using (142) we can write (131) as
D
j
ml
= jm|R|jl = {jm|R|jl}. (145)
We can also use (142) to write (136) as,
{jm|RJ|jl} = LD
j
ml
. (146)
Multiplying (146) on the left by x
3
and using (145) and (144) we nd
{jm|R

J
3
|jl} = lD
j
ml
=

L
3
D
j
ml
. (147)
18
Making the substitution m m and l l in (147) and using (137) we nd

L
3
D
j
ml
= lD
j
ml
. (148)
Multiplying (146) on the left by x
+
= x
1
x
2
and using (145) and (144) we nd
{jm|R

J

|jl} =

(j l)(j l + 1)D
j
ml1
=

D
j
ml
. (149)
Making the substitution m m and l l in (149) and using (137) we nd

D
j
ml
=

(j l)(j l 1)D
j
ml1
. (150)
In summary, the operators

L
3
and

L

have no eect on the superscript j or the rst subscript


m of the D-Function D
j
ml
. As shown in (150), the operators

L

lower and raise the second


subscript l, the opposite of what one might expect. As shown in (148), multiplying D
j
ml
by
the operator

L
3
is equivalent to multiplication by l.
Multiplication by L L. Writing 2L L = L
+
L

+ L

L
+
+ 2L
2
z
and using (138) and
(141) we nd
L LD
j
ml
= j(j + 1)D
j
ml
. (151)
We nd the same expression (151) if we write 2L L =

L
+

+

L

L
+
+ 2

L
2
z
and use (148)
and (150).
3.3 About-face rotations
In addition to the angular momentum quantum numbers j and m, the energy eigenstates
for rotating rigid bodies have additional quantum numbers that are helpful in classifying the
states, and in determining which radiative transitions are allowed. The most important of
these symmetries are connected with how the eigenfunctions are aected by about face
rotations around body-xed axes x
k
. An about-face rotation about the body-xed axis x
k
can be described by the dyadics

Q
1
= + x
1
x
1
x
2
x
2
x
3
x
3
= U 2

S
1

S
1
,

Q
2
= x
1
x
1
+ x
2
x
2
x
3
x
3
= U 2

S
2

S
2
,

Q
2
= x
1
x
1
x
2
x
2
+ x
3
x
3
= U 2

S
3

S
3
. (152)
We see that the eect of the about-face rotations on body-xed unit vectors is

Q
k
x
j
=
_
x
j
if j = k
x
j
, if j = k.
(153)
Only two of the about-face rotations are independent, since one can verify from 152) that

Q
1
=

Q
2


Q
3
=

Q
3


Q
2
,

Q
2
=

Q
3


Q
1
=

Q
1


Q
3
,

Q
3
=

Q
1


Q
2
=

Q
2


Q
1
. (154)
19
We can use (70) (74) to write

Q
1
= exp(i

S
1
), (155)

Q
2
= exp(i

S
2
) = exp(iS

3
) exp(i

2
) exp(iS

3
), (156)

Q
3
= exp(i

S
3
) = exp(iS

3
). (157)
If the rotation X of (55) is parameterized by the Euler angles , , , the rotation

Q
l
X will
be parameterized by dierent Euler angles,
k
,
k
,
k
. Since two about-face rotation about
the same axis restore the body-xed unit vectors x
k
to their original values, we must have
(
k
)
k
= , (
k
)
k
= , and (
k
)
k
= . We can dene a wave function

Q
k
(, , ) that has
been subject to an about-face transformation as

Q
k
(
k
,
k
,
k
) = (, , ) or

Q
k
(, , ) = (
k
,
k
,
k
) (158)
Suppose that is an eigenvalue of

Q
k
with eigenvalue
k
so that

Q
k
=
k
. (159)
Multiplying both sides of (159) by

Q
k
and noting that

Q
2
k
= 1 we nd that we must have

2
k
= 1 so the possible eigenvalues of the about-face rotation are

k
= 1. (160)
Taking k = 3 we nd from (55) and (157),

Q
3
X = exp(i[ + ]

3
) exp(iS

2
) exp(iS
3
), (161)
from which we see that the about-face Euler angles are
_
_

3
_
_
=
_
_

+
_
_
. (162)
Taking k = 2 and noting that S

2
= S

2
we can use (55) and (156) to nd

Q
2
X = exp(i

S
2
) exp(i

3
) exp(iS

2
) exp(iS
3
)
= exp(i

3
) exp(iS

2
) exp(iS

2
) exp(iS
3
)
= exp(i

3
) exp(i[ + ]S

2
) exp(iS
3
)
= exp(i[ ]

3
) exp(i[ ]S

2
) exp(i[ + ]S
3
) (163)
If the angle is in the required range 0 < , the angle + that occurs in the third
line of (163), will exceed the limit and must be replaced by a rotation of
2
= about
the axis x
1
, which is parameterized (modulo 2) by the Euler angle
2
= + . In order
to have the axes x
2
continue to point in the initial direction, as it must for an about face
rotation about x
2
, the third Euler angle must be
2
= . In summary, the about-face
Euler angles are
_
_

2
_
_
=
_
_
+


_
_
. (164)
20
Because

Q
1
=

Q
2


Q
3
, we must have
_
_

1
_
_
=
_
_
(
3
)
2
(
3
)
2
(
3
)
2
_
_
=
_
_
+

_
_
. (165)
Using the about-face angles (162), (164) and (165) in (158), with = D
j
ml
we nd

Q
1
D
j
ml
= (1)
j
D
j
m,l

Q
2
D
j
ml
= (1)
j+l
D
j
m,l

Q
3
D
j
ml
= (1)
l
D
j
ml
. (166)
Here we used the identity (see formula 4.4. (1) of reference[2]),
d
j
ml
( ) = (1)
j+m
d
j
ml
(). (167)
The about-face quantum number of D
j
ml
are therefore

1
= (1)
j

2
= (1)
j+l

3
= (1)
l
. (168)
3.4 Rotations of

L
j
In analogy to (??), if we rotate the operator

L
1
by an angle around the axis x
3
we nd a
rotated version Y
1
= Y
1
() of the operator, which we denote by
Y
1
= e
i

L
3
L
1
e
i

L
3
. (169)
Dierentiating Y
1
with respect to and using the commutation relation [

L
3
,

L
1
] = i

L
2
from (104) we nd
d
d
Y
1
= Y
2
(170)
where
Y
2
= e
i

L
3
L
2
e
i

L
3
. (171)
Dierentiating (171) again we nd in analogy to (170)
d
d
Y
2
= Y
1
(172)
Dierentiating (170) once again and using (172), we nd the simple dierential equation for
Y
1
d
2
d
2
Y
1
= Y
1
, (173)
21
which has the general solution
Y
1
= Acos + Bsin , and Y
2
=
d
d
Y
1
= Asin Bcos , (174)
Setting = 0 in (169), (171) and equating to the corresponding values from (174) we nd
A =

L
1
, and B =

L
2
, (175)
and therefore rotated versions of

L
1
and

L
2
are
Y
1
= e
i

L
3
L
1
e
i

L
3
=

L
1
cos

L
2
sin
Y
2
= e
i

L
3
L
2
e
i

L
3
=

L
1
sin +

L
2
cos . (176)
For the special case of an about-face rotation by = we nd from (176)

Q
3

L
1

Q

3
=

L
1
,

Q
3

L
2

Q

3
=

L
2
,

Q
3

L
3

Q

3
=

L
3
.
(177)
There are identities analogous to (177) for about turns around the axes x
1
and x
2
.
4 Rotational Energies
In molecular spectroscopy for asymmetric-top molecules it is customary to denote three
principal moments of inertia, I
a
, I
b
and I
c
, where the indices abc are some permutation of
123. The corresponding body xed axes by x
a
, x
b
and x
c
. In the customary wave-number
units of molecular spectroscopy, the Hamiltonian (8)

H =
H
hc
= A

L
2
a
+ B

L
2
b
+ C

L
2
c
. (178)
The rotational coecients are related to the principal moments of inertia by
A =

4cI
a
, B =

4cI
b
, C =

4cI
c
. (179)
Following a common convention, we usually will order the moments as
I
a
I
b
I
c
, or A B C. (180)
If we let

L
a
=

L
3
,

L
b
=

L
1
and

L
c
=

L
2
we can write (178) as

H = B

L
2
1
+ C

L
2
2
+ A

L
2
3
=
B + C
2
L L +
2A B C
2

L
2
3
+
C B
4
(

L
2
+
+

L
2

). (181)
with

L

=

L
1
i

L
2
.
22
4.1 Quantum numbers
From inspection of (178) we see that we can select four independent operators that commute
with the Hamiltonian and with each other,
[L L,

H] = 0, [L
3
,

H] = 0, [

C
1
,

H] = 0, [

C
2
,

H] = 0. (182)
We can therefore choose the molecular wave function to be the simultaneous eigenfunction
of

H, L L, L
3
,

C
1
and

C
1
. We can use the eigenvalues of these ve operators to label the
wave functions and energies and write

H
{

E
1

2
}
jm
=

E
{

E
1

3
}
jm
,
L L
{

E
1

2
}
jm
= j(j + 1)
{

E
1

2
}
jm
,
L
3

{

E
1

2
}
jm
= m
{E
1

2
}
jm
,

Q
1

{

E
1

2
}
jm
=
1

{

E
1

2
}
jm
,

Q
2

{

E
1

2
}
jm
=
2

{

E
1

2
}
jm
. (183)
Parity For planar, three-atom molecules like H
2
O and CO
2
, we will assume that the prin-
cipal axis x
1
is normal to the plane of the molecule. In this case, the about-face rotation

Q
1
is equivalent to the parity operation, P, which reects the coordinates of all particles
through the origin of the coordinate system, the center of mass in our case, so
P =

Q
1
. (184)
Then the parity quantum number p is the same as the about-face quantum number
1
,
p =
1
= 1. (185)
4.2 Schroedinger equation
To simplify notation we will use a single index k to label symmetric top wave functions which
we write as
|k =

2j + 1
8
2
D
j
mk
, for k = j, j 1, . . . , j. (186)
We can use (148) and (150) to write

L
3
|k = k|k, (187)

|k =

(j k)(j k + 1)|k 1, (188)

L

L|k = j(j + 1)|k. (189)
We have omitted the quantum numbers j and m from the labels of the symmetric top basis
vectors (186) since the operators

L
3
and

L

, from which the asymmetric top Hamiltonian


(183) is constructed, cannot change the quantum numbers j or m when applied to one of
the basis states (186).
23
We write the asymmetric top wave function |
n
as a superposition of the basis states
(186),
|
n
=

k
|kv
kn
(190)
Then the Shroedinger equation becomes
(

H

E
n
)|
n
= 0, or

k
(

H
ik


E
n

ik
)v
kn
= 0. (191)
The matrix elements Hamiltonian between the symmetric-top basis functions (186) are

H
ik
= i|

H|k =
2j + 1
8
2

D
j
mi

HD
j
mk
dsin dd. (192)
Using (187)(189), we nd that the only non-zero matrix elements of the Hamiltonian (181)
are
k|

H|k =
1
2
_
(B + C)j(j + 1) + (2A B C)k
2

, (193)
k + 2|

H|k =
(C B)
4

(j k)(j k 1)(j + k + 1)(j + k + 2), (194)


k 2|

H|k =
(C B)
4

(j + k)(j + k 1)(j k + 1)(j k + 2). (195)


Except for small rotational quantum numbers j, where explicit expressions for eigenfunctions
and energies can be found, it is necessary to numerically diagonalize the (2j + 1) (2j + 1)
matrix with nonzero elements given by (195 to nd the eigenfunctions |
n
(the column
vectors v
kn
) and energies

E
n
of (191).
4.3 Asymmetric top with j = 1
Some insight can be gained by considering the simple asymmetric top states for angular-
momentum quantum numbers j = 1. The basis states |k of (186) become
|1 =
_
_
1
0
0
_
_

3
8
2
D
1
m1
,
|0 =
_
_
0
1
0
_
_

3
8
2
D
1
m0
,
| 1 =
_
_
0
0
1
_
_

3
8
2
D
1
m1
(196)
Using the bases (196) we can represent the operators of (187)(189) as the 3 3 matrices
24
L L =
_
_
2 0 0
0 2 0
0 0 2
_
_
,

L
2
3
=
_
_
1 0 0
0 0 0
0 0 1
_
_

L
2
+
=
_
_
0 0 2
0 0 0
0 0 0
_
_
,

L
2

=
_
_
0 0 0
0 0 0
2 0 0
_
_
. (197)
Using (197) we nd a 3 3 matrix representation of the Hamiltonian (192),

H =
1
2
_
_
2A + B + C 0 C B
0 2B + 2C 0
C B 0 2A + B + C
_
_
. (198)
From inspection of (198) we see that the Hamiltonian has three eigenvectors, |
n
,
|
x
=
1

2
_
_
1
0
1
_
_
, |
y
=
1

2
_
_
1
0
1
_
_
, |
z
=
_
_
0
1
0
_
_
. (199)
with energies

E
x
= A + C, E
y
= A + B, E
z
= B + C. (200)
If the three energies for j = 1 have been measured, we can determine the coupling coecients,
A, B, C from the inverse of (200),
2A = +E
x
+ E
y
E
z
2B = E
x
+ E
y
+ E
z
2C = +E
x
E
y
+ E
z
. (201)
The about-face rotations of (152) can be represented with the matrices

Q
1
=
_
_
0 0 1
0 1 0
1 0 0
_
_
,

Q
2
=
_
_
0 0 1
0 1 0
1 0 0
_
_
,

Q
3
=
_
_
1 0 0
0 1 0
0 0 1
_
_
. (202)
Using the signs of the eigenvalues,
1
,
2
,
3
for

Q
1
,

Q
2
,

Q
3
to denote the about-face symme-
tries of the eigenstates (199), we nd that they have the symmetries

3
= +, +, +. (203)
The parity symmetries are
p =
1
= , +, . (204)
Two of the states with j = 1 have negative parity and one has positive parity.
25
4.4 State labels
[BEING REVISED.] For asymmetric tops (but not for symmetric tops), the energies

E =

E
n
of (191) are non-degenerate, so we can use the 2j +1, distinct but non-integer values of the
energies

E, together with the intgegers j and m as state labels, and we can denote the states
simply as
{

E}
jm
. As discussed above, for j = 1 it is not necessary to use

E as a state label
since it is possible to use integer about-turn quantum numbers
1
and
2
together with j and
m to uniquely label the states as
{
1

2
}
jm
. Since there are only 3 dierent energies for j = 1,
the four possible values of {
1
,
2
} are more than enough to label the states. An alternate
labeling of the states of the previous section is
x
=
{+}
jm
,
y
=
{+}
jm
, and
z
=
{}
jm
.
For states with j 2 where 2j +1 5, the four possible combinations of {
1
,
2
} do not
provide enough labels to replace the energy

E as a quantum number, and there will be two
or more independent states with the same quantum numbers jm
1

2
. However it is possible
to uniquely label the states of an asymmetric top with quantum numbers j and m with a
pair of integers, k
a
and k
c
instead of the rather uninformative energy quantum number

E.
The 2j + 1 allowed pairs {k
a
k
c
} are exactly the number of labels required. The pair labels
also also uniquely dene the state in the symmetric-top limits, when there are only j + 1
values of

E, not enough to provide 2j + 1 distinct labels. For symmetric tops, all but one
of the states occur as parity doublets, with the same quantum numbers jm

E but with
opposite parity quantum numbers, p =
1
= 1.
To determine the labels k
a
and k
c
of a state
{k
a
k
c
}
jm
of an asymmetric top, we imagine let-
ting the eigenfunction
{k
a
k
c
}
jm
of the Hamiltonian (178) adiabatically change as the coupling
coecient B approaches either the smallest rotational coecient C or the largest rotational
coecient A and the Hamiltonians become,

H = CL L + (A C)

L
2
a
, the prolate limit, A > B = C, (205)

H = AL L (A C)

L
2
c
, the oblate limit, C < B = A. (206)
Prolate limit. For the prolate limit (205) we see that we have [

L
2
a
,

H] = 0, so energy
eigenfunctions can be chosen to be eigenfunctions

L
2
a
. We dene the rst state index k
a
as
the positive square root of the eigenvalue k
2
a
of

L
2
a
, as dened by the eigenvalue equation

L
2
a

{k
a
k
c
}
jm
= k
2
a

{k
a
k
c
}
jm
for the prolate limit, A > B = C. (207)
For the prolate limit, the energies Cj(j + 1) + (A C)k
2
a
with k
2
a
= 0, 1, 4, 9, . . . j
2
. Except
for k
a
= 0, the states
{k
a
k
c
}
jm
will be two-fold degenerate in the prolate limit, since there are
two basis functions with the same value of k
2
a
, namely D
j
m,k
a
. If we were to arrange the
energies

E as a column vector, with the highest energy at the top of the column and the
26
lowest at the bottom, the corresponding values of the indices k
a
would be
{k
a
} =
_

_
j
j
j 1
j 1
.
.
.
1
1
0
_

_
. (208)
Oblate limit. For the oblate limit (205) we see that we have [

L
2
c
,

H] = 0, so energy
eigenfunctions can be chosen to be eigenfunctions

L
2
c
. We dene the second state index k
c
as the positive square root of the eigenvalue k
2
c
of

L
2
c
, as dened by the eigenvalue equation

L
2
c

{k
a
k
c
}
jm
= k
2
c

{k
a
k
c
}
jm
for the oblate limit, C < B = A. (209)
For the oblate limit, the energies Aj(j +1) (AC)k
2
c
decrease with increasing values of
k
2
c
= 0, 1, 4, 9, . . . j
2
. Except for k
c
= 0, the states
{k
a
k
c
}
jm
will be two-fold degenerate, since
there are two independent basis functions with the same value of k
2
c
, namely D
j
m,k
a
. If we
were to arrange the energies

E as a column vector, with the highest energy at the top of the
column and the lowest at the bottom, the corresponding values of the indices k
c
would be
{k
c
} =
_

_
0
1
1
2
2
.
.
.
j 1
j 1
j
j
_

_
. (210)
The allowed label pairs, with labels higher in the column corresponding to higher energies,
are therefore
{k
a
k
c
} =
_

_
j 0
j 1
j 1 1
j 1 2
.
.
.
.
.
.
1 j 1
1 j
0 j
_

_
. (211)
27
From (211) we see that
k
a
+ k
c
= j, or k
a
+ k
c
= j + 1. (212)
In the prolate limit and for k
a
> 0 we can choose two linear combination of the basis functions
D
j
m,k
a
that are eigenfunction of the parity operator P =

C
1
,

{k
a
k
c
}
jm
=

2j + 1
16
2
_
D
j
mk
a
+
1
(1)
j
D
j
m,k
a
_
. (213)
Using (??), (??) and (??) we see that the wave functions (213) are simultaneous eigenfunc-
tions of the about-face operators,

C
j
,

C
j

{k
a
k
c
}
jm
=
j

{k
a
k
c
}
jm
, (214)
where the quantum numbers
j
are
_
_

3
_
_
=
_
_

1
(1)
k
a

1
(1)
k
a
_
_
. (215)
If we gradually increase B from the prolate limit B = C to the oblate limit B = A the
wave function (212) will adiabatically evolve into a more complicated linear combination
of D-functions, but it will retain the same eigenvalue
j
of (214). In the oblate limit with
B = A the symmetry axis will be x
1
= x
c
. In analogy to (214) where the symmetry axis
was x
3
= x
a
, the about-turn quantum numbers must be
_
_

3
_
_
=
_
_
(1)
k
c

3
_
_
. (216)
Comparing (216) with (214) we see that the important quantum numbers for selection rules
can be written in terms of the pair indices, k
a
and k
c
, as
_
_

3
_
_
=
_
_
(1)
k
c
(1)
k
a
+k
c
(1)
k
a
_
_
. (217)
For the special case of k
a
= 0 we can write

{k
a
k
c
}
jm
=
{0k
c
}
jm
=

[j]
8
2
D
j
m0
. (218)
One can readily verify that the relations (213) and (217) remain valid for (218).
References
[1] L. Landau and E. Lifschitz, Quantum Mechanics (Nonrelativistic Theory), Addison-
Wesley, Reading, Mass. 1965.
[2] D. A. Varshalovich, A. N. Moskalev and V. K. Khersonskii, Quantum Theory of Angular
Momentum (World Scientic, Singapore, 1988).
[3] G. Herzberg, Infrared and Raman Spectra Van Norstrand, (Princeton, NJ, 1945).
28

Vous aimerez peut-être aussi