19 October 2001

Chemical Physics Letters 347 (2001) 149±156

www.elsevier.com/locate/cplett

The electronic adiabatic-to-diabatic transformation

matrix and the Wigner rotation matrix
Michael Baer *
Department of Applied Physics, Soreq NRC, Yavne 81800, Israel
Received 12 July 2001; in ®nal form 20 August 2001

Abstract
In this publication we consider electronically multi-fold degeneracy with the aim of revealing the connection between
the adiabatic-to-diabatic transformation (ADT) matrices and Wigner's irreducible representation of the rotation group.
To form the connection we constructed simpli®ed models of two-, three- and four-states all (electronically) degenerate
at a single point and employed the relevant non-adiabatic coupling matrices (NACMs). We found that once these
matrices were properly quantized [Chem. Phys. 259, (2000) 123] the ADT matrices and Wigner's d j -rotation matrices are
related via a similarity transformation. Ó 2001 Elsevier Science B.V. All rights reserved.

1. Introduction that the NACTs are usually singular (in addition

to being vectors) calls for more mathematical ori-
During the last few years major e€orts were ented approaches in order to understand their role
made to understand the features of the non-adia- in molecular physics. An important methodology
batic coupling terms (NACTs) and their role in the in this respect is to assume ad hoc models and to
theory of molecular physics [1±9]. The NACTs look for the features that the NACTs have to
owe their existence to the Born±Oppenheimer possess in order to be applicable for molecular
(BO) assumption which says that the fast moving systems [17±24]. During the last decade we fol-
electrons can be treated separately from the slowly lowed both courses but our main interest was
moving nuclei [10,11]. pointed towards the physical±mathematical fea-
The ordinary way to get acquainted with ob- tures of the NACTs [21,22,24,25]. In this process
jects like the NACTs is to derive them from ®rst we revealed the necessity of being able to form
principles, via ab initio calculations [12±16], and sub-Hilbert spaces (SHSs) [22] in the given region
probe their spatial structure somewhat reminiscent of interest in con®guration space (CS) and the fact
of the way potential energy surfaces (PESs) are that the non-adiabatic coupling matrix (NACM)
studied. However, this way is not enough. The fact has to be quantized for this SHS if they are sup-
posed to be of physical signi®cance [22,24]. This
quantization requirement enhanced the recognition
* that the NACMs are closely related, much more
Present address: Department of Chemistry, H.C. Orsted
Institute, University of Copenhagen, 2100 Copenhagen O,
than anticipated, to the angular momentum op-
Denmark. erators. In the present Letter this approach is
E-mail address: mmbaer@netvision.net.il (M. Baer). pursued. In particular we show, for special cases,
0009-2614/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 2 6 1 4 ( 0 1 ) 0 0 9 7 2 - 1
150 M. Baer / Chemical Physics Letters 347 (2001) 149±156

that the adiabatic-to-diabatic transformation 2. Theoretical background

(ADT) matrix is closely related to Wigner's d j -
rotation matrices which form the irreducible rep- In their treatment of the mixed systems of nu-
resentation of the rotation group [26,27]. clei and electrons BO derived the Schr odinger
The model we consider for this purpose is a case equation (SE) for the nuclei which in present day
of N surfaces all degenerate at one single point in notation can be written as [22,25,28,29]
CS where the NACM, s…s†, is of the following
form [22]: 1 2
…r ‡ s† W ‡ …u E†W ˆ 0; …5†
2m
s…s† ˆ gt…s†: …1†
where r is the usual (mass-scaled) gradient opera-
Here g is an N  N anti-symmetric constant ma- tor, E is the total energy, W is a column matrix which
trix, t…s† is a vector and s is a point in CS. In contains the nuclear functions fwi ; i ˆ 1; 2; . . .g, u is
what follows we consider integrals along con- a diagonal matrix which contains the adiabatic po-
tours. In particular we shall be interested in the tentials and s is a matrix which contains the above-
integral: mentioned NACTs:
Z s sij ˆ hfj jrfi i: …6†
c…s† ˆ ds0
t…s0 † …2† This derivation holds for a complete Hilbert space
0
but it can be shown to hold also for an SHS of
carried out along C where both points, i.e., s ˆ 0 ®nite dimension N if certain conditions are ful®led
and s itself are on C. Here ds is a vectorial di€er- [22,25].
ential length along the contour and the dot stands From Eq. (5) it is noticed that s, like the grad
for a scalar product. Next we introduce the angle a operator itself, is a momentum operator. With
de®ned as the value of the above integral for a respect to this, a relevant question is: What
closed contour C. Thus: happens if one integrates over s along a closed
I contour C? In other words, is there any demand
aˆ ds
t…s†: …3† to be associated with the following di€erential
C dK:
In what follows we shall be interested only in dK ˆ ds
s? …7†
contours that surround the above-mentioned point
of degeneracy. Next, it is assumed that for each The answer to this question is, as was found out
dimension N the g-matrix elements and the vector recently [17,18,20,25], connected with the diabatic
t…s† are arranged in such a way as to guarantee t…s† representation. Diabatic potentials can be formed
is normalized as follows: in various ways but the more straightforward way
is by employing a set of electronic eigenstates as
a…C† ˆ 2p: …4†
calculated at one single point in CS (in this sense it
In the present Letter we derive conditions to be can be said that diabatic potentials are a result of
ful®led by the g-matrix elements in order for the the BO treatment). From the way these potentials
s…s†-matrix to be a matrix of physical meaning. are calculated it can be seen that the diabatic po-
This we do in three steps: (a) For the general tentials, just like adiabatic potentials, are single-
case. (b) For the s…s†-matrix de®ned in Eq. (1). (c) valued in CS. Having single-valued potentials is an
For the three special cases, namely: the two-state essential requirement because it is not possible to
case …N ˆ 2†, the tri-case …N ˆ 3† and the tetra- solve the relevant (nuclei) SE unless the potentials
case …N ˆ 4†. Whereas the ®rst two cases were are single-valued.
already worked out by us on various occasions Next we brie¯y describe a way of deriving the
[21,22] the third case is not only new but presents adiabatic framework which is based on eliminating
also novel features not encountered in ®rst two the s-matrix from Eq. (5). This is achieved by
cases. employing the ADT matrix A, which is derived as
 M. Baer / Chemical Physics Letters 347 (2001) 149±156 151

a solution of the following ®rst-order di€erential  I 

equation [25]: D ˆ } exp ds
s : …13†
C
rA ‡ sA ˆ 0: …8†
It can be proved [20] that the necessary condition
Once A is calculated, the diabatic potential matrix for having single-valued diabatic potentials is that
W follows from the expression: the D-matrix ful®ls the following commutation
relation:
W ˆ Ay uA; …9†
y
where A is the complex conjugate matrix of A. It ‰D; u…s†Š ˆ 0 …14†
is important to emphasize that to derive the dia-
batic potentials this way is as valid as deriving at every point in the region of interest. Since u…s† is
them employing the `direct' way. This does not a diagonal matrix this condition can be satis®ed if
mean that the two diabatic matrices are necessarily and only if the D-matrix is diagonal. Moreover,
identical, but it means that under certain circum- since the D-matrix, just like the A-matrix, is an
stances they have to be related by a (constant) orthogonal matrix the diagonal numbers have to
unitary transformation. be of norm 1, namely, numbers that are either
Next we brie¯y discuss the conditions for Eq. …‡1†s or … 1†s. In this sense Eq. (13) is a quanti-
(8) to have a solution. Being a ®rst-order di€er- zation condition. Thus it is indeed a quantization
ential equation, Eq. (8) has to be solved along a related to the (di€erential) expression in Eq. (7)
contour. The condition for the existence of a so- but in a more complicated way. In particular the
lution along this contour is ful®lment of the `Curl' quantized expression is solely dependent on the s-
condition [22,24,25]. Thus if p and q are any two matrix
R and is related to a contour constructed by
(Cartesian) coordinates then the `Curl' condition C
ds.
implies:
o o 3. The treatment of the model non-adiabatic cou-
sq sp ˆ ‰sq ; sp Š: …10†
op oq pling matrix
This equation can also be written more compactly
as: For the model NACM as presented in Eq. (1)
the D-matrix in Eq. (13) can be written as:
Curl s ˆ ‰s  sŠ: …11†
D ˆ exp… ga†; …15†
Let us now return to Eqs. (8) and (9). Assuming
the `Curl' condition is ful®led along C then Eq. (8) where a is de®ned in Eq. (3) or due to the nor-
has a solution, which can be presented in the fol- malization condition imposed on t…s† it can also be
lowing way [30,31] presented as:
0
 Z  D ˆ exp… 2gp†: …15 †
s
A…s; s0 † ˆ } exp ds
s ; …12† Since g is a full matrix we still have to evaluate the
s0
exponential function. If G is the matrix that di-
agonalizes g and if kk ; k ˆ 1; . . . ; N are the corre-
where s and s0 are two points on C and the symbol
sponding eigenvalues, then the D-matrix can be
} is introduced to indicate that this integral has to
presented as:
be carried out in a given order. In other words, } is
a path ordering operator. In presenting A…s; s0 † in
D ˆ GE…a†Gy ˆ GE…2p†Gy ; …16†
this way it is assumed that the boundary condition
for the solution is the unit matrix.
where the matrix E is a diagonal matrix with the
Next we introduce a matrix D … A…s0 ; s0 ††,
elements:
hence termed the topological matrix, de®ned as
[20±24] Ejk ˆ djk exp… kj a† ˆ djk exp… 2pkj †: …17†
152 M. Baer / Chemical Physics Letters 347 (2001) 149±156

It might be important to remind the reader that the  

cos…2pg1 † sin…2pg1 †
ADT matrix de®ned in Eq. (12) can be written in a Dˆ : …22†
sin…2pg1 † cos…2pg1 †
similar form like Eq. (15) namely:
Since D has to be a diagonal matrix the allowed
A ˆ exp‰ gc…s†Š …18†
values for g1 are:
or also
g1 ˆ n=2; …23†
0
A ˆ GE…c…s††Gy ; …18 †
where n is an integer. In case n is an odd integer we
where the E-matrix is similar to the one de®ned in encounter the Jahn±Teller model [32,33] and in
Eq. (17) except that c replaces a: case it is an even integer we have the Renner±
Ejk ˆ djk exp… kj c…s††: …19† Teller model [34,35].
From Eqs. (18), (19) and (23) we get for the
ADT matrix:
 
cos…c=2† sin…c=2†
4. The treatment of special cases Aˆ ; …24†
sin…c=2† cos…c=2†
For reasons of convenience and also for the where c is given in Eq. (2) and is de®ned in the
sake of showing the similarities between our vari- ‰0; 2pŠ interval.
ous ADT matrices and the Wigner's d j -matrices
we assume the g-matrix to have non-zero elements 4.2. The trifold degeneracy case
along the two tri-diagonals only (this comment is
not relevant for the two-state case). Thus we as- The g-matrix is given in the form:
sume for g the following structure: 0 1
0 g1 0
gjk ˆ djk 1 gj djk‡1 gk : …20† g ˆ @ g1 0 g2 A : …25†
We shall refer to three special cases, namely, 0 g2 0
the two-state degeneracy the tri-state degeneracy For this matrix we get the following eigenvalues:
and the tetra-state degeneracy. The ®rst two q
cases were already discussed before and will be k1;2 ˆ ix; k3 ˆ 0; where x ˆ g12 ‡ g22
mentioned here just for the sake of completeness
…26a†
but we will elaborate more on the tetra-surface
case …N ˆ 4† which is presented here for the ®rst and the corresponding matrix, G, that diagonalizes
time. it:
0 p 1
4.1. The two-state degeneracy 1 @ g1 g1 g2 2
G ˆ p ix ix 0p A: …26b†
x 2 g2 g2 g1 2
The g-matrix in this case is given in the form:
  Substituting Eq. (26b) in Eq. (16) where in Eq. (17)
0 g1
gˆ : …21a† the three k's are replaced by Eq. (26a) yields the
g1 0
following D-matrix:
The matrix G that diagonalizes g is: 0 2 1
  g2 ‡ g12 C g1 xS g1 g2 …1 C†
g1 1 1 D ˆ x 2@ g1 xS x2 C g2 xS A;
G ˆ p …21b†
2 i i g1 g2 …1 C† g2 xS g1 ‡ g22 C
2

and the corresponding eigenvalues are k1;2 ˆ ig1 . …27†

Substituting Eq. (21b) in to Eq. (16) where in Eq.
where
(17) the two k's are replaced by g1 yields the
following D-matrix: C ˆ cos…2px† and S ˆ sin…2px†: …28†
 M. Baer / Chemical Physics Letters 347 (2001) 149±156 153

Since the D-matrix has to be diagonal this re- and

quirement can be ful®led if and only if [20,21] q
xˆ …g12 ‡ g22 ‡ g32 †: …36†
x ˆ n; …29†
where n is an integer (see also Eq. (26a)). Substi- It is seen that both p and q are real functions and
tuting Eqs. (28) and (29) in Eq. (27) shows that the that p > q.
diagonal of the D-matrix contains only (+1)s, The corresponding G-matrix that diagonalizes
which implies that the D-matrix is, in fact, a unit the matrix s is:
0 1
matrix. kq r kq r kp r kp r
From Eqs. (18), (19) and (29) we get for the 1 B ipkq ipkq iqkp iqkp C
ADT matrix (assuming n ˆ 1): G ˆ p B @
C;
A
r 2 k p r k p r k q r k qr

0 1 iqkp iqkp ipkq ipkq

g22 ‡ g12 C…c† g1 S…c† g1 g2 …1 C…c†† …34b†
B C
A…c† ˆ @ g1 S…c† C…c† g2 S…c† A; where r stands for g1 and kp and kq are de®ned as:
g1 g2 …1 C…c†† g2 S…c† g12 ‡ g22 C…c† s s
p 2 r2 r2 q2
…30† kp ˆ ; kq ˆ : …37†
p 2 q 2 p 2 q2
where we recall that:
q Substituting Eq. (34b) in Eq. (16) where, in Eq.
g12 ‡ g22 ˆ 1 …31† (17), the four k's are replaced by Eqs. (34a), yields
the following D-matrix:
and D11 ˆ k2q Cp ‡ k2p Cq ; D12 ˆ …pk2q Sp ‡ qk2p Sq †=r;
C…c† ˆ cos…c† and S…c† ˆ sin…c†: …32† D13 ˆ kp kq …Cp Cq †; D14 ˆ kp kq … qSp ‡ pSq †=r;
Here c is given in Eq. (2), where we recall that it is D22 ˆ …p2 k2q Cp ‡ q2 k2p Cq †=r2 ; D23 ˆ kp kq …pSp qSq †=r;
de®ned along the ‰0; 2pŠ interval. D24 ˆ pqkp kq …Cp 2
Cq †=r ; D33 ˆ k2p Cp ‡ k2q Cq ;
D34 ˆ …qk2p Sp ‡ pk2q Sq †=r; D44 ˆ …q2 k2p Cp ‡ p2 k2q Cq †=r2 ;
4.3. The tetrafold degeneracy case D21 ˆ D12 ; D31 ˆ D13 ; D32 …c† ˆ D23 ;
D41 ˆ D14 ; D42 ˆ D24 ; D43 ˆ D34 ;
The g-matrix for this case is: …38†
0 1
0 g1 0 0 where
B g1 0 g2 0C
gˆB
@ 0
C: …33† Cp ˆ cos…2pp†; Sp ˆ sin…2pp†; …39a†
g2 0 g3 A
0 0 g3 0 Cq ˆ cos…2pq†; Sq ˆ sin…2pq†: …39b†

For this matrix we get the corresponding eigen- The conditions for the D-matrix to become diag-
values: onal is that p and q ful®l the following conditions:
pˆn …40a†
k1;2 ˆ ip;
…34a† and
k3;4 ˆ iq;
qˆ` …40b†
where p and q are:
 where n (>1) and `, de®ned in the range n > ` P 0,
q1=2
1 are allowed to be either integers or half integers
p ˆ p x2 ‡ x4 4…g1 g3 †2 ;
2 but m ˆ n ` can only attain integer values. The
 q1=2 …35† di€erence between the case where n and ` are in-
1 2 tegers and the case where both are half integers is
q ˆ p x2 x4 4…g1 g3 † ;
2 as follows: Going through the expressions in Eq.
154 M. Baer / Chemical Physics Letters 347 (2001) 149±156

(38) it is noticed that in the ®rst case all diagonal i…m0 a‡mc† j
Djm0 m …h† ˆ e dm0 m …b†: …43†
elements of D are …‡1† (so that, D is, in fact, the
unit matrix) and in the second case we get that all It is noticed that if we are interested in ®nding a
four diagonal elements are ( 1) (so that, D is, in relation between the ADT matrix and Wigner's
fact, the minus unit matrix). rotation matrices we should concentrate on the d j -
Since p and q are directly related to the matrix. However before going into a detailed
gj ; j ˆ 1; 2; 3 (see Eqs. (35) and (36)) the two comparison between the two types of matrices it
conditions in Eqs. (40a) and (40b) imply `quanti- could be of interest to consider the Jy -matrix which
zation' conditions for the values of the g-matrix is responsible for the formation of the d j -matrix.
elements. Employing Eqs (2.18) and (2.28) of Ref. [27] it can
It is interesting to note that this is the ®rst time be shown that:
that in the present framework the quantization is 1 p
formed by two quantum numbers: a quantum hjmjJy jjm ‡ ki ˆ d1k …j ‡ m ‡ 1†…j m†;
2i
number n to be termed the principal quantum
…44a†
number and a quantum number ` to be termed the
1 p 
secondary quantum number. This case is reminis- hjm ‡ kjJy jjmi ˆ d1k …j m ‡ 1†…j ‡ m†:
cent of the two quantum numbers that character- 2i
ize the hydrogen atom or the three-dimensional …44b†
harmonic oscillator. De®ning now J~y as:
To obtain the ADT we employ again Eqs. (38)
where p and q are replaced by n and `, respectively, J~y ˆ iJy …45†
and Cp ; Sp ; Cq and Sq in Eqs. (40a) and (40b) are it is seen that the J~y -matrix is an anti-symmetric
replaced by: matrix just like the g-NACM (compare Eq. (20)
Cp ˆ cos…cp†; Sp ˆ sin…cp†; …41a† with those of Eqs. (44a),(44b) and (45)). Moreover
Cq ˆ cos…cq†; Sq ˆ sin…cq†; …41b† since the d j -matrix can be de®ned as:

respectively. Here, as in the previous cases, c is d j …b† ˆ exp… ibJy † ˆ exp…bJ~y † …46†
de®ned in Eq. (2) and attains values in the interval it is expected that the d j -matrices belong to the
‰0; 2pŠ. same group as the ADT matrices if the angle b is
identi®ed with the phase c (see Eq. (18)). Next we
inspect the three above-studied cases:
5. The adiabatic-to-diabatic transformation matrix (1) For the two-state case (i.e., j ˆ 1=2 ), the J~y -
and Wigner's rotation matrix matrix is of the form:
 
1 0 1
The explicit elements of Wigner's three-dimen- J~y ˆ …47†
sional rotation matrix Dj …h† are written as [26] 2 1 0
thus the value of the corresponding g1 in Eq. (21a)
Djm0 m …h† ˆ hjm0 jR…k; h†jjmi
is (1/2) which is its value if in Eq. (23) we assume
i…m0 a‡mc†
ˆe hjm0 je ibJy
jjmi; …42† n ˆ 1 (i.e., the Jahn±Teller case).
(2) For the tri-state case (i.e., j ˆ 1), the J~y -
where R…k; h† is the ordinary angular momentum
matrix is of the form:
operator in the limit h ! 0; m and m0 are the 0 p 1
components of the total angular momentum op- 0 2 0
erator J along the z and the z0 axes, respectively, 1 B p p C
J~y ˆ @ 2 0 2 A: …48†
a; b and c are the corresponding three Euler an- 2 p
0 2 0
gles and jjmi is an eigenfunction of the Hamil-
tonian, of J 2 and of Jz . As a result Eq. (42) will Comparing Eqs. (48) and (25) it is noticed that the
be written as: J~y -matrix and the g-matrix become identical when
 M. Baer / Chemical Physics Letters 347 (2001) 149±156 155

p can expose realistic situations which may exist but

g1 ˆ g2 ˆ 2=2 ) x ˆ 1: …49†
could not be predicted otherwise.
From Eq. (29) it is noticed that the integer n has to We know already that the NACMs have to be
be equal to 1. For this case it is easy to see that quantized in order to form single-valued diabatic
A…c† in Eq. (30) becomes identical to d 1 …c†, given in potentials. In other words this kind of NACM is
Ref. [27]. the only one that will lead to ADT matrices which
(3) For the tetra-state case (i.e., j ˆ 3=2), the J~y - guarantee physically correct diabatic potentials.
matrix is of the form: We showed that the J~y -matrix, (J~y ˆ iJy where Jy is
0 p 1 the y-component of the total angular momentum)
0 
p 3 0 0
1B C which is responsible for the formation of the irre-
3 0 2 p0 C:
J~y ˆ B @ …50† ducible group of the rotation matrices, is an anti-
2 0 2 0 
p 3A
symmetric matrix just like the NACM. In fact we
0 0 3 0
showed more than that. We found (at least for the
Comparing Eqs. (50) and (33) it is noticed that the 2  2, the 3  3 and the 4  4 matrices) that all
J~y -matrix and the g-matrix become identical when relevant NACMs have eigenvalues that turn out to
p be identical to those of the relevant J~y -matrices.
g1 ˆ g3 ˆ 3=2; g2 ˆ 1: …51†
Thus, in both cases the eigenvalues of the 2  2
Employing Eqs. (35) and (36) it can be seen that matrices are (1/2,)1/2), of the 3  3 matrices are
for this choice the two numbers p and q become: (1; 0; 1) and of the 4  4 matrices (3=2; 1=2;
p ˆ n ˆ 3=2 and q ˆ ` ˆ 1=2: …52† 1=2 3=2). Assuming that this situation applies
for any dimension, N , this implies that all quan-
These possibilities are discussed in the paragraph tized g-matrices of the type presented in Eq. (1)
that follows Eqs. (40a) and (40b). Since Wigner's and the J~y -matrices are similar, namely, related via
rotation matrix for this case, i.e., d 3=2 …c†, is usually an unitary transformation. It is obvious, because
not given in the textbooks it is presented here for of this fact, that the same relation holds between
the sake of completeness: the ADT matrices (generated by the NACMs) and
0
3
p 2 p 2 3
1 Wigner's d j -rotation matrices namely, they too,
pC 2 3C S 3S C pS 2 are related via a similarity transformation.
B 3C S C…1 3S † S…1 3C 2 †
2 C
d 2 …c† ˆ B
3
p 2 p3S 2C C; At this stage the implications of the various
@ 3S C S…1p3C 2 † C…1 2
3C S A
p 3S †
®ndings reported in this article are not clear to us.
S3 3S 2 C 3C 2 S C3
We said at the start that the models to be studied
…53†
may yield insight into the nature of the NACMs.
where C ˆ cos…c=2† and S ˆ sin…c=2†. However we are not sure whether these situations
exist in `real' systems. In one of our recent publi-
cations [9], we speculated that this type of situation
6. Discussion and summary may be encountered at the position of one of the
`other' atoms that make up the molecule. For in-
In this publication we considered multi-fold stance in case of the C2 H molecule it may happen
degeneracy with the aim of revealing the connec- when the hydrogen surrounds one of the carbons.
tion between the ADT matrices and the irreducible Thus it could take place, if at all, only in situations
representation of the rotation group. To form the where all values of the potentials at the conical
connection we constructed simpli®ed models of intersections become in®nite.
two-, three- and four-states all degenerate at a The present study revealed one novel feature
single point and employed a general but relevant namely that the quantization of the 4  4 matrix
NACMs. This type of models may not exist in real yields two quantum numbers i.e. n and `. Since for
chemical systems and if they do they are far from the two other cases we had only one quantum
being common. Nevertheless we think that such a number this is clearly an unexpected result. Usu-
study, which also leads to some interesting results, ally more complicated systems (such as the three-
156 M. Baer / Chemical Physics Letters 347 (2001) 149±156

dimensional harmonic oscillator or the hydrogen [14] H.-J. Werner, P.J. Knowles, J. Chem. Phys. 82 (1985) 5033.
atom which is associated with two modes of mo- [15] P. Saxe, B.H. Lengs®eld, D.R. Yarkony, Chem. Phys. Lett.
113 (1985) 159.
tion) yield two quantum numbers but the origin of [16] H.-J. Werner, P.J. Knowles, MOLPRO is a package of ab
the two quantum numbers in our case is not clear initio programs, with contributions from J. Almlof et al.
at this stage. It is hoped that the importance of [17] R. Baer, D. Charutz, R. Koslo€, M. Baer, J. Chem. Phys.
these as well as the other ®ndings will be revealed 105 (1996) 9141.
following additional studies. [18] S. Adhikari, G.D. Billing, J. Chem. Phys. 111 (1999) 40.
[19] A. Alijah, M. Baer, J. Phys. Chem. A 104 (2000) 389.
[20] M. Baer, A. Alijah, Chem. Phys. Lett. 319 (2000) 489.
[21] M. Baer, J. Phys. Chem. A 104 (2000) 3181.
References [22] M. Baer, Chem. Phys. 259 (2000) 123.
[23] M. Baer, S.H. Lin, A. Alijah, S. Adhikari, G.D. Billing,
[1] D.R. Yarkony, J. Chem. Phys. 105 (1996) 10456. Phys. Rev. A 62 (2000) 032506-1.
[2] G. Chaban, M.S. Gordon, D.R. Yarkony, J. Phys. Chem. [24] M. Baer, Phys. Rep., in press.
101A (1997) 7953. [25] M. Baer, Chem. Phys. Lett. 35 (1975) 112.
[3] D.R. Yarkony, J. Phys. Chem. 101 (1997) 4263. [26] E.P. Wigner, Gruppentheorie, FriedrichVieweg und son,
[4] D.R. Yarkony, Acc. Chem. Res. 31 (1998) 511. Braunschweig, 1931.
[5] Z.R. Xu, M. Baer, A.J.C. Varandas, J. Chem. Phys. 112 [27] M.E. Rose, Elementary Theory of Angular Momentum,
(2000) 2746. Wiley, New York, 1957 (chapter IV).
[6] A. Mebel, M. Baer, S.H. Lin, J. Chem. Phys. 112 (2000) [28] T. Pacher, L.S. Cederbaum, H. K oppel, Adv. Chem. Phys.
10703. 84 (1993) 293.
[7] A. Mebel, M. Baer, V.M. Rozenbaum, S.H. Lin, Chem. [29] V. Sidis, in: M. Baer, C.Y. Ng (Eds.), State-to-State Ion
Phys. Lett. 336 (2001) 135. Molecule Reaction Dynamics, Vol. II, Adv. Chem. Phys.,
[8] A. Mebel, M. Baer, S.H. Lin, J. Chem. Phys. 114 (2001) Vol. 82, 1992, p. 73.
5109. [30] M. Baer, Mol. Phys. 40 (1980) 1011.
[9] A. Mebel, M. Baer, A. Yahalom, R. Englman, J. Chem. [31] C. Cohen-Tannoudji, B. Diu, F. Laloe, Quantum Me-
Phys. 115 (2001) 3673. chanics, Herman, Paris, 1977.
[10] M. Born, J.R. Oppenheimer, Ann. Phys. (Leipzig) 84 [32] H.A. Jahn, E. Teller, Proc. R. Soc. London Ser. A 161
(1927) 457. (1937) 220.
[11] M. Born, K. Huang, Dynamical Theory of Crystal [33] R. Englman, The Jahn-Teller E€ect in Molecules and
Lattices, Oxford University Press, New York, 1954. Crystals, Wiley (Interscience), New York, 1972.
[12] R.K. Preston, J.C. Tully, J. Chem. Phys. 54 (1971) 4297. [34] R. Renner, Z. Phys. 92 (1934) 172.
[13] R. Krishnan, M. Frisch, J.A. Pople, J. Chem. Phys. 72 [35] J.W. Zwanziger, E.R. Grant, J. Chem. Phys. 87 (1987)
(1980) 4244. 2954.