• Aucun résultat trouvé

A single continuum description of molecular dissociations

N/A
N/A
Protected

Academic year: 2021

Partager "A single continuum description of molecular dissociations"

Copied!
9
0
0

Texte intégral

(1)

HAL Id: jpa-00210242

https://hal.archives-ouvertes.fr/jpa-00210242

Submitted on 1 Jan 1986

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

A single continuum description of molecular dissociations

J. Robert, J. Baudon

To cite this version:

J. Robert, J. Baudon. A single continuum description of molecular dissociations. Journal de Physique,

1986, 47 (4), pp.631-638. �10.1051/jphys:01986004704063100�. �jpa-00210242�

(2)

A single continuum description of molecular dissociations

J. Robert and J. Baudon

Laboratoire de Physique des Lasers (*),

Université Paris-Nord, Avenue J. B. Clément, 93430 Villetaneuse, France

(Reçu le 12 août 1985, révisé le 28 octobre, accepte le 6 decembre 1985)

Résumé.

-

On rappelle la définition des coordonnées d’Eckart pour un système de N particules, ainsi que l’expres-

sion hamiltonienne de l’énergie cinétique classique. L’importance du Jacobien dans la définition du domaine de

configuration est soulignée. On montre en outre que la fonction d’onde ne souffre d’aucune singularité à la frontière

de ce domaine. Les dissociations du système en p sous-systèmes (2 ~ p ~ N) peuvent être décrites, en coordonnées d’Eckart, dans l’unique continuum r1 ~ oo, où r1 est la plus grande des coordonnées d’Eckart radiales. L’iden- tification d’une voie de dissociation se fait au moyen d’un vecteur unitaire de RN-1, 0393, appelé « pointeur chimique» :

dans une dissociation p-aire, r a pour lieu une variété d’ordre p - 2. On construit une algèbre de matrices de Gram

décrivant les partitions du système en plusieurs sous-systèmes ; un développement en ordre par rapport à la coor- donnée du continuum unique permet d’expliciter le comportement de 0393 au voisinage d’un lieu de dissociation.

Abstract.

-

The definition of Eckart coordinates for a N-particle system and the Hamiltonian form of the classical kinetic energy are given. The important role of the Jacobian in the definition of the configuration domain is empha-

sized. It is shown that no singularity of the wavefunction occurs at the frontier of this domain. The dissociations of the system into p subsystems (2 ~ p ~ N) are described, in Eckart coordinates, within the single continuum

r1 ~ oo, where r1 is the largest radial Eckart coordinate. Each dissociation channel is identified by means of a

unit RN-1 vector 0393 called the « chemical pointer» : in a p-ary dissociation, the locus of 0393 is a (p - 2) manifold.

A Gram matrix algebra is constructed to describe the partitions of the system into several subsystems. Owing to

a power expansion on the single-continuum variable, the behaviour of 0393 in the vicinity of its loci is explicited.

Classification

Physics Abstracts

34.10

1. Introduction.

In their original form [1], the Eckart coordinates of a system of N particles are particularly well adapted to

the description of molecular collisions. This is essen-

tially due to their inertial and canonical characters (which provide a direct way of quantization and give

a rather simple form of the Hamiltonian), and to

their ability in describing simply the dissociation channels.

In a previous paper [2], it has been shown that the

use of the nuclear Eckart frame (i.e. the principal

inertia axes of the nuclei) as the rotating frame to

which the electrons are referred, leads to an expres- sion of the Hamiltonian of a polyatomic molecule

which is very similar to that commonly used for

diatomic molecules. Purely nuclear, or electro-nuclear

dynamical couplings appearing when a Born-Oppen-

heimer basis set is used, have been given explicitly.

(*) L.A. du CNRS no 282.

When the electro-nuclear couplings are neglected,

the nuclear motion reduces to an adiabatic motion

on a single potential surface.

This problem of the motion of several atoms,

including eventually an infinite extension for some of the coordinates, has received much attention for a long

time. Many curvilinear coordinates have been studied

[3]. In particular, a great deal of attention has been, and is, devoted to the hyperspherical coordinates [4-5].

For N particles whose centre of mass is at rest these

coordinates generally consist in one radius p and 3N-4 angles. For any dissociation of the system, in which some of the mutual distances become infinite,

the hyperspherical radius p tends to infinity and all

dissociation belong to this unique continuum. As it will be seen further this property is also true for Eckart coordinates. In addition however these latter coor-

dinates have the great advantage to provide a very direct way of identification of the various dissociation channels. This point has been already studied in [2]

for binary dissociations. The aim of the present paper

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01986004704063100

(3)

632

is to give a generalization of this property to dissocia- tions of any order. For this purpose, the key point is the expression of the mutual distances between particles.

In the present case, this expression is particularly simple. It gives a general and simple method to identify

the various dissociations into 2, 3,

...

subsystems.

Moreover it makes very direct the obtention of the

relationship existing between Eckart coordinates and

hyperspherical ones, and it exhibits quite clearly the

role of the kinematic rotations in the rearrangement

processes.

In part 2 the Eckart coordinates and the classical Hamiltonian expression of the kinetic energy are

briefly recalled. The important role of the Jacobian

in the quantization is also examined : the relevant

configuration space is delimited and it is shown that

no singularity occurs at its frontiers. Dissociation and rearrangement problems are studied in part 3. The link existing between internal and external coordinates in a dissociation is studied in part 4 by means of a Gram

matrix algebra.

2. Eckart coordinates [1].

2.1 DEFINITION.

-

Let us consider a system of N particles whose centre of mass is at rest. The configu-

ration of the system is completely defined by means of

n

=

N - 1 configurational vectors R,,, which can always be chosen in such a way that the Lagrangian

form of the kinetic energy T is expressed as :

The Gram matrix of the system is the n x n matrix G whose elements are :

As no more than 3 configurational vectors are linearly independent, no more than 3 eigenvalues of G are

non-zero. In addition these are obviously real positive, namely : ri, r2, r3.

G is diagonalized by an orthogonal n x n matrix C entirely defined by 3 n - 6 independent dimensionless

quantities q,, ..., q3n-6.

The 3 n - 3 coordinates : ri, r2, r3, ql, ..., q3n-6

only specify the relative positions of the particles.

They need to be completed by a set of 3 Euler angles

a, P, y defining the absolute position in space of the instantaneous solid. The body-fixed or rotating frame

is unambiguously defined by considering the 3 ortho- normalized vectors a; (i

=

1, 2, 3) defined by :

ri, r2, r3 are the instantaneous radii of gyration and

ai, a2, a3 lie along the instantaneous principal axes

of inertia. In the rotating frame, any configurational

vector can be expressed as :

The classical Hamiltonian form of T has been given by Eckart [1] :

with: Bi = (rk - rj)’I(rk’ + rJ); Ci

=

(rf - r;)2/

(2 rk rj), where i, j, k is a cyclic permutation of 1, 2, 3.

The momenta are defined as follows :

Mi

=

M aim

=

ðT/ðwi where M is the total angular

momentum and Wi

=

w a;, w being the angular velocity vector,

with

,

2.2 QUANTIZATION. IMPORTANT ROLE OF THE JACO- BIAN. - When canonical coordinates Q1, Q2, - - ol Q3n,

such as the Eckart ones, are used, the most straight-

forward quantization is carried out by applying the simple rule [6] :

and defining the scalar product by :

In order to recover the usual Cartesian coordinates

XI,

...,

X3n, in which no special topological problem

occurs, one has to consider Cartesian wavefunctions Oc whose scalar product is :

where :

For Eckart coordinates :

(4)

where a is the (3 n-6)-th order determinant whose k-th row is :

with :

As both scalar products must lead to identical physical results, the two wave-functions are necessarily related by :

where a is an arbitrary constant phase factor. In the

following one takes a

=

0; as usual, this gives in the

Hamiltonian an additional « extra-potentials term [3].

ln Cartesian coordinates, the configuration space is R 3". Because of the existence of nodal surfaces J

=

0,

e.g. ri

=

r2, etc..., the transformation X -+ Q is

not univocal. As a consequence the configuration

space in canonical coordinates cannot be R 3" : in fact R 3" must be quotiented by the Jacobian kernel, i.e. it

must be reduced to a domain limited by nodal sur-

faces. In particular, for coordinates ’1’ r2, r3, this means

that any configuration of the system of particles is

describable within the restricted domain ri > r2 > r3.

On the nodal surfaces J

=

0, 0 necessarily vanishes identically and no singularity occurs.

Example : the triatomic case.

The configuration of 3 particles of masses mi, m2, m3 is specified by 2 vectors : pi2 joining particle 1 to particle 2, and P12,3 joining the centre of mass 1,2 to particle 3. The configurational vectors are :

where

G is a 2 x 2 matrix whose eigenvalues are the roots

of the characteristic equation :

Obviously, RI, R2 being given, one has to choose

which root r2, r2 is the largest one.

C is here a 2 x 2 rotation matrix depending on one angle q :

where

The rotating frame { ai } is defined by :

The Jacobian is : J = r1 r2(ri - r2) sin p.

The nodal surface (r1

=

r2) corresponds to the geometry :

Making the choice ri > r2, one has to examine the

asymptotic behaviour of 0 when ri - r2 - 0+.

Let us define :

As shown in a previous paper [2] the triatomic

Hamiltonian is :

where V is the potential energy.

When v -+ 0+, the asymptotic behaviours of the relevant quantities are as follows :

V does not present any singularity and it is independent of q when ri

=

rZ. It can then be simply written : V(r l’ ri)

=

V(u).

The asymptotic form of the Hamiltonian is :

It is worthnoting that variables q and y are actually

combined into the unique variable 0 = q + y. In the case ri = r2, there is an ambiguity in the definition of the Eckart vectors ai, a2 (these are two orthogonal

unit vectors in the plane of the three particles) ; howe-

ver this ambiguity is of no consequence, since a’change

in the definition of a1, a2 (i.e. a rotation around a3)

modifies at the same time q and y in such a way that 0 remains the same.

The wavefunction 0(u, v, q, a, p, y) is separable in

the form :

(5)

634

The separated differential equations are readily obtai-

ned :

where Jl2, m2 E N, and E1 + E2

=

E.

The important point here is that G is the regular

solution of an equation containing a pure repulsive centrifugal potential, i.e. a regular Bessel function.

As v-+O+, G-v’ with 21= (m2 + 1)il2 + 1. On

another hand, J asymptotically behaves as U3 v.

Therefore, the Cartesian wavefunction GC behaves as

i.e. it is also regular.

3. Dissociations.

3.1 REARRANGEMENTS. - In a binary dissociation,

the system of N particles breaks into 2 subsystems of N1, N2 particles (N1 + NZ

=

N, 0 N1 N), whose

centres of mass are separated by an infinitely large

distance. In general, several binary dissociations exist,

each of them corresponding to a specific rearrangement channel. These considerations are readily generalized

to ternary,... p-ary dissociations. The definition of a rearrangement channel is intimately related to the

choice of the configurational vectors {Rex}. This

choice itself is related to a specific way of constructing

the centre of mass of the system. This construction can

be carried out iteratively, in the following way : one takes first the centre of mass of 2 particles, then one joins it to a third particle (or to the centre of mass of

several other particles, already determined); this

determines a new partial centre of mass, etc... The relative vectors obtained along this construction, once conveniently weighted, form a possible set of confi- gurational vectors. Let a(N) be the number of different

sets of this kind. One considers here that any change

of the type : Rex -+ - Rex leaves invariant the configu-

rational set. Using the tree theory, one can show that

[7]:

or equivalently that a(N) obeys the recurrent rela-

tion [8] :

Let [’1’ r2, r3, C(qi, ..., Q3n-6), 0152, P, y] be the Eckart

coordinates defined upon one of the previous configu-

rational sets, { R,,, 1. One has to examine how these

coordinates change when another reference set of the

same kind, { R’ I is taken in place of { Ra }. The new

set can be expressed as :

where the X fJa. depend only on the masses.

As the kinetic energy must have the same form in both sets, then :

Therefore :

This means that the matrix X, whose elements are Xa,

has the property :

X is then a rotation in R ", depending only on the

masses. This transformation X is one of the a(N)

kinematic rotations [4].

As already shown in a previous paper [2] the Eckart

coordinates transform in the following way :

where the transformation q -+ q’, or equivalently

C -+ C’ is one of the kinematic rotations :

Obviously there are more than a(N) transformations which obey to conditions (12) and (13). Starting from

a set { R. }, one may generate another possible set by setting

where X is any constant orthogonal n x n matrix.

Generally, a vector :1B)’ is not obtained by weighting

a relative vector joining two partial centres of mass.

In fact however, as soon as dissociation problems

are considered, supplementary conditions are imposed

to the configurational vectors : the specification of a given fragmentation of the system into two subsystems

or more, implies to relate directly some of the configu-

rational vectors to distances between specific partial

centres of mass, and this relation is automatically

realized when a set of the type Ra is used. A set of this type will be called a proper set. In summary 3C corres-

ponds to an arbitrary change of origin on the variables q> whereas X describes a transition between two

rearrangements.

(6)

3.2 MUTUAL DISTANCES IN ECKART COORDINATES.

-

The consideration of mutual distances is evidently important in a discussion about the dissociations.

It is also important in the derivation of the expression

of the potential in Eckart coordinates.

The relative vector PJlV joining two particles M, v,

can be expanded over the configurational vectors R0153,

as :

where the quantities Yay depend only on the masses.

Using (4), equation (17) gives :

where

For a given couple (Jl, v), Zuy depends on the masses

and on the variables qk.

It is worthnoting that Zffv remains invariant under

a change of configurational vectors, since this quantity

is the scalar product of p,v by the invariant vector

ri ai. Obviously this invariance is preserved for all configurational sets, proper or not.

Finally the mutual distance p,, is given by :

By means of this expression of the mutual distances,

one sees that, when no particular constraint is imposed

to the coefficients ZP’, (i.e. no particular condition imposed to variables qk), then when, 1 is made infinite,

r2 and r3 being finite, all relative distances become infinite, i.e. the system breaks completely, into N particles. All dissociations of lower orders (dissocia-

tions into N - 1, N - 2,..., 2 subsystems) are also

obtained within the same unique continuum (ri infinite,

r2 and r3 finite), by imposing a certain number of

conditions to Z!". This point will be examined in the

following paragraph.

3.3 DISSOCIATIONS. CHEMICAL POINTER.

-

Let us

consider a dissociation of the system into p sub- systems S1, ..., Sp, containing respectively N1, ..., Np particles, with N1 + ... + Np

=

N and 2 p N.

In this p-ary dissociation, it is assumed that all mutual distances within the subsystems remain finite, whereas

the mutual distances between the centres of mass

(or as well between particles belonging to different subsystems) tend to infinity.

To constrain the mutual distances internal to a given subsystem, say Sk, to be finite, it is necessary and sufficient that Nk -1= nk internal mutual distances PIlV remain finite when ri -+ oo (r2, r3 finite).

Consequently :

or :

O O

The number of such equations is equal to the number

of couples in Sk, i.e. 2 nk(nk + 1). However, only nk equations among them are independent. Similar equa-

tions are obtained for all subsystems 5’i,...,Sp.

Let us now introduce the following R "-vectors :

r is a unit vector depending only on the variables qk.

On another hand, the vectors Y’’’’ (and even their improper version 1JPV) depend only on the masses.

Then equations (21-22) become :

There are nl+n2+w+np=n-p+1=N-p

independent equations. They express the fact that

T(qk) is orthogonal to n - p + 1 fixed vectors. As

1 r 12

=

1, this means that, in a givenp-ary dissociation,

the locus of r on the hypersphere of radius unity is

a (p - 2) manifold, from which the intersections with the (p - 2) manifolds corresponding to the other

p-ary dissociations are removed (these intersections include all dissociations of lower order : p - 1,

p - 2, ..., 2).

Actually, as T, as well as Y#’v, are defined except for

an arbitrary orthogonal transformation X (which

fixes the origin of the qk and the orientation of the

corresponding curvilinear frame), there is a double

determination for each dissociation channel. For

instance, in the case p

=

2, where the loci of rare reduced to simple points, one binary dissociation is associated with 2 diametrically opposite points on

the hypersphere. This is in fact of no importance since

the domains of variation of qk can be restricted to

[0, n[ without any loss of generality. In the following,

all vectors will be chosen on a half hypersphere.

The number of (p - 2) manifolds is equal to the

number of ways of partitioning a set of N elements into p non-empty subsets, i.e. the Stirling number of

the second kind [9] :

The direction of the unit vector r is the label of a

specific dissociation channel, in the unique rl-conti-

nuum. For this reason, r may be called the chemical

pointer.

Example : system of 4 particles.

In this case n

=

3, then r and the 6 vectors Y#,V belong

to R 3. For each binary dissociation (in 2 + 2 or

3 + 1 particles), r is perpendicular to 2 vectors Y#’v,

and thus it points in a fixed direction (0-manifold).

(7)

636

The number of such points on the half-sphere is equal

to S’

=

7 (see points A, B, ..., in Fig. 1). In the same

manner, for each ternary dissociation channel (2 + 1 + 1 particles) r is perpendicular to one vector YIlV and

its locus is a (halo circle of radius 1, from which the

intersection points with the other similar (halo circles

must be removed (Fig. 1). Note that there exist 83

=

6

ternary dissociation channels. Finally, there is only

one quaternary dissociation (explosion of the system)

in which r is no longer constrained except that it

cannot belong to the previous loci.

Fig. 1.

-

Dissociation of a system of 4 particles (1, 2, 3, 4).

(a) Definition of the configurational vectors : R1

=

p’/2 12, R2

=

/,11/2 34, R3

=

/11/2 C34 C12’ where C34 is the centre

of mass 3, 4 and C12 that of 1, 2.

(b) Loci of the chemical pointer r on the sphere of radius 1, for the different dissociations of the system. The figure is

drawn in the case ml = m2

=

m3

=

m4. When r points

on a circle labelled by p, v the relative distance p,, remains finite when r 1 -+ oo. The circle p, v is in the plane perpendicu- lar to a vector YIlV (see text). Only the directions of these vec- tors are shown. The intersecting points of these 6 circles cor- respond to the 7 possible binary dissociations (only a half sphere, e.g. xi a 0, has to be considered) : A : 12 + 34,

B : 23 + 14, C 13 + 24 (intersections of two circles);

D : 123 + 4, E : 234 + 1, F : 341 + 2, G : 412 + 3 .(inter-

sections of three circles).

Let us take the following choice of configurational

vectors :

In such conditions, the components of the vectors YIl" are as follows :

4. Internal and external Eckart coordinates. Gram matrix algebra.

4.1 INTERNAL AND EXTERNAL COORDINATES. - In order to analyse the partitioning of a system into several subsystems, or conversely the collapse of

several subsystems into a unique one, using Eckart

coordinates only, it is necessary to examine the link

existing between the coordinates of the subsystems (internal coordinates) plus those defining their relative

positions (external coordinates) and the coordinates of the whole system.

Let us consider a system of N particles as the sum

of subsystems Sl, S2,

...,

Sp, containing respectively NI, N2, ..., Np particles, with given relative positions (subsystem Sd). This is equivalent to make a choice of

a proper configurational set of n vectors { Ra}, such

that it can be partitioned into p + 1 subsets : nk

configurational vectors Rp belonging to subsystem

Sk (k

=

1, ..., p), plus nd = p - I configurational vec-

tors Ry specifying the relative positions of the sub- systems Sk (i.e. subsystem Sd) :

For each subsystem, either internal or external, one defines Eckart coordinates, i.e. a Gram matrix G k

with characteristic values t1,2,3’ an Eckart frame

a1,2,3 and a rotation matrix Ck, where k

=

0 (sub- system Sd), 1, ..., p. Similar quantities, namely G,

’1,2,3’ a1,2,3 and C, are defined for the whole system.

4. 2 COMBINATION OF THE SUBSYSTEMS.

-

Each vector

Rf can be expanded as :

(8)

and :

All vectors e) can be gathered into a unique family

{ efJ} (f3

=

l, ..., n), using the same ordering of the subsystems as that used in (26). This definition of efJ includes also cases in which subsystems contain less

than 4 particles, since in such cases the number of

non-zero characteristic values reduces to 2, 1, 0, respectively for 3, 2, 1 particles (or centres of mass).

Let us now define a n x n matrix C’ as being the block-diagonal matrix, whose blocks are Cd, C 1,

...,

C P.

By means of C’, one can write any configurational

vector as :

From (28), the total Gram matrix is readily derived :

or :

O O

where P is the n x n matrix of elements : P rzp

=

ea ep.

Taking into account the definition of e,, the P matrix takes the following form :

where crosses and hatched areas correspond to non-

zero elements (respectively (rt)2 and rt rl’ ak k’).

Let C p be the rotation which diagonalizes P :

where A is the diagonalized Gram matrix of the total system.

Using the definition of C, G

=

C L1C and (29),

one gets :

and thus :

In summary, knowing the partial Eckart coordinates of the subsystems Sd, Si,

...,

Sp, one is able to construct

the coordinates for the total system : from (32) one

deduces the values of r 2 and the Eckart frame { a, 1,

i.e. the Euler angles of the total system. Then equa- tion (31) gives the total rotation matrix C.

It may be noticed that this combination of partial

Eckart coordinates into the total ones does not imply

the choice of proper configurational sets, neither for

subsystems S1, ..., Sp, nor for Sd. However if one needs

simple asymptotic forms as ri -+ oo (cf. § 3.1) when

the system breaks into S1 + S2 + ... + Sp, the confi- gurational set of Sd is necessarily a proper one. More

generally, in a collision with rearrangement, it is

necessary to take all configurational sets as proper

ones.

4.3 BEHAVIOUR OF r IN THE VICINITY OF A DISSOCIA- TION LIMIT POINT.

-

Let R 1 be the unique (proper) configurational vector of Sd, along which the binary

dissociation takes place. From the expansion (4)

one has :

In particular :

When ri -+ oo, R1 only becomes infinite and it takes the asymptotic form :

On another hand, Ra# 1 being finite, Cla tends to zero,

then as expected F -+ { 1, 0 ... 0 1. From the above expression of Cia, one sees that the vanishing corm- ponents of r behave as :

D1ex is the projection of an internal configurational

vector, belonging to S1 or S2, on the direction of the dissociation (Fig. 2), which means that D1ex actually

become an internal variable in S1 or S2.

In fact this property can be generalized to any kind of dissociation. Let us consider a well defined p-ary dissociation. As shown before, the chemical pointer r, defined on the proper set of configurational vectors Rex, belongs to a given p - 2 manifold. By a convenient

rotation X in R ", one can pass to an improper set Sip

such that the components of the chemical pointer are { 1, o ... o }, which means that ?1 only becomes infinite,

with 11 r, a 1 (oo ), i.e. that an improper binary

dissociation occurs. Then the components C1a of the improper rotation matrix C behave as before, and one

can deduce the behaviour of the components of the

(9)

638

Fig. 2.

-

Binary dissociation of a system into 2 subsystems Sl + S2, described in a proper set of configurational vectors : R1 (unique vector in Sd), becoming infinite when ’1 --+ oo, and Rex (systems S1, S2) remaining finite. To the first order in

ri 1, the chemical pointer behaves as :

proper matrix C, by applying the rotation X which

depends not solely on the masses, but also on the

relative asymptotic direction of those proper vectors

Ra(E Sd) along which the p-ary dissociation takes

place.

5. Conclusion.

From a dynamical point of view, the inertial and canonical characters of the Eckart coordinates sim-

plify significantly the expression of the kinetic part of the Hamiltonian of a N-particles system. These coordinates appear a priori as curvilinear coordinates suitable for the description of the motion of a many-

body system. The difficulties which could come of the

non univocal character of the transformation from Cartesian coordinates into Eckart ones are actually

ruled out as soon as one considers the relevant confi-

guration domain, whose frontiers are the nodal surfaces of the Jacobian. Just as does the r

=

0 frontier in

spherical coordinates, the nodal surfaces act as

« mirrors » on which the wavefunction vanishes.

Consequently any molecular geometry can be des- cribed, and any dynamical problem can be solved in

principle, within this limited domain, without any risk of singularity at the frontiers.

In addition to these interesting dynamical properties,

the Eckart coordinates have proven to be particularly

°

efficient in the description of all possible rearrangement

channels. Firstly all these channels belong to a single

continuum (ri oo, where r1 is the largest radial coor- dinate). Secondly, these coordinates lead to a vectorial

representation of the rearrangements since the tran- sition from one rearrangement to another one is

equivalent to a rotation in space R n (n

=

N - 1), depending only on the masses (kinematic rotation).

This transformation involves only the dimensionless Eckart variables qk, i.e. the elements of C, the rotation

in R" which diagonalizes the Gram matrix of a confi-

gurational vector set. The unit R "-vector r, whose components are the elements of the first line of C,

allows a straightforward identification of the disso- ciation channels. This vector acts as a chemical

pointer : to a specific p-ary dissociation corresponds

a locus of r, consisting in a (p - 2) manifold. In

particular, definite points on the hypersphere of

radius 1 correspond to all possible binary dissociations.

The asymptotic behaviour of r in the vicinity of these points can be derived, by taking as a coordinate axis in R" the direction related to a specific binary disso-

ciation. To the first order in ri 1, the components T a ( a # 1) behave as D1,)r1’ where D 1 a is the pro-

jection of the internal configurational vector R.

on ai(oo), i.e. on the direction of the separation of the

2 fragments. This property can be extended to disso- ciations of higher order, since formally any given

p-ary dissociation is equivalent to a binary disso-

ciation in a definite improper configurational set.

This behaviour provides some insight on the proper

way of expanding the potential and the wave function

over angular harmonics.

The main remaining tasks are then to derive the corresponding system of coupled equations and solve

it in some specific cases, using more or less approxi-

mate methods. From this point of view, the case of a

diatom-atom collision involving only rotational exci-

tation, appears as the simplest one. Moreover it is

a test case since such collisions have been already extensively studied in using standard coordinates

[10,11].

Acknowledgments.

We are grateful to Drs. A. Nauts and X. Chapuisat

for providing them with their article prior to publi-

cation. We thank Pr. F. Combes and Dr. M. Ducloy

for their interest in this work.

References

[1] ECKART, C., Phys. Rev. 46 (1934) 383.

[2] ROBERT, J. and BAUDON, J., J. Phys. B (1985), to be published.

[3] NAUTS, A. and CHAPUISAT, X., Mol. Phys. 54 (1985),

to be published and references therein.

[4] SMITH, F. T., J. Chem. Phys. 31 (1959) 1352.

[5] FANO, U., Rep. Prog. Phys. 46 (1983) 97.

LAUNAY, J. M. and LE DOURNEUF, M., Proc. XIIIth

ICPEAC, Berlin (1983), Invited Papers, p. 635.

[6] DIRAC, P. A. M., The Principles of Quantum Mecha-

nics (Clarendon Press) 1957, p. 87.

[7] COMBES, F., private communication.

[8] DUCLOY, M., private communication.

[9] ABRAMOWITZ, M. and STEGUN, I. A., Handbook of

Mathematical Functions (Dover Publications) 1972,

p. 824.

[10] REBENTROST, F. and LESTER Jr., W. A., J. Chem. Phys.

67 (1977) 3367.

[11] LAUNAY, J. M., J. Phys. B 9 (1976) 1823.

Références

Documents relatifs

The energy stores in the z modes cannot come from charges as it is stored in the wide space between the plates. This energy can only come from the time derivation of the

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

I conduct the research in two steps: (1) evaluate the impacts of transport investment on intercity accessibility patterns; and (2) identify the influence of accessibility

Deux triangles sont semblables si les longueurs de leurs côtés sont proportionnelles.. Exemple : Les triangles ABC et DEF

Pour tous les patients de notre étude, le délai de deux heures ± 15 minutes pour la précision des dosages san- guins réalisés deux heures après la prise, de même que le maintien

The following chapter addresses mechanisms of internal appetite control, starting by discussing intuitive eating, a recently introduced concept that describes eating in response

Therefore, transgenic 5xFAD mice (bearing 5 familial Alzheimer disease-linked mutations) which shows accentuated amyloidogenesis, was crossed with mice having a deletion of the

Following Lerner, now define the external debt burden by the steady-state trade surplus needed to finance public debt service.. This steady-state debt logically is ”paid for” by