Deutsch

Symmetry Breaking
in Chemical Reactions

Wolfgang Kliesch

The website provides a short introduction to the mathematical description of the symmetry of molecules and the symmetry breaking in molecular systems. A model of chemical reactions which considers the symmetries of the reactant, the transition configuration and the product is introduced.
The reader who is interested in greater details is referred to [1].

1. Molecular Systems

Molecules are systems that consist of nuclei and electrons. Due to the lower mass of the electrons, the electrons move much faster than the nuclei. Therefore the nuclei may be regarded to be stationary compared with the electrons, i. e. the electrons follow the movements of the nuclei without any delay (Born-Oppenheimer-Approximation) [2]. Thus the movement of the nuclei may be separated from the movement of the electrons, and the movement of the nuclei can be treated by Newton's principles whereas the movement of the electrons is treated by the principles of quantum mechanics. Within this semi-classical framework the nuclei of a molecule form a well-defined geometrical figure, the nuclear skeleton of the molecule. Every geometrical arrangement of the nuclei (nuclear configuration) is assigned an energy value which results from the movements of the electrons. These considerations lead to the concept of molecular systems.

1.1 Molecular Systems

A molecular system ist a system which consists of n nuclei and a function E that assigns an energy value to each nuclear configuration. The energy value is the eigenvalue of the corresponding Schrödinger equation of the movement of electrons; see [2].
The nuclei are moving according to Newton's principles in the potential field generated by the energy function E.

In general a molecular system has several nuclear configurations which are the nuclear skeleton of a molecule. The nuclear skeletons are local minimizers of the energy function E.

The shape of a nuclear configuration can be illustrated in the three-dimensional Euclidean space by assigning a point of the three-dimensional Euclidean space to every nuclei of the molecular system. The position vectors of these points are denoted by pi , i=1, …, n, in the following, where n is the number of the nuclei of the system.

Henceforth the nuclear configurations are assumed to be non-linear, i. e. the nuclei of a configuration are not laying on a line of the three-dimensional Euclidean space.

1.2 The Symmetry of a Nuclear Configuration

A nuclear configuration is symmetric if the set of the nuclei of the configuration is mapped onto itself by at least one symmetry operation different from the identity.
There are exactly five symmetry operations on the three-dimensional Euclidean space:

Ck : rotation through 360/k degrees, k>1,
σ: reflection at a plane,
Sk : rotary reflection = σ ⋅ Ck (the rotational axis is perpendicular to the plane of reflection),
I: inversion,
E: identity = rotation through 360°.

The set of the symmetry operations which map a nuclear configuration onto itself forms the symmetry group of the nuclear configuration. The symmetry group of a non-linear nuclear configuration is always finite; see [1, Theorem 11.1]. It describes the symmetry of the nuclear configuration completely. The symmetries that a molecule can have, are described in detail for example in [1, 3, 4]. The group C1={E} is the symmetry group of the asymmetric nuclear configurations.
Suppose a nuclear configuration has the symmetry of the group G. If s is a symmetry operation of the group G then for the position vector pi of each nucleus there is exactly one position vector pji so that

s ( pi ) = pji,  i=1, …, n ,  (1)

where

s(pi) ≠ s(pk) ,  if   i≠k .  (2)

Since the symmetry operations are isometric functions, every symmetry group G is isomorph to a group Ĝ of orthogonal (3, 3)-matrices [1, Theorem 8.2], i. e. if an orthonormal base is chosen in the three-dimensional Euclidean space then every symmetry operation s of the group G is assigned one and only one orthogonal (3, 3)-matrix Os of the group Ĝ. Equations (1) turn into the equations

Ospi = pji,  i=1, …, n.   (1a)

By equation (2) every symmetry operation s∈G generates a permutation πs on the set of the nuclei of a nuclear configuration which has the symmetry of the group G. Therefore the symmetry of a nuclear configuration can also be described by permutations. Equations (1) define the permutation

πs= ( 12…n j1 j2 … jn ) .  (3)

The set Π of the permutations πs,  s∈G, is a sub­group of the symmetrical group, the group of all permutations of n objects. It is isomorphic to a group P of permutation matrices of order n; see [1, Lemma 3.5]. So there is a permutation matrix Ps∈P for every permutation πs∈Π.

2. Symmetry Breaking in Molecular Systems

Here symmetry breaking means the distortion of a symmetric nuclear configuration to a nuclear configuration of lower symmetry or an asymmetric nuclear configuration, where the loss of symmetry is minimal.
The change of symmetry is caused by a force acting on the nuclear configuration. The force determines the extent of the loss of symmetry considerably.

For the mathematical description of symmetry breaking in molecular systems it is favorable to summarize the n position vectors pi of a nuclear configuration in a vector

p= ( p1 ⋮ pn )   (4)

so that every nuclear configuration is represented by a point of the 3n-dimensional Euclidean space ℝ3n. Note that only the vectors p of which the position vectors pi fulfil the condition

pi ≠ pj ,  if   i≠j,

represent a nuclear configuration. They are called configuration vectors.
By equation (4) equations (1a) can be rewritten as

( Osp1 ⋮ Ospn )= ( pj1 ⋮ pjn )= ( Ps⊗I3 ) p,  (5)

where Ps is the permutation matrix defined by permutation (2), and I3 is the identity matrix on ℝ3. The operator ⊗ is the Kronecker product of two matrices. Further more the equation

( Osp1 ⋮ Ospn )= (In⊗O s) p   (6)

is valid, where In is the identity matrix on ℝn.

2.1 The Symmetry of a Configuration Vector

If the nuclear configuration given by the configuration vector p ∈ℝ3n/has the symmetry of the group G then by equations (5) and (6)

(In⊗O s) p = ( Ps⊗I3 ) p.

As the matrix ( Ps⊗I3 ) is orthogonal, and the orthogonal matrices are regular, the equation

( Ps⊗I3 )-1 (In⊗O s) p =  p   (7)

is true, i. e. a configuration vector p of the 3n-dimensional vector space ℝ3n/represents a nuclear configuration which has the symmetry of the group G if and only if it fulfils equation (7) for all matrices Os of the group Ĝ. Per definitionem, an arbitrary vector p∈ ℝ3n/ has the symmetry of the group G if it fulfils equation (7) for all matrices Os of the group Ĝ. For simplicity, a vector that has the symmetry of the group H is called H-symmetric.
The mapping

Γn(s) = ( Ps⊗I3 )-1 (In⊗O s) = PsT⊗ O s

sends the symmetry operations of the group G into the group of the orthogonal (3n, 3n)-matrices. It is a faithful matrix representation of the group G (s. [1, Lemma 14.1]) and generates the faithful linear representation

Κn: G→I0 (ℝ3n/).

The group I0 (ℝ3n) is the group of the isometries on ℝ3n/ which map the origin of ℝ3n onto itself; see [1, Section 14.1].
A vector p∈ℝ3n/ is G-symmetric if it is a fixed point of the mapping Κn(s) for every symmetry operation s∈G, i. e.

Κn (s) (p )=p  ∀ s∈G.

Thus, a nuclear configuration has the symmetry of the group G if and only if the configuration vector is a fixed point of the previous equation.
If p∈ ℝ3n/ then the set

Σp(Κn (G))={Κn (s)∈ Κn(G) | Κn (s) (p )=p}

is a sub­group of the group Κn(G) , the isotropy sub­group of the vector p in the group Κn(G). The vector p has at least the symmetry of the group G if Σp(Κn (G))=Κn (G). If H is a maximal sub­group of the group G and Σp(Κn (G))=Κn (H) then the vector p has exactly the symmetry of the group H.
By means of the linear representation Κn the 3n- dimensional Euclidean space can be decomposed in a direct sum of m invariant sub­spaces 𝕍k so that

ℝ3n/= 𝕍1+ … +𝕍m   (m≤3n) ,    Κn(s) ( 𝕍k)⊆𝕍k   ∀s∈G   (1≤k≤m),  (8)

and the sub­spaces 𝕍k do not contain any further invariant sub­spaces; see [1, Theorem 5.2].
The vectors of the vector space ℝ3n/ that have the symmetry of the group G form a sub­space of ℝ3n/, the fixed point subspace

Fix(ℝ3n/;  Κn( G))= {p∈ ℝ3n| Κn (s) (p )=p  ∀ s∈ G}.

If H is a sub­group of the group G then the inclusion

Fix(ℝ3n;  Κn( G))⊆ Fix(ℝ3n/;  Κn( H))

is true. If a vector p∈ ℝ3n/ has the isotropy sub­group Σp(Κn (G)) then it is an element of the fixed point sub­space Fix( ℝ3n/;  Κn( G)). The the fixed point sub­space Fix( ℝ3n/;  Κn( G)) is the direct sum of the one-dimensional sub­spaces 𝕍k of the direct sum (8); see [1, Proposition 14.11].
If G=C1 then Fix(ℝ3n/;  Κn( G))=ℝ 3n/.

2.2 Symmetry Breaking

The symmetry breaking described above implies that a symmetry breaking force acting on a G-symmetric nuclear configuration has to be given by a vector y∈ ℝ3n/ which has the symmetry of a subgroup of the group G. Additionally the vector y has to meet certain conditions to break the symmetry minimally.

A sub­group H of the finite group G, H ≠G, is called a bifurcation sub­group of the group G in the subspace 𝕍 of the direct sum (8) if (i) there is a vector y,  y≠0, so that Κn(H) =Σy(Κn (G)) and (ii) dim(𝕍∩ Fix(ℝ3n/; Κn(H)))=1; see [1, Section 16.1]. The vector y is called a bifurcation vector. It is H-symmetric.
A bifurcation vector is uniquely determined up to length and sign.

If H and H' are two bifurcation subgroups of the group G in the sub­space 𝕍 of the direct sum (8) then H=H'; see [1, Proposition 16.1]. Therefore a bifurcation sub­group of the group G cannot contain further bifurcation sub­groups of the group G. In this sense the loss of symmetry caused by the associated bifurcation vector is minimal.

If H is a bifurcation sub­group of the group G in the sub­space 𝕍 of the direct sum (8) then 𝕍 is one-dimensional if and only if H is a normal sub­group of the group G (i. e. H satisfies the condition gH=Hg for all g∈G); see [1, Theorem 16.1]. Hence, the dimension of the sub­space 𝕍 is greater than one if and only if H is not a normal sub­group of the group G.
If 𝕍 is one-dimensional then H is the only bifurcation sub­group in 𝕍 and the geometrical deformation of a G-symmetric nuclear configuration is uniquely determined up to a reflection or a rotation through 180°. If dim 𝕍≥2 then also the sub­groups of the group G which are conjugate to the sub­group H, are bifurcation sub­groups in 𝕍 so that the distortion of a nuclear configuration which is caused by a bifurcation vector of 𝕍, takes place in different ways. The bifurcation vectors which belong to the sub­groups conjugate to the group H, generate geometrical different, but chemically identical nuclear configurations.

In [1] the bifurcation subgroups of the symmetry groups of molecules are determined up to the symmetry groups Td, Oh and Ih. The bifurcation sub­groups of a symmetry group can be derived from the irreducible characters of the symmetry group. Every symmetry breaking that may occur in a nuclear configuration, is determined by a character of the symmetry group of the nuclear configuration.

3. A Model of chemical Reactions

In a chemical reaction the reactant (initial molecule(s)) is transformed to a product (final molecule(s)). The transformation is connected with the surmounting of the energy barrier between reactant and product. In the Transition State Theory [5, 6] a saddle point of first order of the energy function of the considered molecular system determines the energy barrier between reactant and product. The saddle point defines the transition configuration of the chemical reaction.

If reactant and product of a chemical reaction have different symmetries then the transition configuration cannot be a saddle point of first order [1]. So a model of chemical reactions has also to consider transition configurations which are defined by a saddle point of higher order; see Section 3.1.

The symmetry groups Hre and Hpr of the reactant and the product configuration of a chemical reaction, respectively, have to be sub­groups of the symmetry group G of the transition configuration; cf. [7]. Therefore the following types of chemical reactions can be distinguished:

I Hre=Hpr =G
II Hre, Hpr ≠G
IIIa Hre≠G, H pr=G
IIIb Hre =G, Hpr≠G

The chemical reactions of type II and III are always connected with a symmetry breaking.

3.1 Transition Configurations

Since — as proven in [1] — the transition configuration of a chemical reaction of type II or III cannot be a saddle point of first order the concept of transition configurations has to be redefined.

A vector v∈ℝ3n is called a descent vector with respect to the point p∈ℝ3n if for small values of h∈ℝ the inequality

E(p+hv) -E(p)<0   (9)

is fulfilled.
If H is a bifurcation sub­group of the symmetry group G in a subspace 𝕍 then 𝒟G(H) denotes the set of all points of the fixed point subspace Fix(ℝ3n ; Κn(G)) for which there is a descent vector that is a bifurcation vector with respect to the group H in a sub­space of the direct sum (8).
A configuration vector p0∈ Fix(ℝ3n;  Κn( G)) is called a transition configuration with respect to the group H if p0∈𝒟G (H) and E(p0)=min {E(p)| p∈𝒟G (H)}.
By the definition a transition configuration need not be a saddle point. The Hessian matrix of a transition configuration need not have a negative eigenvalue. It is sufficient if the descent vector is a element of the kernel of the Hessian matrix of the transition configuration.
If the symmetry groups of the reactant and the product of a chemical reaction are different then the chemical reaction has two transition configurations in general, one transition configuration with respect to the symmetry group of the reactant and a second one with respect to the symmetry group of the product. The transition configurations have the same symmetry. Therefore they are elements of the same fixed point subspace.

If the configuration vector p is a saddle point of first order of the energy function E then the Hessian matrix H(p) has exactly one negative eigen­value λ- and the eigen­vector v - which belongs to λ- satisfies inequality (9). If p is G-symmetric then v- is contained in a one-dimensional sub­space 𝕍 of the direct sum (8) and Κn(s) ( v- )=±v- for all s∈G. If Κn(s) ( v- )=v- for all s∈G then the eigenvector v- is an element of the fixed point sub­space Fix(ℝ3n ; Κn(G)). Thus the saddle point p is a transition configuration of a chemical reaction of type I. If there are symmetry operations s in the group G for which Κn(s) ( v- )=-v- then the eigen­vector v- is a bifurcation vector with respect to a sub­group of the group G and the saddle point p is a transition configuration of a chemical reaction of type II or III. As the bifurcation vectors of a bifurcation sub­group are uniquely determined (up to sign and length) [1], a saddle point of first order can be a transition configuration with respect to reactant and product of a chemical reaction if and only if the symmetry groups of the reactant and the product are equal. The chemical reaction is of type II, where reactant and product are chemically identical.

3.2 Chemical Elementary Reactions

In the present section p and p' denote the configuration vectors of the reactant and the product, respectively, of the chemical reaction

p→p'   (10).

Let G be the symmetry group of the transition configuration(s) of chemical reaction (10). Suppose that the symmetry group H of the reactant is a bifurcation sub­group of the group G in a sub­space 𝕍 of the direct sum (8). Thus the configuration vector p is an element of the fixed point sub­space Fix(ℝ3n ; Κn(H)).

The normalizer NG(H)={s ∈G | Hs= sH} of the sub­group H in the group G satisfies the relation

H⊆NG(H) ⊆ G.

If H≠NG(H) then there is an image vector p∗= Κn(s)( p) of the reactant vector p in the fixed point subspace Fix(ℝ3n ; Κn(H)), s∈NG(H) \ H, where Κn(s)( p)= Κn(s' )(p ) for all s,  s'∈NG(H ) \ H; see [1, Lemma 14.11 und Theorem 16.2]. The nuclear configuration defined by the configuration vector p∗ is a reflection, a rotation through 180° or an inversion of the reactant configuration which is given by the configuration vector p. If p'=p ∗ is set in equation (10) then the chemical reaction

p→p ∗  (11)

in the fixed point subspace Fix(ℝ3n ; Κn(H)) is called a symmetric elementary reaction. The transition configuration of chemical reaction (11) equals the transition configuration of chemical reaction (10) with respect to the reactant. If H= NG(H) then there is not any symmetric elementary reactions.
If NG(H)≠G then there are sub­groups H′ in the group G which are conjugate to the group H (i. e. H′ =sHs-1, s∈G \ NG (H)). Also these sub­groups are bifurcation sub­groups of the group G in the sub­space 𝕍 [1, Lemma 16.3], where dim 𝕍≥2. If H′ =sHs-1, s∈G \ NG (H), then the image vector Κn(s)( p) is an element of the fixed point sub­space Fix(ℝ3 n; Κn(H′ )). If in addition H≠NG(H) then also the image vector Κn(s)( p∗) is an element of the fixed point sub­space Fix(ℝ3 n; Κn(H′ )) so that the chemical reaction

Κn(s)( p)→ Κn(s)( p∗)  (12)

is a symmetric elementary reaction in the fixed point sub­space Fix( ℝ3 n; Κn(H′ )). The symmetric elementary reactions (12) of the symmetry operations s∈G \ NG (H) together with the symmetric elementary reaction (11) form a complex of symmetric elementary reactions. They have a common G-symmetric transition configuration. Since the transition configuration has a descent vector for every fixed point sub­space Fix(ℝ3 n; Κn(H ′)), the reactant configuration is assigned a product configuration in no uniquely defined manner. Thus a complex of symmetric elementary reactions also contains asymmetric elementary reactions of the kind

Κn(s)( p)→ Κn(s') (p),  Κn(s)( p)→ Κn(s')( p∗),  Κn(s)( p∗)→ Κn(s')( p∗).

If NG( H)=G then in the group G there are no sub­groups which are conjugate to the group H. Therefore a complex of symmetric elementary reactions consists only of reaction (11).
If H=NG(H) then the chemical reactions

p→ Κn(s)( p),  Κn(s)( p)→ Κn(s') ( p)  (13)

form a complex of asymmetric elementary reactions. They have a common G-symmetric transition configuration. An asymmetric elementary reaction is always connected with two fixed point sub­spaces, the fixed point sub­space containing the reactant configuration and the fixed point sub­space containing the product configuration.

Also the product configuration of chemical reaction (10) is connected with a complex of elementary reactions like the reactant configuration. The transition configurations of the reactant and the product complex are not equal in general.

3.3 A Model of Chemical Reactions

As above, the symmetry groups of the reactant and the product of a chemical reaction are denoted by H and H', respectively.

Assuming that a chemical reaction (10) satisfies the conditions

the following assertions are true:

(1) If the transition configuration of a chemical reaction is a saddle point of first order then the chemical reaction is of type I or II, and H=H' [1, Theorem 24.1].

(2) A non-elementary chemical reaction of type II consists of either two complexes of symmetric or asymmetric elementary reactions or a complex of symmetric elementary reactions and a complex of asymmetric elementary reactions [1, Theorem 24.2].

(3) If the symmetry group H or H' is a symmetric bifurcation sub­group of the group G (i. e. NG (H)≠H or NG( H') ≠H', respectively) then a chemical reaction of type III is either a narcissistic reaction or a complex of symmetrical elementary reactions. If the symmetry group H or H' is an asymmetric bifurcation sub­group of the group G (i. e. NG (H)=H or NG( H') =H', respectively) then a chemical reaction of type III involves a complex of asymmetric elementary reactions [1, Theorem 24.3].

In the case of a non-elementary chemical reaction of type II or III the transformation of the reactant to the product is a transition of the complex of the elementary reactions of the reactant to the complex of the elementary reactions of the product. The transition configuration of the reactant complex is transformed to the transition configuration of the product complex in the reactive domain. Both transition configurations have the same symmetry. The transformation is connected with a compressing or stretching of the bonds and/or a change of the angles while maintaining the symmetry. The energy values of the transition configurations are different in general.
If the transition configurations of a chemical reaction of type II is a saddle point and if the symmetry groups H and H' are different then the saddle point is at least of order two.

References

[1] Kliesch W, Symmetry and Symmetry Breaking in Chemistry, Walter de Gruyter Berlin/Boston 2025.
[2] Primas H, Müller-Herold U, Elementare Quantenchemie, Teubner Stuttgart 1990.
[3] Hollas JM, Die Symmetrie von Molekülen, Walter de Gruyter Berlin 1975.
[4] Steinborn D, Symmetrie und Struktur in der Chemie, VCH Weinheim 1983.
[5] Eyring H, Polanyi M, Z Physikal Chem B12(1931) 279.
[6] Laidler KJ, King MC, J Phys Chem 87(1983) 2657.
[7] McIver JW, Accounts Chem Res 7(1974) 72.