## Qualitative and quantitative tools for predicting the
products and mechanisms of chemical reactions

**
Lundi 11 décembre 2008, 11**^{h}00

Salle Raphael 1, site Ivry-sur-Seine

**Paul W. Ayers**

Department of Chemistry, McMaster University, Hamilton, Canada

Given a set of reagents in a specific reaction environment, does a reaction
occur and, if so, what are the products and what is the mechanism? This talk
will feature two techniques for predicting chemical reagents' reactivity. At
a qualitative level, recent advances in conceptual density-functional theory
(DFT) will be discussed. Unlike most other conceptual tools for elucidating
the reactivity of a molecule, the reactivity predictors from conceptual DFT
lie within the framework of a formally exact quantitative theory (DFT). A
particularly useful approach to conceptual DFT is what may be termed the
"perturbative" perspective on chemical reactivity, wherein the reactivity of a
reagent is described by measuring its response to a small number of different
"model perturbations", which are chosen to simulate different types of reagents
(hard acids, soft bases, etc.). The perturbative perspective provides a
unifying framework for disparate DFT-based reactivity indices. Here, I'll
discuss applications of the perturbative perspective to situations where naïve
frontier molecular orbital theory fails.
At a quantitative level, I will discuss recent work on finding reaction
coordinates on multi-dimensional potential energy surfaces. Most existing
methods for determining reaction coordinates require prior knowledge of the
reactants, products, and.for multi-step reactions.reaction intermediates. (In
addition, knowledge of the transition state structures is frequently helpful,
and sometimes required.) The approach I'll describe is based on the
"fast-marching" algorithm commonly used for front-propagation problems, and is
distinguished from most conventional methods because it only requires
knowledge of the reactants. Reactive intermediates, transitions states, and
products are output from, rather than input to, the computational method.
Consequently, the fast marching method provides a rigorous mathematical
approach to predicting not only the products, but also the mechanism, of a
chemical reaction.