Abstract. This is an overview of the use of empirical force fields in the study of reaction mechanisms. Em- pirical-valence-bond-type methods (including reactive force field and multiconfigurational molecular mechan- ics) produce full reaction surfaces by mixing, in the simplest case, known force fields describing reactants and products. The SEAM method instead locates ap- proximate transition structures by energy minimization along the intersection of the component force fields. The transition-state force-field approach (including Q2MM) designs a new force field mimicking the transition structure as an energy minimum. The scope and appli- cability of the various methods are compared. Keywords: Molecular mechanics – Transition states – Empirical force fields 1 Introduction Molecular modeling based on empirical force fields is today a mature field with applications in many areas of chemistry. The primary uses of force fields are to predict equilibrium structures, to perform conformational sam- pling, and for molecular dynamics. With well-chosen parameters, it is possible to determine distortion energies and barriers to conformational change, for example, bond rotation barriers, with good accuracy. Using current modeling software with graphical user interfaces, these tasks no longer require expertise in computational methods, but can be used as black-box techniques. Basic modeling is also entering the curriculum in more and more chemical educational institutions, even at under- graduate levels. Molecular mechanics started out as a method for determining structures and conformational energies of simple organic molecules in vacuo. In this overview, the term ‘‘molecular mechanics’’ will be used interchange- ably with ‘‘empirical force fields’’. Owing to its success in this area, the methodology has been augmented to allow application in other areas of chemistry, for example, calculations of heat of formation data to allow com- parison of different molecules [1] and application to metal complexes [2, 3]. Empirical force fields are also widely used for simulations in condensed phases [4], because they allow sampling of a large number of con- figurations. Some of these methods may be considered mature, and are available in contemporary software, but applications beyond the area of simple organics require some experience to evaluate and validate the results. The use of molecular mechanics for describing transi- tion states (TSs)/structures is not yet a black-box technique. A transition structure is the geometry of a first- order saddle point on a potential-energy surface (PES). A TS is the hypersurface in phase-space dividing reactant and product, commonly taken to be an ensemble of mol- ecules with the geometry of a transition structure at a given temperature. Often the two concepts are used in- terchangeably without a clear distinction. Since molecular mechanics relies on explicit bonding information, TSs may be divided into two groups: conformational changes (bond rotations and atom inversions) and chemical reac- tions (bond breaking/formation). The first type poses no new fundamental problems for molecular mechanics methods, beside those inherent in the force-field parame- Correspondence to: P.-O. Norrby e-mail: pon@kemi.dtu.dk Feature article Transition states from empirical force fields Frank Jensen 1 , Per-Ola Norrby 2 1 Department of Chemistry, University of Southern Denmark, Campusvej 55, 5230 Odense, Denmark 2 Department of Chemistry, Organic Chemistry, Technical University of Denmark, Building 201, Kemitorvet, 2800 Lyngby, Denmark Received: 17 April 2002 / Accepted: 26 July 2002 / Published online: 4 November 2002 Ó Springer-Verlag 2002 Theor Chem Acc (2003) 109:1–7 DOI 10.1007/s00214-002-0382-6