The different run types are characterized by how the geometry is manipulated:
SinglePoint
The SCF solution is computed for the input geometry.
GeometryOptimization
The atomic coordinates are varied in an attempt to find a (local) energy minimum. One may let all coordinates free or only a subset, keeping the others frozen at their initial values.
TransitionState
Search for a saddle point. Similar to a GeometryOptimization, but now the Hessian at the stationary point presumably has one negative eigenvalue.
LinearTransit
The geometry is modified step by step from an initial to a final configuration. All of the coordinates or only a subset of them may be involved in the transit. The coordinates to be modified are the LinearTransit parameters. For each of the LinearTransit points (geometries) the computation may be a Single Point SCF calculation or a GeometryOptimization. In the latter case only those coordinates (or a subset of them) can be optimized that are not LinearTransit parameters. The LinearTransit feature can be used for instance to sketch an approximate reaction path in order to obtain a reasonable guess for a transition state, from where a true TransitionState search can be started.
IRC or IntrinsicReactionCoordinate
Tracing a reaction path from a transition state to reactants and/or products. A fair approximation of the transition state must be input. The end-point(s) - reactants / products - are determined automatically.
Frequencies
Computation of force constants and from these the normal vibrational modes and harmonic frequencies. The force constants can be calculated by numerical differentiation of the energy gradients at the equilibrium geometry and the slightly deviating geometries (making small displacements of the atoms). There is however also a possibility with the post-ADF program SD, to calculate the second derivatives analytically. This should usually be faster and in some cases more robust (no SCF convergence problems in displaced geometries, as only a single SCF is required). Note that this program still has some limitations. Most importantly, it can only handle X$alpha and VWN (LDA), but not GGA, calculations at this moment.
CINEB
Calculation of the reaction path and transition state search using Climbing-Image Nudged Elastic Band method. This method is further referrer to as NEB or CI-NEB. Using this method one can find a transition state between two known states, further referred to as initial and final states. The choice which state is initial and which is final is arbitrary. During calculation with this method, a number of replicas, or images, of the system is calculated. These images can be considered as being linked by an elastic band. Each image is optimized in such a way that on each step the forces parallel to the reaction path are removed and spring forces are added that keep distances to this image's neighbors equal. At the end of the optimization the images are evenly distributed along the reaction path, the image highest in energy being the transition state (if the climbing-image option is on, the default).
For all features that involve changes in geometry, i.e. all run types except the SinglePoint, it is imperative that you use single-atom fragments. Larger molecular fragments can only be applied in SinglePoint calculations.
Four keys are involved in the specification of the geometry and its manipulation:
atoms
sets the atomic (starting) positions.
geometry
Controls the run type and strategy parameters, such as convergence thresholds and the maximum number of geometry steps to carry out.
atoms and geometry
These two keys together are sufficient for a straightforward Optimization, TransitionState search, IRC run or a Frequencies computation. (Of course, you also need to specify the Fragments or BASIS key.
constraints
May be used to impose constraints for geometry optimizations, LT, and TS, in the new branch for optimization, which is the default for these optimizations. This key can not be used for IRC, NEB, or for the old branch of optimizations.
geovar
May be used to impose constraints, for instance when only a subset of all coordinates should be optimized. This key should be used for IRC, NEB, or for the old branch of optimizations, to impose constraints. GeoVar may also be used in a LinearTransit or NEB run to define the LinearTransit or NEB parameters, respectively, and their initial and final values.
Constraints and LinearTransit parameters in the old branch of optimizations may also be controlled within the atoms block if a MOPAC-style input format is used, see below.
Runtype control and strategy parameters
