For examples of data-sets for solids, please see: SOLIDS
A solid-state data set can be identified by the presence of three "Tv" atoms at the end of the geometry. A working definition would be as follows:
The geometry used represents a unit cell that can be translated in three dimensions to form an infinite crystal. This means that the unit cell is a volume defined by three sets of two planes. No space-group symmetry can be used. The unit cell used must be large enough to hold a sphere of radius 4 Angstroms. This is not a problem when the primitive unit cell is large, but if it is small then two or more primitive unit cells must be put together to make a unit cell that is large enough to hold the sphere. Thus in rock salt, halite, the primitive unit cell is a cube of size 5.64 Angstroms. In order to hold a sphere of diameter 8 Angstroms, eight primitive unit cells would need to be used. This would give rise to a unit cell of edge 11.28 Angstroms.
The unit cell should be defined in Cartesian coordinates, with atom 1 at the origin, that is, at coordinates (0.0, 0.0, 0.0).
The translation vectors are represented in a unique way. At first, it might appear unnecessarily complicated, but in practice it is quite easy to use.
An atom in the unit cell is selected. The choice of atom is not important computationally, but can help in visualizing the structure later on. For example, if 1-Chloronaphthalene were being calculated, the chlorine atom would be an obvious choice, as it is easily located in the solid. This atom is put at the start of the geometry, that is, it is defined as atom 1. Its coordinates are
Let the unit cell used be defined as lattice point (0,0,0), then the three translation vectors are defined by lattice points (1,0,0), (0,1,0), and (0,0,1). Each translation vector will move atom 1 to the corresponding position in the adjacent unit cell. These positions are defined using the symbols "Tv". That is, the position of each Tv is defined as being where atom 1 would be if it were translated through one unit cell.
A good starting point is to begin with a cif file and use MERCURY to generate a starting geometry. Open the cif file with MERCURY, and set up a 2 by 2 by 2 set of lattice points by using "Calculate" => "Packing/Slicing" => "Pack" then change the "1.0" to "2.0" for "a", "b", and "c". Save the resulting structure as an "ent" file.
Open the "ent" file in a graphics package, e.g., CAChe, and identify the atom to be used as atom 1. Re-arrange the atoms so that the atom selected is atom 1.
Identify the three atoms that atom 1 would become it it were translated through one lattice point in each of the three directions. Rename these atoms "Tv".
Delete all atoms that are not in lattice point (0,0,0). This might be tedious.
Save the file as Cartesian coordinates.
Edit the Cartesian coordinates to add optimization flags.
The lattice parameters a, b, and c, a, b, and g, are not used directly, but are used by MERCURY in generating the lattice points.
Add keywords, title and comment, and run the data set using MOPAC.