A systematic scheme to split the volume of molecular crystals into additive increments is discussed. In contrast to earlier procedures, it relies on the definition of atom types on the basis of their geometrical rather than chemical environment. In addition, the role of the relevant structural features of the compounds is explicitly taken into account. This approach provides insight into the relative influence of chemical bonds, hydrogen bonds and rings on the volume of organic crystals. Compared with group-contribution techniques, it yields very similar results with many fewer empirical parameters. Applied to estimate the densities of 42 880 crystals containing elements up to chlorine and measured at different temperatures, an average absolute deviation from experiment close to 2% is obtained.
A method to estimate the lattice energies E(latt) of nitrate salts is put forward. First, E(latt) is approximated by its electrostatic component E(elec). Then, E(elec) is correlated with Mulliken atomic charges calculated on the species that make up the crystal, using a simple equation involving two empirical parameters. The latter are fitted against point charge estimates of E(elec) computed on available X-ray structures of nitrate crystals. The correlation thus obtained yields lattice energies within 0.5 kJ/g from point charge values. A further assessment of the method against experimental data suggests that the main source of error arises from the point charge approximation.
In view of its interest for high pressure simulations of molecular crystals, the self-consistent field density functional tight-binding model is coupled to a symmetry-constrained optimization routine and applied to a stringent test case: the open structure of the acetonitrile crystal. The symmetry-constrained relaxation is compared with two symmetry-unconstrained relaxations, one starting from the observed X-ray structure and the other one from a structure obtained after a optimization of the atoms positions within the observed unit cell. According to the procedure employed for the crystal relaxation, three different energy minima with similar energies are found.
A method to estimate heats of formation of ionic molecular crystals prior to their actual synthesis is put forward. It relies on systematic searches for stable polymorphs through ion packing, explicit evaluations of the cohesive energies using these computer-generated crystal structures, and separate density-functional estimations of the gas-phase contributions. Preliminary results indicate that suitable trial structures are readily found even with a moderate sampling of the search space. Moreover, the density functional procedure employed turns out to be satisfactory for charged species. On the other hand, using observed crystal geometries, a point-charge model yields acceptable values for the cohesive energies. Finally, the need for a transferable force field to relax the trial structures appears to be the most critical point.
Standard group volumes which can be used to estimate the crystal densities of molecular salts and hydrates are reported, as a complement to values derived recently for the functional groups of neutral organic compounds. These new parameters were derived from a least-squares fit of cell volumes for a set of 1132 ionic molecular crystals from the Cambridge Structural Database. Their values point to the unusual overlap between monovalent O atoms and neighbouring H atoms. Using the new group volumes presently obtained, the crystal densities of the salts are predicted with an average error of <2.5%, while previous atom-based schemes yield average errors of >3%. To illustrate the possible application of the present database, the problem of designing environmentally friendly propellants is addressed.