In most cases in the literature a hydrogen bond is either defined trough the distance between an acceptor and a donor atom [see for example Hamilton and Ibers] or by a cut off criteria which includes all atoms below a certain distance and bond angle and exclude everything else.
The first definition is especially useful in protein structures, as the hydrogen positions are not resolved here. Nevertheless it is important to know that a short distance between a donor and an acceptor is only a side-effect of the hydrogen bond, allowing two electronegative atoms to come closer together as normally investigated.
False hits may arise, if other forces produce the same effect and this way mimic a hydrogen bond. Problems arise for very weak hydrogen bonds and for furcated hydrogen bonds.
Cut off criteria are based on the analysis of many crystal structures. They show, that there is a gap for O-H...O distances around 2.2 Å for the H...O distance, where not many hits are observed. But the pictures in the last chapter showed, that distances and angles around hydrogen change continually, even if some configurations are more common than others. There is also no physical reason for a certain cut off.
In a crystal (and in principle also in solvents, but motion makes the problem here even more complicated) there is a simple way, how to define, if two atoms are coordinated to each other and in this case bonded to each other or not.
Everything necessary, is a complete division of a crystal in atomic fragments and some rules which tell, if two neighbouring fragments are bonded to each other or not.
Three different models will be presented, which are capable of this task:
Last Updated: 26 October 1996