STOCHASTIC GENERATOR OF CHEMICAL-STRUCTURE .1. APPLICATION TO THE STRUCTURE ELUCIDATION OF LARGE MOLECULES

This paper presents an original computer-assisted structure elucidation system based on a stochastic approach. Using a randomized technique, it is shown that the number of chemical structures that match a set of analytical data can be approximated in a reasonable computational time. Furthermore, it is demonstrated that a sample of three-dimensional models can be generated and be statistically representative of the entire population of potential models. The analytical data introduced in the system can be derived from experimental techniques as diverse as elemental analysis, functional group analysis,H-1, C-13, and Si-29 NMR, mass spectrometry, pyrolysis, gas chromatography, pycnometry, N-2 and CO2 adsorption, mercury intrusion, and SAXS. The stochastic structure elucidation system is applied to macromolecular compounds that are studied in biochemistry (lignin), geochemistry and fuel science (coal), and material sciences (amorphous silica gel). For these compounds as well as for other amorphous chemical structures cited in the paper the proposed stochastic approach is the first technique that correlates a large diversity of analytical data and three-dimensional molecular models.