Chemometric methods in structure elucidation with MS and IR

Research project P 14792 Chemometric methods in structrue elucidation with MS and IR Kurt VARMUZA 09.10.2000 Central problems in chemistry are the at on of chemical substances in samples and the chemical structure elucidation. Both problems are related: identification refers to already known compounds with appropriate data available, structure elucidation is the more challenging part and refers to new compounds - for instance from chemical syntheses in laboratories or found in biological, environmental or food samples. Chemical structures cannot be determined directly by instruments. Modem chemistry uses experimental spectroscopy for substance identification and structure elucidation, mainly nuclear magnet resonance spectra (NMR), mass spectra (MS), and infrared spectra (IR). However, the relationships between spectral data measured from a compound and its chemical structure are very complex and cannot be described sufficiently by rules or algorithms that are applicable to practical problems. Although parts of the spectra interpretation process are already supported by computerized methods still a wide field of unexplored methods and ideas exist. The aim of the project is a significant contribution to cornputer-assisted spectra interpretation of MS and IR- These two spectroscopic methods have been chosen for the project because they can be applied for the analysis of substances present only in very low concentrations, and because of the experience, databases and software available in the laboratory of the applicant. The principal concepts of the project comprise the use of large spectroscopic and structural databases, the development of new spectra similarity searches ` the application of multivariate data analysis including neural networks and genetic algorithms, and utilization of today`s high performance of personal computers. Besides the basic research an important goal is the realization of a software prototype dedicated to practical use in laboratories. It is expected that this software will be of substantial help in cases currently available methods fail, especially for a structureoriented evaluation of spectra sin-Oarity searches, exploratory investigations of spectra-structure relationships, and the automatic recognition of chemical substance classes which in combination with isomer generation leads to a systematic structure elucidation. For important parts of the project existing collaborations with other research groups will be intensified: mathematicians at the University of Bayreuth (Germany), spectroscopists at the University of Plovdiv (Bulgaria), and analytical chemists at the University of Vienna.