Investigation of the Metal Carbonyl Bonding Problem by Pattern Recognition.
Abstract
The application of pattern recognition techniques is expanded to the analysis of bonding models. Supervised learning techniques are used to compare sigma and pi models of bonding in metal carbonyl systems. The three symmetry allowed C = O stretching frequencies and six other ligand and metal properties for each of 67 compounds of the type L - M(CO)5 provide evidence strongly supporting a sigma bonding model to explain variations in carbonyl stretching frequencies. Unsupervised learning techniques provide evidence that both models are inadequate but that carbonyl stretching frequencies may be predicted from simple ligand and metal properties. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1974
- Accession Number
- AD0781291
Entities
People
- Bruce R. Kowalski
- Douglas S. Dierdorf
Organizations
- University of Washington