Structure prediction from infrared spectra

The modeling approach is generally useful if the query structure is not contained in the database. Although the methods for structure prediction from infrared spectra introduced here can give quite accurate proposals, several considerations have to be taken into account. 

We have already met one tool that can be used to investigate the links that exist among data items. When the features of a pattern, such as the infrared absorption spectrum of a sample, and information about the class to which it belongs, such as the presence in the molecule of a particular functional group, are known, feedforward neural networks can create a computational model that allows the class to be predicted from the spectrum. These networks might be effective tools to predict suitable protective glove material from a knowledge of molecular structure, but they cannot be used if the classes to which samples in the database are unknown because, in that case, a conventional neural network cannot be trained. 

In a surprising recent development, Kratschmer et al. have shown that certain all-carbon molecules are produced in large quantities in the evaporation of graphite and can be isolated as soluble, well-defined solids. The major species was identiHed as molecular C ) through mass spectrometry and by comparison of the infrared spectrum with theoretical predictions for the celebrated truncated-icosahedron structure, which had earlier been proposed to account for cluster beam observations. The solid material, described as a new form of elemental carbon in a nearly pure state, has a disordered hep lattice of packed quasi-spherical molecules, but determination of the precise molecular structure awaits diffraction from well-ordered crystals. 

FIGURE 6.4 (continued) (c) Derivation of the 3D structure of a compound from its infrared spectrum. After training, the query infrared spectrum is used to predict the RDF descriptor, and a structure database is searched for the most similar descriptor. The corresponding structure is retrieved as the initial model. 

Database Approach is a specific method for deriving the molecular structure from an infrared spectrum by predicting a molecular descriptor from an artificial neural network and retrieving the structure with the most similar descriptor from a structure database. 

Though this is not an acid salt, and does not even involve any carboxyl groups, its structure is analogous to that of a Type A acid salt, as was predicted by Hadzi and Novak 36) from its infrared spectrum, and confirmed by X-ray work 37). 

The far-infrared spectrum of Rh4(CO)i2 has been measured (7) and the four bands observed have been assigned as 225 cm", c 200 cm , e 176 cm , Uj and 128 cm, a. These bands are not markedly different from those observed for Ir4(CO)i2 334) despite their significantly different crystal structures 377). The single force constant, however, obtained from calculations based on a pseudotetrahedral structure could not satisfactorily predict the band positions. Using an M M model, values of 0.95 mdyne/A were obtained for the basal stretching force constant and 1.35 mdyne/A for the M—M force constant. 

The infrared spectrum of Fc2(CO), was first obtained by Sheline and Pitzer, who observed two terminal (2C80 and 2034 cm ) and one bridging (1828 cm ) CO stretching bands. This result agrees with that expected from structure VI, determined by X-ray analysis. Again according to X-ray analysis/ the structure of Mn2(CO),o is that shown by VII of Fig. 111-40. This D4j structure predicts four Raman- and three infrared-active CO stretching bands. 

Molecular dynamics attempts to solve the dynamically evolving ensemble of molecules given the interactions between molecules. The form of the forces between molecules or atoms, the number of interactions (i.e., two- or three-body interactions), and the number of molecules that can be tackled by the program determine the success of the model. Molecular dynamics simulations can predict the internal energy, heat capacity, viscosity, and infrared spectrum of the studied compound and form an integral part in the determination and refinement of structures from X-ray crystallography or nuclear magnetic resonance (NMR) experiments. 

RDF descriptors exhibit a series of unique properties that correlate well with the similarity of structure models. Thus, it would be possible to retrieve a similar molecular model from a descriptor database by selecting the most similar descriptor. It sounds strange to use again a database retrieval method to elucidate the structure, and the question lies at hand Why not directly use an infrared spectra database The answer is simple. Spectral library identification is extremely limited with respect to about 28 million chemical compounds reported in the literature and only about 150,000 spectra available in the largest commercial database. However, in most cases scientists work in a well-defined area of structural chemistry. Structure identification can then be restricted to special databases that already exist. The advantage of the prediction of a descriptor and a subsequent search in a descriptor database is that we can enhance the descriptor database easily with any arbitrary compound, whether or not a corresponding spectrum exists. Thus, the structure space can be enhanced arbitrarily, or extrapolated, whereas the spectrum space is limited. 

It should be noted, however, that this method does not give a clear-cut answer if the predicted numbers of infrared- and Raman-active fundamentals are similar for various probable structures. Furthermore, a practical difhculty arises in determining the number of fundamentals from the observed spectrum, since the intensities of overtone and combination bands are sometimes comparable to those of fundamentals when they appear as satellite bands of the fundamental. This is panicularly true when overione and combination bands are enhanced anomalously by Fermi resonance (accidental degeneracy). For example, the frequency of the first overtone of the 1/2 vibration of CO2 (667cm ) is very close to that of the > vibration (1337 cm ). Since these two vibrations belong to the same symmetry species ( p, they interact with... 

---