QSPR Models for NMR Spectra

In the past, the interpretation of the NMR spectra was based on empirical correlations derived from the study of crystalline materials with known stiucture, [62] and, successively, on correlations with stmctural descriptors computed by ab-initio calculations on crystals or model clusters [63-68]. However, crystaUine systems generally exhibit a limited diversity of local stiuctures in contrast to the disorder and variety of stmctural units (different coordination numbers and Q distributions) present in multicomponent glasses. Moreover, the cluster approach does not account for the correlations between stmctural factors that exist in solids and disorder in glasses [69-71]. 

To overcome these limitations, several studies focused on semi-quantitative comparisons between information derived from NMR spectra and stmctural features obtained from molecular dynamics simulations on large glass samples. They make use of connectivity between different types of Q species and the related 

7 The structural inversion QSPR approach to extract structural distributions from NMR data. The example of the Si-O-T distribution for sodium silicate glasses is given [87] 

Alkaline/alkaline earth silicate [85, 87, 89-92] 5iso of each Q species Mean Si-O-T angle (T denotes the Q connected tetrahedron) 

---