LAUSR

Abstract In the study of Nd:phosphate laser glass modified by introducing Li2O (0–8 mol%) and Na2O (0–8 mol%), respectively, glass transition temperature (Tg), thermal expansion coefficient (α), nonlinear refractive index (n2) and effective linewidth (Δλeff) were measured. Statistical based composition-structure-property (C-S-P) models were systematically developed. For the single-component study, a limitation of the C-P model approach is first discussed, i.e., which cannot adequately simulate glass property responses. To overcome the C-P mode limitation from the single-component study, C-S and S-P models were subsequently explored. The alternative methodology, transforms a problem of a given glass property response to a single (one dimensional) oxide change to a problem of the same property response to multi-directional glass structural changes for building a corresponding statistical S-P model. The same methodology is also applied to build C-S model. Plausible glass network structural units (or distributions) were derived from measured Fourier transform Infra-red (FTIR) spectra aided by using a standard curve-fitting method. As a result, S-P models showed significant improvements in simulating the measured glass properties. The C-S-P models, as a glass design tool box, were validated by using four designed glasses in the current independent design of experiment (DOE). For the four DOE derived glass compositions, the measured Li2O and Na2O concentrations and measured properties of the four glasses were very close to the C-S-P model predictions within the experimental errors, except for two cases where differences of Tg were significantly large.