Genome-based approach to new Nd: Phosphate glass: Composition-structure-property statistical modeling and validation

Abstract

Modeling statistical relationships between the network structural groups (S) derived from Fourier transform infrared (FTIR) spectroscopic analysis and various properties (P) is discussed in this article, focusing on a wide range of K2O, MgO, and Al2O3 in a single baseline Nd: phosphate glass composition (C). Properties covered by this study are emission cross-section, effective linewidth, fluorescence lifetime, liquidus temperature, coefficient of thermal expansion, and nonlinear refractive index. Both S-P models and C-S models were established; the former demonstrated sufficiently high predictability (R2>0.97 and Radj2>0.96) for all cases. In designing a new glass composition that simultaneously meets multiple target properties, especially for dealing with nonlinear property responses to composition variations, the final C-S-P database is constructed by combining S-P and C-S models. Moreover, the C-S-P validation tests were performed by using “bi-directional” modeling steps, P→S→C and C→S→P.