Solid state light (SSL) bulbs are rapidly replacing incandescent and fluorescent lighting due to their high efficiency and long operating lifetimes. SSLs generate white light by combining a blue or near-UV LED with a rare-earth doped inorganic phosphor, which partially or fully down-converts the LED emission to longer wavelengths. The search for new phosphors used for this application has recently centered on identifying inorganic compounds that form rigid crystal structures, as probed by their Debye temperature (ΘD). Materials with a high ΘD tend to have decreased non-radiative relaxation and consequently a high photoluminescent quantum yield (PLQY). In addition to a high ΘD, these crystal structures also require a wide electronic band gap (Eg) for the rare-earth ion to down-convert the absorbed photons to the visible light region. This project, therefore, is developing a procedure to optimize these two material parameters by using a combination of machine learning and electronic structure calculations. Here, over 2,000 Debye temperatures are estimated using a support vector machine (SVM) model and their band gaps obtained by density functional theory (DFT). Materials with the highest ΘD and largest Eg are then selected for experimental analysis. Among these compounds, NaBaB9O15 shows an outstanding ΘD (729 K) and a wide Eg (5.5 eV). We subsequently synthesized this phase with Eu2+ substitution on the Ba2+ sites using high temperature solid state reactions. The pure phase product was characterized with high-resolution synchrotron X-ray powder diffraction while photoexcitation and emission studies reveal deep blue emission when excited by UV-light with a maximum PLQY of more than 70% and excellent thermal stability. These results support our proposed machine learning approach to screen new phosphors for use in solid state white lighting.

Accelerating the discovery of inorganic phosphors with the assistance of machine learning

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.