In ACS applied materials & interfaces ; h5-index 147.0
High-efficiency long-wavelength emission near-infrared (NIR) phosphors are the key to next-generation LED light sources. However, high-efficiency phosphors usually exhibit narrow-band emission at shorter wavelengths due to the crystal field intensity. In this paper, we utilize multi-objective optimization to discover the NIR phosphor Gd3Mg0.5Al1.5Ga2.5Ge0.5O12:0.04Cr3+. It exhibits a broadband NIR emission from 650 to 1100 nm peaking at 763 nm, with a full width at half maximum (FWHM) of 150 nm, an internal quantum efficiency (IQE)/external quantum efficiency (EQE) of 90%/53.1%, and good thermal stability (85.3% @ 150 °C). The packaging results show that ∼53.2 mW of output power is achieved at 915 mW input power, which suggests promising applications for NIR pc-LED. Our approach is based on the data of emission wavelength (WL) and IQE for garnet:Cr NIR phosphors to construct machine learning models. An active learning strategy is used to make tradeoffs between WL and IQE, and we are able to find the targeted phosphor after only four iterations of synthesis and characterization.
Jiang Lipeng, Jiang Xue, Wang Changxin, Liu Pei, Zhang Yan, Lv Guocai, Lookman Turab, Su Yanjing
2022-Nov-09
Cr3+, garnet, machine learning, multi-objective optimization, near-infrared
