Recently, demand of infrared phosphors is increasing because they can be used not only for lighting applications such as night vision camera and plant growth but also in various fields such as biomarkers that emit light by binding to abnormal sites in living organisms. Since the Ni2+-doped phosphors are drawing attentions as potential infrared phosphors, analysis of multiplet energies and emission mechanism of Ni2+ in crystals is very important for the development of novel infrared phosphors. For Mn4+, Cr3+ and Ni2+ in crystals, Brik et al. analyzed a large amount of experimental data in literature and indicated the importance of the Racah parameter C in addition to B and proposed a new Nephelauxetic parameter. Considering their results, we have recently created prediction models of 2Eg levels of Mn4+ and Cr3+ in oxides and fluorides based on the Racah parameters B and C using machine learning.For theoretical design of infrared phosphors, a similar prediction model for 1Eg level of Ni2+ would be useful. According to the Tanabe-Sugano diagram for the d8 configuration, the energy of 1Eg is almost independent of Δ/B which is a measure of the strength of the ligand field normalized by the Racah parameter B. Therefore, the energy of 1Eg seems to be determined by the Racah parameter B. However, as pointed out by Brik et al., in actual experimental values in literature, there are Ni2+-doped crystals which have different 1Eg energies despite having almost the same B values, indicating the necessity of consideration of the Racah parameter C as well. In this work, we created systematic model clusters for Ni2+ ions in crystals and performed first-principles calculations of the multiplet energies using the DV-Xα and DVME methods. Using these systematic data as the training data, predictive models of 1Eg based on the Racah parameters B and C were created by machine learning and the influence of these parameters on the multiplet energies was investigated in detail.
Read full abstract