Machine Learning Guided Design of Single–Phase Hybrid Lead Halide White Phosphors

Designing new single‐phase white phosphors for solid‐state lighting is a challenging trial–error process as it requires to navigate in a multidimensional space (composition of the host matrix/dopants, experimental conditions, etc.). Thus, no single‐phase white phosphor has ever been reported to exhibit both a high color rendering index (CRI ‐ degree to which objects appear natural under the white illumination) and a tunable correlated color temperature (CCT). In this article, a novel strategy consisting in iterating syntheses, characterizations, and machine learning (ML) models to design such white phosphors is demonstrated. With the guidance of ML models, a series of luminescent hybrid lead halides with ultra‐high color rendering (above 92) mimicking the light of the sunrise/sunset (CCT = 3200 K), morning/afternoon (CCT = 4200 K), midday (CCT = 5500 K), full sun (CCT = 6500K), as well as an overcast sky (CCT = 7000 K) are precisely designed. Through the guidance of machine learning tools, the host composition and dopants concentration of hybrid lead halides can be finely controlled to design a series of single‐phase white phosphors with ultra‐high color rendering (above 92) mimicking the light of the sunrise/sunset, morning/afternoon, midday, full sun, as well as an overcast sky.