Researchers at Dalian Minzu University and Harbin Institute of Technology have developed a machine learning-guided single-photon upconversion strategy based on cascade pumping, according to a study published in Nature Communications.
The technique uses a photon-stacking process that turns the intermediate state of lanthanide ions into a virtual ground state. This allows direct single-photon pumping to target energy levels. Photon upconversion in infrared photonics typically faces limits from narrow spectral response ranges, low efficiency, and slow operational speeds. In proof-of-concept tests using NaYS2:Ho3+, the team increased selective upconversion emissions by two to three orders of magnitude through precise population control.
Testing showed the efficient upconversion response range extended to approximately 2100 nanometers while reducing response time from 30 milliseconds down to 54 microseconds. The authors confirmed the strategy generalizes to other lanthanide ions, including thulium, praseodymium, and erbium. Optimizing energy transfer in holmium-sensitized systems produced near-pure red, green, and blue light emissions.
In gas sensing experiments, the team demonstrated low-threshold carbon dioxide detection using narrowband photodetection, achieving a sensitivity of 6.4×10−4 ppm−1. Qi Xiao, Wen Xu, Xiumei Yin, Na Zhou, Xinyao Dong, Ge Zhu, Xixian Luo, and Bin Dong co-authored the paper, which was received in March 2026 and published on July 31, 2026. The research received support from the National Key Research and Development Program of China, the National Natural Science Foundation of China, and regional research foundations in Dalian and Liaoning Province.
