The team led by Prof. Gengtao Fu from Nanjing Normal University has successfully enhanced the electrocatalytic oxidation performance of 5-hydroxymethylfurfural (HMF) by employing a rare-earth Cerium (Ce) doping strategy to tune the geometric and electronic structure of a CuO catalyst. This catalyst achieves efficient conversion of HMF to FDCA with near-theoretical-limit Faradaic efficiency (98.4%) and high production rate. By enabling optimal configuration matching, it effectively reduces steric hindrance, offering a novel approach for biomass electrocatalytic conversion.

A universal potential for all-purpose atomic simulations has been pursued for decades, but remains challenging due to limitations in model expressiveness and dataset construction. Now, writing in the journal Science China Chemistry, the research team led by Professor Zhi-Pan Liu at Fudan University has developed a generalized global neural network potential (GG-NN) based on a newly proposed low-cost, high-accuracy High-order Pair-reduced Neural Network (HPNN) model and a comprehensive global potential energy surface dataset of 5.84 million configurations covering 83 elements. GG-NN achieves high-precision predictions for both energies and forces, offering the potential to significantly accelerate the prediction and design of molecular and materials systems.

Biochar has been widely promoted as a sustainable solution for improving soil health, cleaning polluted water, and capturing carbon. However, a comprehensive review warns that its environmental safety may be more complex than previously thought.

Researchers from Koç University have developed a new artificial intelligence model that predicts where people direct their attention in 360-degree videos by integrating both audio and visual cues, addressing a key challenge in immersive media where viewers are free to explore their surroundings rather than being guided by a fixed camera frame; unlike conventional approaches that rely mainly on visual information, the model jointly analyzes sound and image data to better reflect how humans perceive complex, dynamic environments, resulting in more accurate attention prediction in virtual reality and 360-degree video settings, with potential applications ranging from immersive media design and video production to human–computer interaction and content optimization.

Osaka Metropolitan University researchers proposed a geometry-aware framework for automatically removing human noise from TLS point clouds by projecting 2D instance segmentation masks into 3D space and validating candidates through multi-stage geometric filtering.