Search for: All records

Structurally stabilized composites are promising for using phase change materials in high‐temperature thermal energy storage (TES). However, conventional skeleton materials, which typically comprise 30–50 wt% of the composite, mainly provide sensible heat storage and contribute minimally to overall energy density. This study introduces a new class of redox‐active oxide‐molten salt (ROMS) composites that overcome this limitation by combining sensible, latent, and thermochemical heat storage in a single particle. Specifically, porous, redox‐active Ca₂AlMnO_5+δ(CAM) complex oxide particles were demonstrated as a suitable support matrix, with the pores filled by eutectic NaCl/CaCl₂salt. X‐ray diffraction confirms excellent phase compatibility between CAM and the salt. Scanning electron microscopy/energy dispersive X‐ray spectroscopy and nano X‐ray tomography show good salt infiltration and wettability within the CAM pores. Thermogravimetric analysis reveals that a 60 wt% CAM/40 wt% salt composite achieves an energy density of 267 kJ kg⁻¹over a narrow 150 °C window, with ≈50 kJ kg⁻¹from thermochemical storage. Additionally, the composite shows higher thermal conductivity than salt alone, enabling faster energy storage and release. ROMS composites thus represent a novel and efficient solution for high‐performance TES. 

It is challenging to deploy 3D Convolutional Neural Networks (3D CNNs) on mobile devices, specifically if both real-time execution and high inference accuracy are in demand, because the increasingly large model size and complex model structure of 3D CNNs usually require tremendous computation and memory resources. Weight pruning is proposed to mitigate this challenge. However, existing pruning is either not compatible with modern parallel architectures, resulting in long inference latency or subject to significant accuracy degradation. This paper proposes an end-to-end 3D CNN acceleration framework based on pruning/compilation co-design called Mobile-3DCNN that consists of two parts: a novel, fine-grained structured pruning enhanced by a prune/Winograd adaptive selection (that is mobile-hardware-friendly and can achieve high pruning accuracy), and a set of compiler optimization and code generation techniques enabled by our pruning (to fully transform the pruning benefit to real performance gains). The evaluation demonstrates that Mobile-3DCNN outperforms state-of-the-art end-to-end DNN acceleration frameworks that support 3D CNN execution on mobile devices, Alibaba Mobile Neural Networks and Pytorch-Mobile with speedup up to 34 × with minor accuracy degradation, proving it is possible to execute high-accuracy large 3D CNNs on mobile devices in real-time (or even ultra-real-time). 

Orbital current has attracted significant attention in recent years due to its potential for energy-efficient magnetization control without the need for materials with strong spin–orbit coupling. However, the fundamental mechanisms governing orbital transport remain elusive. In this study, we systematically explore orbital transport in Ti/Ni bilayers through orbital pumping, drawing an analogy to spin pumping. The orbital current is generated and injected into the Ti layer via the microwave-driven orbital dynamics in Ni, facilitated by its strong spin–orbit correlation. We employed thickness-dependent ferromagnetic resonance measurements and angular-dependent inverse orbital Hall effect (IOHE) detection to probe orbital transport in Ti based on the conventional spin-pumping methodology. The observed enhancement in the damping factor indicates an orbital-diffusion length of ∼5.3 ± 3.7 nm, while IOHE-based estimation suggests a value of around 4.0 ± 1.2 nm, which confirms its short orbital-diffusion length. Furthermore, oblique Hanle measurements in the longitudinal configuration reveal an orbital relaxation time of approximately 16 ps. Our results establish that orbital pumping, analogous to the conventional spin-pumping framework, can serve as a robust technique for elucidating orbital transport mechanisms, paving the way for the design of efficient spin-orbitronic devices. 

Graph Neural Networks (GNNs) have excelled in diverse applications due to their outstanding predictive performance, yet they often overlook fairness considerations, prompting numerous recent efforts to address this societal concern. However, most fair GNNs assume complete demographics by design, which is impractical in most real-world socially sensitive applications due to privacy, legal, or regulatory restrictions. For example, the Consumer Financial Protection Bureau (CFPB) mandates that creditors ensure fairness without requesting or collecting information about an applicant’s race, religion, nationality, sex, or other demographics. To this end, this paper proposes fairGNN-WOD, a first-of-its-kind framework that considers mitigating unfairness in graph learning without using demographic information. In addition, this paper provides a theoretical perspective on analyzing bias in node representations and establishes the relationship between utility and fairness objectives. Experiments on three real-world graph datasets illustrate that fairGNN-WOD outperforms state-of-the-art baselines in achieving fairness but also maintains comparable prediction performance. 

Searching for land refugia becomes imperative for human survival during the hypothetical sixth mass extinction. Studying past comparable crises can offer insights, but there is no fossil evidence of diverse megafloral ecosystems surviving the largest Phanerozoic biodiversity crisis. Here, we investigated palynomorphs, plant, and tetrapod fossils from the Permian-Triassic South Taodonggou Section in Xinjiang, China. Our fossil records, calibrated by a high-resolution age model, reveal the presence of vibrant regional gymnospermous forests and fern fields, while marine organisms experienced mass extinction. This refugial vegetation was crucial for nourishing the substantial influx of surviving animals, thereby establishing a diverse terrestrial ecosystem approximately 75,000 years after the mass extinction. Our findings contradict the widely held belief that restoring terrestrial ecosystem functional diversity to pre-extinction levels would take millions of years. Our research indicates that moderate hydrological fluctuations throughout the crisis sustained this refugium, likely making it one of the sources for the rapid radiation of terrestrial life in the early Mesozoic. 

This study introduces a new family of redox-active oxide molten salt (ROMS) composites for high-capacity thermal energy storage. Porous perovskite oxides serve as active support materials, facilitating thermochemical energy storage...