Search for: All records

Abstract Over the past decade, lead halide perovskites have gained significant interest for ionizing radiation detection, owing to their exceptional performance, and cost-effective fabrication in a wide range of form factors, from thick films to large single crystals. However, the toxicity of lead, limited environmental and thermal stability of these materials, as well as dark current drift due to ionic conductivity, have prompted the development of alternative materials that can address these challenges. Bismuth-based compounds (including perovskite derivatives and nonperovskite materials) have similarly high atomic numbers, leading to strong X-ray attenuation, but have lower toxicity, tend to be more environmentally stable, and can have lower ionic conductivity, especially in low-dimensional materials. These materials are also advantageous over commercial direct X-ray detectors by being able to detect lower dose rates of X-rays than amorphous selenium by at least two orders of magnitude, are potentially more cost-effective to mass produce than cadmium zinc telluride, and can operate at room temperature (unlike high-purity Ge). Given the strong interest in this area, we here discuss recent advances in the development of bismuth-based perovskite derivatives (with 3D, 2D and 0D structural dimensionality), and other bismuth-based perovskite-inspired materials for direct X-ray detection. We discuss the critical properties of these materials that underpin the strong performances achieved, particularly the ability to detect low-dose rates of X-rays. We cover key strategies for enhancing the performance of these materials, as well as the challenges that need to be overcome to commercialize these emerging technologies. Graphical abstract 

Abstract We report a new CO observation survey of LHAASO J0341+5258, using the Nobeyama Radio Observatory 45-m telescope. LHAASO J0341+5258 is one of the unidentified ultra-high-energy (UHE;E> 100 TeV) gamma-ray sources detected by LHAASO. Our CO observations were conducted in 2024 February and March, with a total observation time of 36 hr, covering the LHAASO source (∼0 $\stackrel{\text{°}}{.}$3–0 $\stackrel{\text{°}}{.}$5 in radius) and its surrounding area (1° × 1 $\stackrel{\text{°}}{.}$5). Within the LHAASO source extent, we identified five compact (<2 pc) molecular clouds at nearby distances (<1–4 kpc). These clouds can serve as proton–proton collision targets, producing hadronic gamma rays via neutral pion decays. Based on the hydrogen densities (700–5000 cm⁻³) estimated from our CO observations and archived Hidata from the Dominion Radio Astrophysical Observatory survey, we derive the total proton energy ofW_p(E> 1 TeV) ∼ 10⁴⁵erg to account for the gamma-ray flux. One of the molecular clouds appears to be likely associated with an asymptotic giant branch (AGB) star with an extended CO tail, which may indicate some particle acceleration activities. However, the estimated maximum particle energy below 100 TeV makes the AGB-like star unlikely to be a PeVatron site. We conclude that the UHE emission observed in LHAASO J0341+5258 could be due to hadronic interactions between the newly discovered molecular clouds and TeV–PeV protons originating from a distant SNR or due to leptonic emission from a pulsar wind nebula candidate, which is reported in our companion X-ray observation paper. 

Autonomous edge computing in robotics, smart cities, and autonomous vehicles relies on the seamless integration of sensing, processing, and actuation for real-time decision-making in dynamic environments. At its core is the sensing-to-action loop, which iteratively aligns sensor inputs with computational models to drive adaptive control strategies. These loops can adapt to hyper-local conditions, enhancing resource efficiency and responsiveness, but also face challenges such as resource constraints, synchronization delays in multimodal data fusion, and the risk of cascading errors in feedback loops. This article explores how proactive, context-aware sensing-to-action and action-to-sensing adaptations can enhance efficiency by dynamically adjusting sensing and computation based on task demands, such as sensing a very limited part of the environment and predicting the rest. By guiding sensing through control actions, action-to-sensing pathways can improve task relevance and resource use, but they also require robust monitoring to prevent cascading errors and maintain reliability. Multi-agent sensing-action loops further extend these capabilities through coordinated sensing and actions across distributed agents, optimizing resource use via collaboration. Additionally, neuromorphic computing, inspired by biological systems, provides an efficient framework for spike-based, event-driven processing that conserves energy, reduces latency, and supports hierarchical control-making it ideal for multi-agent optimization. This article highlights the importance of end-to-end co-design strategies that align algorithmic models with hardware and environmental dynamics, improve cross-layer inter-dependencies to improve throughput, precision, and adaptability for energy-efficient edge autonomy in complex environments. 

Abstract Spiking Neural Networks (SNNs) have emerged as a compelling, energy-efficient alternative to traditional Artificial Neural Networks (ANNs) for static image tasks such as image classification and segmentation. However, in the more complex video classification domain, SNN-based methods fall considerably short of ANN-based benchmarks, due to the challenges in processing dense RGB frames. To bridge this gap, we propose ReSpike, a hybrid framework that synergizes the strengths of ANNs and SNNs to tackle action recognition tasks with high accuracy and low energy cost. By partitioning film clips into RGB image Key Frames, which primarily capture spatial information, and event-like Residual Frames, which emphasize temporal dynamics cues, ReSpike leverages ANN for processing spatial features and SNN for modeling temporal features. In addition, we propose a multi-scale cross-attention mechanism for effective feature fusion. Compared to state-of-the-art SNN baselines, our ReSpike hybrid architecture demonstrates significant performance improvements (e.g., >30% absolute accuracy improvement on both HMDB-51 and UCF-101 datasets). Additionally, ReSpike is the first SNN method capable of scaling to the large-scale benchmark Kinetics-400. Furthermore, ReSpike achieves comparable performance with prior ANN approaches while bringing better accuracy-energy tradeoff. 

Lithium metal batteries (LMBs), especially “anode-free“ LMBs, promise much higher energy density than current lithium-ion batteries but suffer from poor capacity retention. While novel electrolytes have been designed to extend cycle life in anode free LMBs, most of them contain a high fraction of fluorinated solvents or diluents that may cause environmental concerns. Herein, we report the design and synthesis of a group of nonfluorinated ether solvents (termed xME solvents). By substituting the methyl terminal group of 1,2-dimethoxy ethane (DME) with different alkyl groups, the solvation power of xME solvents was tuned to be weaker, leading to more ion pairing in electrolyte solvation structure. In anode free type Cu/LiFePO₄(Cu/LFP) cells, xME electrolytes in general show better capacity retention than DME-based electrolyte. Some xME electrolytes also show better oxidative stability than DME against aluminum and LiNi_0.8Mn_0.1Co_0.1O₂(NMC811) electrodes. While the general improvement in LMB cycle life and oxidative stability can be attributed to more ion pairing, the local variation within xME electrolytes indicates other factors are also important. Our work proposes a molecular design strategy to fine-tune ion solvation structure of nonfluorinated ether electrolytes for LMBs. 

Composite polymer electrolytes that incorporate ceramic fillers in a polymer matrix offer mechanical strength and flexibility as solid electrolytes for lithium metal batteries. However, fast Li+ transport between polymer and Li+-conductive filler phases is not a simple achievement due to high barriers for Li+ exchange across the interphase. This study demonstrates how modification of Li7La3Zr2O12 (LLZO) nanofiller surfaces with silane chemistries influences Li+ transport at local and global electrolyte scales. Anhydrous reactions covalently link amine-functionalized silanes [(3-aminopropyl)triethoxysilane (APTES)] to LLZO nanoparticles, which protects LLZO in air. APTES functionalization lowers the poly (ethylene oxide) (PEO)-LLZO interphase resistance to half that of unmodified LLZO and increases effective Li+ transference number, while insulating Al2O3 completely blocks ion exchange and lowers transference number and conductivity in PEO-lithium bis(trifluoromethanesulfonyl)imide (LiTFSI)-LLZO composites. Modeling an inner resistive interphase between LLZO and PEO surrounded by an outer conductive interphase explains non-linear conductivity trends. Solid-state 7Li & 6Li nuclear magnetic resonance shows Li+ only exchanges between PEO-LiTFSI and some LLZO interphase, with no appreciable Li+ transport through bulk LLZO. Surface functionalization is a promising path toward lowering the polymer-ceramic interphase resistance. This work demonstrates that local changes in Li+ transport affect macroscopic performance, highlighting the intricate relationships between all interfaces in inherently heterogeneous composite polymer electrolytes. 

This review explores the intersection of bio-plausible artificial intelligence in the form of spiking neural networks (SNNs) with the analog in-memory computing (IMC) domain, highlighting their collective potential for low-power edge computing environments. Through detailed investigation at the device, circuit, and system levels, we highlight the pivotal synergies between SNNs and IMC architectures. Additionally, we emphasize the critical need for comprehensive system-level analyses, considering the inter-dependencies among algorithms, devices, circuit, and system parameters, crucial for optimal performance. An in-depth analysis leads to the identification of key system-level bottlenecks arising from device limitations, which can be addressed using SNN-specific algorithm–hardware co-design techniques. This review underscores the imperative for holistic device to system design-space co-exploration, highlighting the critical aspects of hardware and algorithm research endeavors for low-power neuromorphic solutions.