Related Articles
Advanced electrode processing for lithium-ion battery manufacturing
Lithium-ion batteries (LIBs) need to be manufactured at speed and scale for their use in electric vehicles and devices. However, LIB electrode manufacturing via conventional wet slurry processing is energy-intensive and costly, challenging the goal to achieve sustainable, affordable and facile manufacturing of high-performance LIBs. In this Review, we discuss advanced electrode processing routes (dry processing, radiation curing processing, advanced wet processing and 3D-printing processing) that could reduce energy usage and material waste. Maxwell-type dry processing is a scalable alternative to conventional processing and has relatively low manufacturing cost and energy consumption. Radiation curing processing could enable high-throughput manufacturing, but binder selection is limited to certain radiation curable chemistries. 3D-printing processing can produce electrodes with diverse architectures and improved rate performance, but scalability is yet to be demonstrated. 3D-printing processing is good for special applications where throughput and cost can be compromised for performance.
Memristors based on two-dimensional h-BN materials: synthesis, mechanism, optimization and application
Memristors offer vast application opportunities in storage, logic devices, and computation due to their nonvolatility, low power consumption, and fast operational speeds. Two-dimensional materials, characterized by their novel mechanisms, ultra-thin channels, high mechanical flexibility, and superior electrical properties, demonstrate immense potential in the domain of high-density, fast, and energy-efficient memristors. Hexagonal boron nitride (h-BN), as a new two-dimensional material, has the characteristics of high thermal conductivity, flexibility, and low power consumption, and has a significant application prospect in the field of memristor. In this paper, the recent research progress of the h-BN memristor is reviewed from the aspects of device fabrication, resistance mechanism, and application prospect.
Infiltration-driven performance enhancement of poly-crystalline cathodes in all-solid-state batteries
All-solid-state batteries (ASSBs) with adequately selected cathode materials exhibit a higher energy density and better safety than conventional lithium-ion batteries (LIBs). Ni-rich layered cathodes are benchmark materials for traditional LIBs owing to their high energy density. Recent studies have highlighted the advantages of using crack-free, single-crystalline cathode materials in ASSBs. In this study, a scalable infiltration sheet-type process was used to fabricate composite electrodes with different cathode-material morphologies for ASSBs. Typically, crack-free single-crystalline materials exhibit better retention performance and lower rate capability (i.e., slower kinetics in charge‒discharge processes) than polycrystalline cathode materials. Li6PS5Cl-infiltrated polycrystalline electrodes showed excellent retention performance and rate capability. Galvanostatic intermittent titration technique analysis and transmission electron microscopy of the single-crystalline electrode confirmed severe polarization and the presence of a rock-salt-structure layer in the cathode particles; these results indicated side reactions within the layered structure of the material. In contrast, composite electrodes consisting of polycrystalline cathode materials infiltrated with the solid electrolyte Li6PS5Cl showed excellent electrochemical performance owing to intimate electrode–electrolyte interfacial contact. The result from this study confirmed the critical influence of interface engineering and material morphology on the overall performance and stability of ASSBs and could facilitate the development of high-performance ASSBs in the future.
Giant memory window performance and low power consumption of hexagonal boron nitride monolayer atomristor
Two-dimensional (2D) monolayers have gained significant attention as ultrathin active layers for fabricating atomic-scale memristor (atomristor) structures due to their crystalline structures and clean surfaces. This study reports on the giant memory window performance and low power consumption of the atomristor structures using a hexagonal boron nitride (h-BN) monolayer and symmetric silver (Ag) metal electrodes through a polypropylene carbonate (PPC) assisted transfer method. The h-BN atomristor exhibits the highest memory window (~4 × 109), the lowest leakage current (~0.24 pA), and the lowest power consumption (~3 × 10−14 W) compared to the other 2D atomristors. Furthermore, the h-BN atomristor achieves significant endurances and yields of up to 10,000 switching cycles and 77%, respectively, due to the superior thermomechanical properties of the PPC support layer for transferring ultrathin and large-area h-BN monolayers. These results represent a significant step toward the realization of high-performance and energy-efficient neuromorphic computing circuits based on 2D monolayers.
A unified acoustic-to-speech-to-language embedding space captures the neural basis of natural language processing in everyday conversations
This study introduces a unified computational framework connecting acoustic, speech and word-level linguistic structures to study the neural basis of everyday conversations in the human brain. We used electrocorticography to record neural signals across 100 h of speech production and comprehension as participants engaged in open-ended real-life conversations. We extracted low-level acoustic, mid-level speech and contextual word embeddings from a multimodal speech-to-text model (Whisper). We developed encoding models that linearly map these embeddings onto brain activity during speech production and comprehension. Remarkably, this model accurately predicts neural activity at each level of the language processing hierarchy across hours of new conversations not used in training the model. The internal processing hierarchy in the model is aligned with the cortical hierarchy for speech and language processing, where sensory and motor regions better align with the model’s speech embeddings, and higher-level language areas better align with the model’s language embeddings. The Whisper model captures the temporal sequence of language-to-speech encoding before word articulation (speech production) and speech-to-language encoding post articulation (speech comprehension). The embeddings learned by this model outperform symbolic models in capturing neural activity supporting natural speech and language. These findings support a paradigm shift towards unified computational models that capture the entire processing hierarchy for speech comprehension and production in real-world conversations.
Responses