Feed aggregator

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D5EE01958C, PaperShang Liu, Shiteng Li, Qijun Yang, Meng Lin
Significant progress has been made in enhancing solar interfacial evaporation (SIE) performance at the laboratory scale, however, translating these improvements to meter-scale systems suitable for practical deployment remains limited by...
The content of this RSS Feed (c) The Royal Society of Chemistry

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D5EE03443D, Perspective Open Access &nbsp This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.Haoxuan Jiang, Adel Rezaeimotlagh, Sahar Nazari, Tianyu Li, Jingwen Huang, Dorna Esrafilzadeh, Renwu Zhou, Ali Rouhollah Jalili
Electrifying ammonia production demands modular systems powered entirely by renewable energy, eliminating dependence on fossil-derived hydrogen. This perspective argues that coupling non-thermal plasma oxidation of air to nitric oxide (NO)...
The content of this RSS Feed (c) The Royal Society of Chemistry

GNN-ACLP: Graph Neural Networks Based Analog Circuit Link Prediction

Circuit link prediction identifying missing component connections from incomplete netlists is crucial in analog circuit design automation. However, existing methods face three main challenges: 1) Insufficient use of topological patterns in circuit graphs reduces prediction accuracy; 2) Data scarcity due to the complexity of annotations hinders model generalization; 3) Limited adaptability to various netlist formats. We propose GNN -ACLP, a graph neural networks (GNNs) based method featuring three innovations to tackle these challenges. First, we introduce the SEAL (learning from Subgraphs, Embeddings, and Attributes for Link prediction) framework and achieve port-level accuracy in circuit link prediction. Second, we propose Netlist Babel Fish, a netlist format conversion tool leveraging retrieval-augmented generation (RAG) with a large language model (LLM) to improve the compatibility of netlist formats. Finally, we construct SpiceNetlist, a comprehensive dataset that contains 775 annotated circuits across 10 different component classes. Experiments demonstrate accuracy improvements of 16.08% on SpiceNetlist, 11.38% on Image2Net, and 16.01% on Masala-CHAI compared to the baseline in intra-dataset evaluation, while maintaining accuracy from 92.05% to 99.07% in cross-dataset evaluation, exhibiting robust feature transfer capabilities.

https://drive.google.com/file/d/1W_VPanXTm3I6oWFV2xhm81H2egEViXhI/view?usp=sharing

• Speaker: Guanyuan Peter Pan
• Thursday 21 August 2025, 12:45-13:45
• Venue: Zoom (Meeting ID: 873 5011 9733, Passcode: 732177).
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list

Direct brain stimulation modulate physiology and behaviour in humans.

Direct brain electrical brain stimulation offers a causal window on human brain function and a route to therapy. This talk contrasts open-loop perturbation, single pulses and short trains used to map effective connectivity and state dependence across wake, sleep and anaesthesia, with closed-loop paradigms that detect neural features in intracranial EEG in real time to trigger stimulation. I’ll show how closed-loop experiments can adjust stimulation based on estimated cognitive state, producing rapid changes in cortical dynamics and measurable effects on behaviour during ongoing tasks. I’ll also discuss high-frequency oscillations and ripples as candidate biomarkers for targeting, and outline opportunities and limits for translating these methods to precision neuromodulation in humans.

The Host for this talk is Tristan Bekinschtein, tb419@cam.ac.uk.

• Speaker: Rina Zelmann, PhD, Neurology at Massachusetts General Hospital & Harvard Medical School
• Tuesday 26 August 2025, 16:00-17:30
• Venue: Ground Floor Lecture Theatre, Department of Psychology.
• Series: Zangwill Club; organiser: Adelaide Schiemer.

Add to your calendar or Include in your list

GNN-ACLP: Graph Neural Networks Based Analog Circuit Link Prediction

Circuit link prediction identifying missing component connections from incomplete netlists is crucial in analog circuit design automation. However, existing methods face three main challenges: 1) Insufficient use of topological patterns in circuit graphs reduces prediction accuracy; 2) Data scarcity due to the complexity of annotations hinders model generalization; 3) Limited adaptability to various netlist formats. We propose GNN -ACLP, a graph neural networks (GNNs) based method featuring three innovations to tackle these challenges. First, we introduce the SEAL (learning from Subgraphs, Embeddings, and Attributes for Link prediction) framework and achieve port-level accuracy in circuit link prediction. Second, we propose Netlist Babel Fish, a netlist format conversion tool leveraging retrieval-augmented generation (RAG) with a large language model (LLM) to improve the compatibility of netlist formats. Finally, we construct SpiceNetlist, a comprehensive dataset that contains 775 annotated circuits across 10 different component classes. Experiments demonstrate accuracy improvements of 16.08% on SpiceNetlist, 11.38% on Image2Net, and 16.01% on Masala-CHAI compared to the baseline in intra-dataset evaluation, while maintaining accuracy from 92.05% to 99.07% in cross-dataset evaluation, exhibiting robust feature transfer capabilities.

• Speaker: Guanyuan Peter Pan
• Thursday 21 August 2025, 12:45-13:45
• Venue: Zoom (Meeting ID: 873 5011 9733, Passcode: 732177).
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D5EE03272E, PaperSai Li, Xianhui Zhao, Zheng Liu, Rang Xiao, Xin Zhang, Binghan Cui, GePing Yin, Pengjian Zuo, Yulin Ma, Chaoyang Li, Ning Wang, Guokang Han, Huaizheng Ren, Chunyu Du
The sluggish interfacial kinetics of graphite anodes restricts the fast-charging capability of lithium-ion batteries (LIBs), inducing severe lithium plating and electrolyte decomposition, which markedly accelerates battery degradation and raises safety...
The content of this RSS Feed (c) The Royal Society of Chemistry

Nature Energy, Published online: 21 August 2025; doi:10.1038/s41560-025-01849-y

Preserving subsistence practices

Nature Energy, Published online: 21 August 2025; doi:10.1038/s41560-025-01848-z

Branching out

Nature Energy, Published online: 21 August 2025; doi:10.1038/s41560-025-01850-5

Manufacturing in full flow

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D5EE02702K, Review Article Open Access &nbsp This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.Sowjanya Vallem, Malayil Gopalan Sibi, Rahul Patil, Vishakha Goyal, Giridhar Babu Anam, E. A. Lohith, K. Keerthi, Muhammad Umer, N. V. V. Jyothi, Matthias Vandichel, Subhasmita Ray, Daniel Ioan Stroe, Mani Balamurugan, Sada Venkateswarlu, Jagadeesh Rajenahally, Radek Zboril, Aristides Bakandritsos
Amidst escalating global concerns over rising atmospheric CO2 levels, the capture and effective utilization of C1 and C2+ sources are crucial not only for advancing a sustainable society but also...
The content of this RSS Feed (c) The Royal Society of Chemistry

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D5EE03339J, PaperCong Geng, Kuanxiang Zhang, Jiwen Jiang, Changhua Wang, Chung Hsien Wu, Jize Wang, Fei Long, Liyuan Han, Yi-Bing Cheng, Yong Peng
High-performance of two-terminal (2-T) perovskite/Cu(In,Ga)Se2 (PVK/CIGS) tandem solar cells (TSCs) is fundamentally limited by submicron-scale topographic irregularities inherent to commercially available CIGS substrates. We demonstrate that these features induce spatial...
The content of this RSS Feed (c) The Royal Society of Chemistry

GNN-ACLP: Graph Neural Networks Based Analog Circuit Link Prediction

Circuit link prediction identifying missing component connections from incomplete netlists is crucial in analog circuit design automation. However, existing methods face three main challenges: 1) Insufficient use of topological patterns in circuit graphs reduces prediction accuracy; 2) Data scarcity due to the complexity of annotations hinders model generalization; 3) Limited adaptability to various netlist formats. We propose GNN -ACLP, a graph neural networks (GNNs) based method featuring three innovations to tackle these challenges. First, we introduce the SEAL (learning from Subgraphs, Embeddings, and Attributes for Link prediction) framework and achieve port-level accuracy in circuit link prediction. Second, we propose Netlist Babel Fish, a netlist format conversion tool leveraging retrieval-augmented generation (RAG) with a large language model (LLM) to improve the compatibility of netlist formats. Finally, we construct SpiceNetlist, a comprehensive dataset that contains 775 annotated circuits across 10 different component classes. Experiments demonstrate accuracy improvements of 16.08% on SpiceNetlist, 11.38% on Image2Net, and 16.01% on Masala-CHAI compared to the baseline in intra-dataset evaluation, while maintaining accuracy from 92.05% to 99.07% in cross-dataset evaluation, exhibiting robust feature transfer capabilities.

• Speaker: Guanyuan Peter Pan
• Thursday 21 August 2025, 12:30-13:30
• Venue: Zoom (Meeting ID: 873 5011 9733, Passcode: 732177).
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list

This work introduces an easy-fabrication and self-stratifying paint recipe that forms an absorptionscattering dual-layer structure, systematically preserving color while enabling subambient daytime radiative cooling. All colored paints exhibit outstanding subambient cooling potential (>90% solar refletance and 0.94 sky window emissivity), and the coolest black paint achieves exceptionally high solar reflectance (∼46%).

Abstract

Although significant energy-saving potential of passive daytime radiative cooling has led to diffusively white or reflective surfaces, achieving coloration expands their application for both aesthetic and functional purposes. However, it presents huge technical challenges in scalable processing. This work presents a novel and general self-stratifying approach for creating radiative cooling paints with various colors using selected narrow-band absorption dyes. Applied as a single coat, these paints self-assemble into an absorption-scattering dual-layer structure that efficiently displays color while minimizing light absorption. Three base colors (cyan, magenta, and yellow) are demonstrated to provide full daytime subambient cooling, achieving 93% solar reflectance and 0.94 sky window emissivity. One newly developed black paint achieves 46.3% solar reflectance and a temperature drop of up to 25 °C, outperforming commercial counterparts. A purple radiative cooling paint (mixed dyes) demonstrates the color customization potential within the CMYK space, addressing the dilemma between color and cooling for colored radiative cooling paints. The mechanical and surface properties, characterized by abrasion, adhesion, and contact angle tests, are on par with or surpassing commercial products. This unique self-stratifying mechanism enhances surface properties and provides new insights for developing high-volume loading paints and water-harvesting coatings without increasing fabrication complexity.

The intelligent electromagnetic functional materials (EFMs) with multi-band compatibility, environment perception, and adaptation has demonstrated prosperous applications in electromagnetic communication, electromagnetic protection, sensing, robots, energy harvesting, wearable electronics, healthcare, and environmental management. This review systematically summarizes the intelligentization of EFMs and provides insights into future challenges and prospects.

Abstract

With advancements in radar detection and electromagnetic (EM) communication technologies, intelligentizing EM functional materials (EFMs) and endowing them with dynamic responsiveness across multi-spectral ranges, along with the ability to perceive and adapt to complex operation environments, is of significant importance. This article provides a comprehensive review of the recent progress in intelligent EFMs. It begins with the fundamentals of intelligent EFMs, with an emphasis on their EM response mechanisms and basic functions. The motivation and necessity of developing intelligent EFMs are then discussed. Thereafter, new advances, particularly in the design of intelligent EFMs for various applications, including EM communication systems, EM protection platforms, sensors, robots, energy harvesting devices, wearable electronics, healthcare, and environmental management, are highlighted. Lastly, this review concludes with an outlook on future directions, critical challenges to address, and possible solutions for intelligent EFMs.

A cation-selective defects strategy is proposed to decouple and optimize activity and durability of A-site ordered layered perovskites air electrode for reversible protonic ceramic cells (RPCCs). A specific cations-deficient Pr(Ba0.5Sr0.5)0.95Co1.5Fe0.5O5+δ displays superior activity and stability when used as an RPCCs air electrode, accelerating its commercialization.

Abstract

Reversible protonic ceramic cells facilitate efficient chemical-electrical energy interconversion, advancing renewable energy utilization. Commercial viability, however, demands intermediate-to-low temperatures (ILT, 400–600 °C) operation, currently constrained by air electrode performance. A-site ordered layered perovskite PrBa0.5Sr0.5Co1.5Fe0.5O5+δ (PBSCF) promises, yet faces activity and stability issues at ILT. Cation defects effectively tune defect structures in simple perovskites, boosting electrochemical performance, but their specific effects in A-site ordered perovskites with dual A-site environments remain unexplored. Here, A-site cation-selective defects are engineered to tune PBSCF's performance, with Pr-deficient (Pr0.95Ba0.5Sr0.5Co1.5Fe0.5O5+δ, p-PBSCF) and Ba/Sr-deficient (Pr(Ba0.5Sr0.5)0.95Co1.5Fe0.5O5+δ, bs-PBSCF) variants revealing distinct defects-performance relationships. Pr defects weaken Co─O covalency to activate Co sites, enhancing oxygen electrocatalytic activity. Concurrently, it lowers oxygen vacancy concentration, inhibiting hydration-induced lattice expansion. This stabilizes Ba─O/Sr─O bonds and mitigates Ba/Sr segregation, enhancing stability. However, the reduced oxygen vacancy concentration inhibits the material's hydration, lowering proton conduction and thus restricting activity enhancement. In contrast, Ba/Sr defects not only weaken Co─O covalency to activate Co sites, but also increase oxygen vacancy concentration, promoting proton and oxygen-ion transport, thereby significantly enhancing electrode activity. Furthermore, despite increased hydration, bs-PBSCF's larger-radius cation defects yield a smaller unit cell versus p-PBSCF, further strengthening Ba─O/Sr─O bonds and inhibiting Ba/Sr segregation, thus leading to superior stability.

This study investigates how the chemical structure of alumina precursors affects their interaction with benzenethiol inhibitor layers on Cu surfaces during area-selective atomic layer deposition (AS-ALD). It reveals that ligand-dependent miscibility and Lewis acidity govern precursor infiltration and degradation of the inhibitor, leading to selectivity loss. The work introduces a new mechanism for selectivity failure in AS-ALD processes.

Abstract

Area-selective atomic layer deposition (AS-ALD) is of increasing importance in nanostructure fabrication, and precursor selection is critical to realizing a successful process. This work explores the role of the precursor in achieving AS-ALD of Al2O3 on a dielectric (SiO2) in the presence of a metal (Cu/CuOx). Four different precursors—dimethylaluminum isopropoxide (DMAI), trimethylaluminum (TMA), triethylaluminum (TEA), and triisobutylaluminum (TIBA)—are tested against a benzenethiol (BT) inhibitor. BT forms monolayers on Cu, whereas on CuOx it forms a thick crystalline multilayer composed of 1D-coordination polymers of Cu-thiolate (CuBT). This work observes that DMAI provides exceptional selectivity: 22 nm of Al2O3 can be deposited on patterned substrates with 99.9% selectivity, and in the case of thick CuBT can achieve excellent pattern transfer on nanoscale patterns with well-defined material interfaces. In contrast, none of the alkylaluminum precursors show significant selectivity, a result attributed to their miscibility in the CuBT multilayer leading to its degradation. This work proposes that the miscibility of the alkylaluminum precursors depends on ligand length and structure. The results show that the ligand-dependent miscibility and subsequent degradation of CuBT impact the location of Al2O3 nucleation. This study highlights new considerations for AS-ALD process design to achieve high selectivity.

Nature Nanotechnology, Published online: 20 August 2025; doi:10.1038/s41565-025-01992-3

This Review discusses single-photon detectors and quantum-light sources for super-resolution microscopy, measurements below classical noise limits and photon-number-resolved spectroscopy as emerging tools for nanoscale electronic materials characterization and bioimaging.

Title to be confirmed

Abstract not available

• Speaker: Kris Cao (Cohere)
• Friday 14 November 2025, 12:00-13:00
• Venue: Hybrid (In-Person + Online). Here is the Zoom link: https://cam-ac-uk.zoom.us/j/4751389294?pwd=Z2ZOSDk0eG1wZldVWG1GVVhrTzFIZz09.
• Series: NLIP Seminar Series; organiser: Suchir Salhan.

Add to your calendar or Include in your list

Title to be confirmed

Abstract not available

• Speaker: Kris Cao (Cohere)
• Friday 14 November 2025, 12:00-13:00
• Venue: Hybrid (In-Person + Online). Here is the Zoom link: https://cam-ac-uk.zoom.us/j/4751389294?pwd=Z2ZOSDk0eG1wZldVWG1GVVhrTzFIZz09.
• Series: NLIP Seminar Series; organiser: Suchir Salhan.

Add to your calendar or Include in your list

Energy Environ. Sci., 2025, Advance Article
DOI: 10.1039/D5EE02676H, PerspectiveCarlos D. Díaz-Marín, Zachary J. Berquist
Unprecedented artificial intelligence (AI) growth poses major electricity, emissions, and water challenges. Essentially all AI electricity use is converted to heat, representing a Gigawatt-scale resource that can critically boost AI sustainability.
To cite this article before page numbers are assigned, use the DOI form of citation above.
The content of this RSS Feed (c) The Royal Society of Chemistry