Nanoscience News (<front>)

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D5EE00211G, PaperJianlong Wang, Zheng Yang, Zhenjie Wang, Xinxian Wang, Yanrui Zhao, Jinbiao Ma, Hengyu Li, Yang Yu, Zhong Lin Wang, Tinghai Cheng
With the challenges of monitoring marine environmental pollution, developing in-situ energy harvesting technologies for self-powered marine sensing nodes has become increasingly urgent for effective environmental protection. As an effective approach...
The content of this RSS Feed (c) The Royal Society of Chemistry

To address off-target organ enrichment and low endosomal escape during mRNA delivery, the study learns from the technology of lipid nanoparticles (LNPs) and integrate cholesterol moieties and zwitterionic species into branched poly(β-amino ester)s (hPAEs) to obtain “four-in-one” LNP-like hPAEs (O-LhPAEs), which have spleen-targeted delivery and have potential for the treatment of silicosis by nebulization.

Abstract

mRNA therapeutics hold tremendous promise for disease prevention and treatment. Development of high-performance mRNA delivery systems with enhanced transfection efficiency and a safety profile will further fulfill their therapeutic potential and expedite their translation. The synthesis of “four-in-one” highly branched poly(β-amino ester)s (O-LhPAEs) is reported by integrating the essential components of lipid nanoparticles (LNPs) for spleen-selective mRNA enrichment and nebulization treatment of silicosis. 60 O-LhPAEs with distinct branched structure and chemical composition, including tertiary/quaternary amines, cholesterol moieties, zwitterionic species, and hydrophobic alkyl tails, are synthesized using sequential Michael addition, ring-opening, and nucleophilic substitution reactions. The unique topological structure and chemical composition collectively enhanced O-LhPAEs/mRNA polyplex serum resistance, cellular uptake, and endosomal escape. The optimal O-LhPAE, 20%b-3C-2P12, exhibits up to 93.1% mRNA transfection across 11 different cell types, including epithelial cells, fibroblasts, cancer cells, stem cells, neurological cells, and astrocytes. Biodistribution study reveals that 20%b-3C-2P12/mRNA polyplexes are mainly enriched in the spleen following systemic administration. Through nebulization, 20%b-3C-2P12 mediated high Tbx2 mRNA expression in the lungs of silicosis mice, effectively restoring lung functions. This study not only establishes a strategy for development of LNP-like O-LhPAEs but also provides promising candidates for highly safe, efficient, and spleen-selective mRNA delivery and nebulization treatment of silicosis.

Robust nonnegative matrix factorization with the beta-divergence and applications in imaging

Data is often available in matrix form, in which columns are samples, and processing of such data often entails finding an approximate factorization of the matrix into two factors. The first factor (the “dictionary”) yields recurring patterns characteristic of the data. The second factor (“the activation matrix”) describes in which proportions each data sample is made of these patterns. Nonnegative matrix factorization (NMF) is a popular unsupervised learning technique for analysing data with nonnegative values, with applications in many areas such as in text information retrieval, recommender systems, audio signal processing, and hyperspectral imaging. In a first part, I will give a short tutorial presentation about NMF for data processing, with a focus on majorization-minimization algorithms for NMF with the beta-divergence, a continuous family of loss functions that takes the quadratic loss, KL divergence and Itakura-Saito divergence as special cases. Then, I will present applications for hyperspectral unmixing in remote sensing and factor analysis in dynamic positron emission tomography, introducing robust variants of NMF that account for outliers, nonlinear phenomena or specific binding.

References C. Févotte, J. Idier. Algorithms for nonnegative matrix factorization with the beta-divergence. Neural computation, 2011. C. Févotte, N. Dobigeon. Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization. IEEE Transactions on Image Processing, 2015. Y. C. Cavalcanti, T. Oberlin, N. Dobigeon, C. Févotte, S. Stute, M. J. Ribeiro, C. Tauber. Factor analysis of dynamic PET images: beyond Gaussian noise. IEEE Transactions on Medical Imaging, 2019. A. Marmin, H. Goulart, C. Févotte. Joint majorization-minimization for nonnegative matrix factorization with the beta-divergence. IEEE Transactions on Signal Processing, 2023.

Bio:

Cédric Févotte is a CNRS research director with Institut de Recherche en Informatique de Toulouse (IRIT). Previously, he was a CNRS researcher at Laboratoire Lagrange (Nice, 2013-2016) & Télécom ParisTech (2007-2013), a research engineer at Mist-Technologies (the startup that became Audionamix, 2006-2007) and a postdoc at University of Cambridge (2003-2006). He holds MEng and PhD degrees in EECS from École Centrale de Nantes. His research interests concern statistical signal processing and machine learning, with particular interests in matrix factorization, inverse problems, source separation and recommender systems. Selected distinctions: IEEE Fellow (2022), ERC Consolidator Grant (2016-2022), IEEE Signal Processing Society Sustained Impact Paper Award (2023), IEEE Signal Processing Society Best Paper Award (2014).

• Speaker: Cédric Févotte, Institut de Recherche en Informatique de Toulouse (IRIT)
• Tuesday 22 April 2025, 11:00-12:00
• Venue: LT6, Baker Building, CUED.
• Series: Signal Processing and Communications Lab Seminars; organiser: Kimberly Cole.

Add to your calendar or Include in your list

Robust nonnegative matrix factorization with the beta-divergence and applications in imaging

Data is often available in matrix form, in which columns are samples, and processing of such data often entails finding an approximate factorization of the matrix into two factors. The first factor (the “dictionary”) yields recurring patterns characteristic of the data. The second factor (“the activation matrix”) describes in which proportions each data sample is made of these patterns. Nonnegative matrix factorization (NMF) is a popular unsupervised learning technique for analysing data with nonnegative values, with applications in many areas such as in text information retrieval, recommender systems, audio signal processing, and hyperspectral imaging. In a first part, I will give a short tutorial presentation about NMF for data processing, with a focus on majorization-minimization algorithms for NMF with the beta-divergence, a continuous family of loss functions that takes the quadratic loss, KL divergence and Itakura-Saito divergence as special cases. Then, I will present applications for hyperspectral unmixing in remote sensing and factor analysis in dynamic positron emission tomography, introducing robust variants of NMF that account for outliers, nonlinear phenomena or specific binding.

References C. Févotte, J. Idier. Algorithms for nonnegative matrix factorization with the beta-divergence. Neural computation, 2011. C. Févotte, N. Dobigeon. Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization. IEEE Transactions on Image Processing, 2015. Y. C. Cavalcanti, T. Oberlin, N. Dobigeon, C. Févotte, S. Stute, M. J. Ribeiro, C. Tauber. Factor analysis of dynamic PET images: beyond Gaussian noise. IEEE Transactions on Medical Imaging, 2019. A. Marmin, H. Goulart, C. Févotte. Joint majorization-minimization for nonnegative matrix factorization with the beta-divergence. IEEE Transactions on Signal Processing, 2023.

Bio:

Cédric Févotte is a CNRS research director with Institut de Recherche en Informatique de Toulouse (IRIT). Previously, he was a CNRS researcher at Laboratoire Lagrange (Nice, 2013-2016) & Télécom ParisTech (2007-2013), a research engineer at Mist-Technologies (the startup that became Audionamix, 2006-2007) and a postdoc at University of Cambridge (2003-2006). He holds MEng and PhD degrees in EECS from École Centrale de Nantes. His research interests concern statistical signal processing and machine learning, with particular interests in matrix factorization, inverse problems, source separation and recommender systems. Selected distinctions: IEEE Fellow (2022), ERC Consolidator Grant (2016-2022), IEEE Signal Processing Society Sustained Impact Paper Award (2023), IEEE Signal Processing Society Best Paper Award (2014).

• Speaker: Cédric Févotte, Institut de Recherche en Informatique de Toulouse (IRIT)
• Tuesday 22 April 2025, 11:00-12:00
• Venue: LT6, Baker Building, CUED.
• Series: Signal Processing and Communications Lab Seminars; organiser: Kimberly Cole.

Add to your calendar or Include in your list

Advanced imaging of the feline biliary tract.

Abby graduated from Cambridge Veterinary School in 2001 and during five years in a busy small animal practice, developed an interest in radiology and ultrasound, and gained the RCVS Certificate in Veterinary Diagnostic Imaging. Abby returned to Cambridge in 2006 to undertake a Residency program in Diagnostic Imaging, and gained the European Diploma in Veterinary Diagnostic Imaging in 2009. Abby is a European Specialist in Veterinary Diagnostic Imaging and an RCVS fellow. She has published many scientific publications and book chapters in the field, and is a former executive board member of ECVDI and former chair of the EAVDI BID . She is current Chair of the ECVDI examination committee. She is interested in all aspects of imaging, and is currently researching advance imaging of the feline biliary tract as part of the Vet MD programme at Cambridge Vet school.

Chaired by Olivier Restif

• Speaker: Abigail Caine, Department of Veterinary Medicine
• Friday 06 June 2025, 08:45-10:00
• Venue: LT2.
• Series: Friday Morning Seminars, Dept of Veterinary Medicine; organiser: Fiona Roby.

Add to your calendar or Include in your list

Long-lived circularly polarized organic long persistent luminescence (CP-OLPL) is achieved by constructing a chiral exciplex system, with green CP-OLPL emission lasting over 1.5 hours and exhibiting an asymmetry factor of 4.5 × 10− 3. Additionally, by incorporating a fluorophore emitter and utilizing synergistic singlet-singlet and chirality energy transfer, an orange-red CP-OLPL with a duration exceeding 1 hour is realized, showcasing its potential for applications in afterglow display and lighting, and multi-level information encryption.

Abstract

Circularly polarized organic long persistent luminescence (CP-OLPL) has garnered significant attention due to its distinctive properties. However, achieving CP-OLPL materials with ultralong durations remains a formidable challenge. Herein, an effective strategy is proposed to obtain long-lived CP-OLPL by constructing a self-designed chiral donor for developing a host–guest chiral exciplex system. The gradual recombination of long-lived charge-separated states enables a green CP-OLPL emission to persist for over 1.5 hours with an asymmetry factor (|g lum|) of 4.5 × 10−3. More intriguingly, doping with rubrene fluorophore yields an orange-red CP-OLPL system, exhibiting a duration over 1 hour and |g lum| of 2.3 × 10−3 through synergistic singlet-singlet and chirality energy transfer. These properties render the development of chiral afterglow display, multi-level information encryption, and afterglow lighting. This work not only represents a significant advancement in the design of chiral donors for ultralong CP-OLPL exciplex system with durations but also provides valuable insights into exciton dynamics.

Ubiquitous defects predominately account for photo-instability and open-circuit voltage losses in wide-bandgap perovskite solar cells (WBG PSCs). This review comprehensively presents the underlying impact mechanisms, summarizes the advanced optimization strategies across various functional layers and their interfaces to develop efficient and stable WBG PSCs, and evaluates their performance in semitransparent solar cells, tandem solar cells, and indoor photovoltaic applications.

Abstract

Wide-bandgap (WBG) perovskite solar cells (PSCs) have garnered considerable attention of late for their potential as semitransparent photovoltaics for building integration, top-cells in tandem configurations, and indoor photovoltaics (IPVs) for Internet of Things (IoT) applications. However, recent investigations have unveiled that underlying defect-mediated phase segregation, ion migration, lattice strain, and other factors can give rise to self-accelerated degradation reactions and the contraction of quasi-Fermi level splitting (QFLS) within devices. Extensive efforts have been undertaken to reduce defect densities in bulks, at surfaces, and across interfaces with charge transport layers (CTLs). This review provides a timely and comprehensive understanding of the intrinsic defect ecosystem in WBG perovskites, and mechanistically elucidates their impacts on device stability and open circuit voltage losses. Subsequently, recent advances in defect passivation strategies are cross-sectionally overviewed, covering various components of devices. The applications of WBG PSCs in semitransparent devices, tandem applications, and IPVs are discussed. Finally, prospects and challenges are proposed, providing insights for future research and technological advancements.

This study demonstrates a p-type cuprous sulfide lattice connectedly capping over Cs3Cu2I5 to form lattice-matched core/shell nanocrystals by controlling the reactivity of sulfur precursor in the synthesis. The resulting pure white PeLEDs achieved by combining with yellow emission CsCu2I3 exhibit a recorded EQE of 3.45% and a high CRI of 91.

Abstract

The copper-based (Cu-based) halide perovskite possesses eco-friendly features, bright self-trapped-exciton (broadband) emission, and a high color-rendering index (CRI) for achieving white emission. However, the limited hole injection (HI) of Cu-based perovskites has been bottle-necking the efficiency of white electroluminescence and thus their application in white perovskite light-emitting diodes (W-PeLEDs). In this study, we demonstrate a p-type cuprous sulfide (Cu2S) lattice-connectedly capping over Cs3Cu2I5 to form lattice-matched core/shell nanocrystals (NCs) by controlling the reactivity of sulfur (S) precursor in the synthesis. Interestingly, the resultant Cs3Cu2I5/Cu2S NCs significantly enhance the hole mobility compared to Cs3Cu2I5 NCs. Besides, the photoluminescence quantum yield of Cs3Cu2I5 NCs increases from 26.8% to 70.6% after the Cu2S lattice-connected capping. Consequently, by establishing the structure of CsCu2I3/Cs3Cu2I5/Cu2S in W-PeLEDs, an external quantum efficiency of 3.45% and a CRI of 91 is realized, representing the highest reported electroluminescent performance in lead-free Cu-based W-PeLEDs. These findings contribute to establishing guidelines and effective strategies for designing broadband electroluminescent materials and device structures of PeLEDs.

Real-space nanoimaging of unidirectional hybridized exciton-polariton (EPs). The resolution to observe near-field SOC can be very high (down to 10 nm) and the contrast can be clearly distinguished at the real-space. The work provides an alternative but more direct approach to uncover the underlying physics of optical SOC and EP properties.

Abstract

Unidirectional excitation of highly confined guided modes is essential for nanoscale energy transport, photonic integrated devices, and quantum information processing. Among various feasible approaches, the mechanism based on optical spin–orbit coupling is investigated for unidirectional routing of surface plasmons and valley exciton-polaritons, without needing the use of complicated magneto-optical effects and parity symmetry breaking. So far, the direct near-field nanoimaging of such exotic polaritonic modes based on optical spin–orbit coupling has remained elusive. Here, the real-space nanoimaging of unidirectional polaritons in van der Waals semiconductors are reported. Specifically, photonic spins are coupled into the tip of a scattering-type scanning near-field optical microscopy for circular dipolar excitations of spin–orbit interactions, thus enabling the unidirectional waveguide exciton-polariton propagation with remarkable unidirectionality (ratio of spectrum amplitudes under opposite circularly polarized illumination) R = 0.291 for TM mode. Via switching to the opposite helicities, the reversed opposite directions are observed. The work offers a promising avenue for detecting and processing spin information for future communication technology at the nanoscale.

The large-pore MSNs with a high specific surface area serve as a platform for covalent PMB grafting and bactericidal CQD loading. CQD@lMSN-PMB exhibits efficient antibacterial activity alongside synchronous endotoxin neutralization. In Gram-negative bacterial keratitis, CQD@lMSN-PMB effectively and comprehensively suppresses the immune-inflammatory response, preserving corneal integrity and transparency.

Abstract

In the treatment of infectious keratitis, therapeutic strategies often prioritize enhancing bactericidal efficacy. However, endotoxins released from Gram-negative bacteria cause inflammatory reaction, leading to corneal structural damage and scar formation. Given that polymyxin B (PMB) can bind and neutralize lipopolysaccharide (LPS), this study employs large-pore mesoporous silica nanoparticles (lMSNs) grafted with PMB as carriers for cationic antibacterial carbon quantum dots (CQDs) to prepare CQD@lMSN-PMB, which enables synchronous sterilization and endotoxin neutralization. In the acidic infectious microenvironment, the accelerated release of CQDs eliminates 99.88% bacteria within 2 h, effectively substituting immune mediated sterilization. Notably, CQD@lMSN-PMB exhibits exceptional LPS neutralization performance (2.22 µg LPS/mg CQD@lMSN-PMB) due to its high specific surface area. In an infectious keratitis model, inflammation subsides significantly within the first day of CQD@lMSN-PMB intervention and is completely resolved by day 3. By day 2, interleukin-1β, interleukin-6 and tumor necrosis factor-α in CQD@lMSN-PMB group decrease by 86.99%, 91.15%, and 77.56%, respectively, compared to the CQDs-only sterilization group. Ultimately, corneal integrity and transparency are preserved, with suppressed expressions of fibrosis-related factors including matrix metalloproteinase 9, transforming growth factor-β and α-smooth muscle actin. Therefore, this synchronous sterilization and endotoxin neutralization strategy outperforms monotherapy strategies focused solely on sterilization or endotoxin neutralization.

This study presents an effective strategy for developing high-performance hybrid organic-inorganic lanthanide halide glass scintillators for X-ray imaging through the dehydration of crystalline precursors. This process not only effectively suppresses melt crystallization but also enhances radioluminescence by improving X-ray-to-visible photon conversion efficiency. Thus, it enables the fabrication of large, transparent scintillators suitable for high-resolution X-ray imaging.

Abstract

Although glass scintillators hold great promise for high-resolution X-ray imaging, the practical application is often limited by thermodynamic instability, leading to uncontrolled glass-to-crystal transformations that degrade imaging resolution. Herein, a novel strategy is presented to synthesize (methyl(triphenyl)phosphonium)3EuCl6 ((MTP)3EuCl6) glass scintillators through the dehydration of their crystalline precursors. The findings reveal that the dehydration process significantly enhances the stability of the glass scintillators by confining the constituent ions within a rigid, highly viscous matrix. This confinement effectively restricts ion mobility and prevents the reorganization required for crystal nucleation. Moreover, the dehydration reduces the trapping of in situ generated charge carriers and increases the photoluminescence quantum yield, leading to enhanced radioluminescence performance. The resulting (MTP)3EuCl6 glass scintillators demonstrate an X-ray detection limit as low as 95.8 nGyair s⁻¹ and achieve a spatial imaging resolution of 14.3 lp mm−1 at a dose rate of 5 mGyair s⁻¹. This work provides valuable insights into designing glass scintillators that integrate long-term thermodynamic stability with optimized scintillation performance, offering significant potential for advanced X-ray imaging applications.

A strategy is implemented to embed dermal fibroblasts in collagen in contact with anisotropic hydrogel fibers. Fibroblasts generate traction forces and remodel the collagen matrix, following the alignment cues of the hydrogel template. The orientational order is transferred to the fibroblasts, inducing collective alignment. Cell alignment generates internally driven, oriented forces that ultimately lead to programmed 2D and 3D shape changes of the cell-laden collagen matrices, such as transforming from square to diamond, or from flat to dome. This approach provides insights into the role of topological defects in morphogenesis and advances new techniques for developing shape-morphing materials with programmable forces.

Abstract

During morphogenesis, cells collectively execute directional forces that drive the programmed folding and growth of the layers, forming tissues and organs. The ability to recapitulate aspects of these processes in  vitro will constitute a significant leap forward in the field of tissue engineering. Free-standing, self-organizing, cell-laden matrices are fabricated using a sequential deposition approach that uses liquid crystal-templated hydrogel fibers to direct cell arrangements. The orientation of hydrogel fibers is controlled using flow or boundary cues, while their microstructures are controlled by depletion interaction and probed by scattering and microscopy. These fibers effectively direct cells embedded in a collagen matrix, creating multilayer structures through contact guidance and by leveraging steric interactions amongst the cells. In uniformly aligned cell matrices, oriented cells exert traction forces that can induce preferential contraction of the matrix. Simultaneously, the matrix densifies and develops anisotropy through cell remodeling. Such an approach can be extended to create cell arrangements with arbitrary in-plane patterns, allowing for coordinated cell forces and pre-programmed, macroscopic shape changes. This work reveals a fundamentally new path for controlled force generation, emphasizing the role of a carefully designed initial orientational field for manipulating shape transformations of reconstituted matrices.

An organic BTHG agent called OTBP with prominent crystallinity is developed. The resonant absorption at 433 nm and prominent crystallinity of OTBP enhance its BTHG efficiency when excited by a 1300 nm femtosecond laser. Tuning the size of OTBP nanocrystals based on Mie scattering theory successfully enables them to achieve high-contrast muti-organ angiography with ultra-low pulse energy (100 pJ) and thereby negligible photodamage risk.

Abstract

Organic materials featuring third harmonic generation (THG) hold great promise for deep-tissue bioimaging due to their good biocompatibility and second near-infrared excitation. However, minimizing photodamage from the incident light necessitates significant improvements in the third-order nonlinear susceptibility. Herein, an organic luminogen called OTBP is developed as a backward THG (BTHG) contrast agent for second near-infrared (NIR-II) angiography. OTBP's intense absorption at 433 nm resonantly enhances its BTHG efficiency when excited by a 1300 nm femtosecond laser. In the aggregate state, the robust intermolecular interactions among OTBP molecules realize excellent crystallinity and the facile preparation of nanocrystals (NCs) with a high refractive index of 1.78. By leveraging Mie scattering theory, the best size of OTBP NCs for BTHG collection is attained. These integrated properties result in a high BTHG efficiency of OTBP NCs. Encapsulating the NCs with F-127 enables ultralow-power but high-contrast 3D vasculature imaging with negligible photodamage and background interference. Further elevating the laser power to 60 mW enables the visualization of microvessels at 500 µm with a high SNR of 143. This study offers insights into material design strategies toward efficient organic BTHG contrast agents and paves the way for the materials-oriented non-linear optics.

A novel interfacial chemical tuning strategy is proposed involving proton transfer between the amine of pyridoxamine (PM) and the phosphonic acid of Me-4PACz. This process enhances charge delocalization via electrostatic attraction between oppositely charged molecules, yielding a Me-4PACz-PM charge polarization interface that improves NiOx p-type conductivity, facilitates band alignment, and supports high-quality perovskite films, showcasing its commercial potential.

Abstract

Interfacial localized charges and interfacial losses from incompatible underlayers are critical factors limiting the efficiency improvement and market-integration of perovskite solar cells (PSCs). Herein, a novel interfacial chemical tuning strategy is proposed involving proton transfer between the amine head of pyridoxamine (PM) and the phosphonic acid anchoring group of [4-(3,6-dimethyl-9H-carbazol-9-yl)butyl]phosphonic acid (Me-4PACz), with simultaneous enhancement of charge delocalization through electrostatic attraction between opposite charged molecules. The Me-4PACz-PM charge polarization interface modulates the nickel oxide (NiOx) charge states and the coordination environment at buried interfaces, consequently enhancing p-type conductivity and obtaining a more compatible band arrangement. The high-coverage and wettability of the NiOx/Me-4PACz-PM underlayer also facilitate the deposition of high-quality perovskite films, releasing lattice strain and mitigating trap-assisted non-radiative recombination. Attributing to the implementation of charge polarization tunable interfaces, small-area devices and modules with an aperture area of 69 cm2 achieved impressive power conversion efficiencies (PCEs) of 26.34% (certified 25.48%) and 21.94% (certified 20.50%), respectively, and unencapsulated devices maintained their initial PCEs ≈90% after aging for 2000 h (ISOS-L-1) and 1500 h (ISOS-D-1). The broad applicability of charge polarization tunable interfaces and the successful scaling of large-area modules provide a reference for expanding PSCs applications.

Advanced Materials, EarlyView.

Out-of-context reasoning/learning in LLMs and its safety implications

Beyond learning patterns within individual training datapoints, Large Language Models (LLMs) can infer latent structures and relationships by aggregating information scattered across different training samples through out-of-context reasoning (OOCR) [1, 2]. We’ll review key empirical findings, including Implicit Meta-Learning (models learning source reliability implicitly and subsequently internalizing reliable-seeming data more strongly [1]) and Inductive OOCR (models inferring other latent structures from scattered data [3]). We’ll explore potential mechanisms behind these phenomena [1, 4]. Finally, we’ll discuss the significant AI safety implications, arguing that OOCR coupled with Situational Awareness [5] underpins threats like Alignment Faking [6], potentially leading to persistent misalignment resistant to standard alignment techniques.

1. Krasheninnikov et al., “Implicit meta-learning may lead language models to trust more reliable sources” https://arxiv.org/abs/2310.15047 2. Berglund et al., “Taken Out of Context: On Measuring Out-of-Context Reasoning in LLMs” https://arxiv.org/abs/2309.00667 3. Treutlein et al., “Connecting the Dots: LLMs can Infer and Verbalize Latent Structure from Disparate Training Data” https://arxiv.org/abs/2406.14546 4. Feng et al., “Extractive Structures Learned in Pretraining Enable Generalization on Finetuned Facts” https://arxiv.org/abs/2412.04614 5. Laine et al., “Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs” https://arxiv.org/abs/2407.04694 6. Greenblatt et al., “Alignment faking in large language models” https://arxiv.org/abs/2412.14093

Teams link available upon request (it is sent out on our mailing list, eng-mlg-rcc [at] lists.cam.ac.uk). Sign up to our mailing list for easier reminders via lists.cam.ac.uk.

• Speaker: Dmitrii Krasheninnikov, Usman Anwar, University of Cambridge
• Wednesday 02 April 2025, 11:00-12:30
• Venue: Cambridge University Engineering Department, CBL Seminar room BE4-38..
• Series: Machine Learning Reading Group @ CUED; organiser: .

Add to your calendar or Include in your list

Title to be confirmed

Chaired by Olivier Restif

• Speaker: Abigail Caine, Department of Veterinary Medicine
• Friday 06 June 2025, 08:45-10:00
• Venue: LT2.
• Series: Friday Morning Seminars, Dept of Veterinary Medicine; organiser: Fiona Roby.

Add to your calendar or Include in your list

Aqueous redox flow batteries are limited by the competing hydrogen evolution reaction (HER) at their negative electrodes. In this work, conductive polymers are conformally coated on porous carbonaceous electrodes to improve the reaction selectivity of hybrid all-iron redox flow batteries. Poly(pyrrole)/PSS coating inhibits HER in acidic media and increases the battery lifetime, while PEDOT/PSS electrode improves the roundtrip efficiency.

Abstract

Aqueous all-iron redox flow batteries are an attractive and economic technology for grid-scale energy storage owing to their use of abundant and environmentally benign iron as the redox active material and water as solvent. However, the battery operation is challenged by the plating/stripping reactions of iron and the competing hydrogen evolution reaction at the negative electrode, which hinder performance and durability. Here, the reaction selectivity of the negative electrode is tailored by introducing conductive polymer coatings onto porous carbonaceous electrodes. Two conductive polymers, poly(3,4-ethylenedioxythiophene) (PEDOT) and poly(pyrrole) (PPy) are conformally coated with the dopant poly(4-styrenesulfonate) (PSS) and the resulting electrochemistry is studied on model electroanalytical platforms and redox flow batteries. Both polymers decrease the hydrogen evolution current on rotating disc electrodes, with PPy/PSS strongly inhibiting the reaction at high overpotentials. In full all-iron redox flow cells, PPy/PSS coating extends cyclability and significantly reduces hydrogen evolution, while PEDOT/PSS coating improves the round-trip efficiency, possibly acting as a redox shuttle for the iron stripping reaction. These findings motivate broader investigation and implementation of conductive polymers to engineer reaction selectivity for flow batteries and other electrochemical technologies.

This research proposes a multistage competitive coordination chromogenic mechanism between metal-polyphenols and amines. Based on this, a metal-polyphenol network colorimetric sensor array (MPN-CSA) for real-time monitoring of meat freshness has been developed, achieving a detection rate of 99.83% through convolutional neural network technology. This sensitive, accurate, stable, eco-friendly, and economical solution can enhance food safety and reduce food waste.

Abstract

A low-cost, high-precision, and secure real-time system for monitoring food freshness can significantly improve spoilage issues, yet traditional colorimetric sensor arrays often suffer from chemical dyes’ high toxicity and limited color changes. Here, a metal-polyphenol network colorimetric sensor array (MPN-CSA) is built for detecting total volatile base nitrogen (TVB-N) markers of meat freshness. The multi-level competitive coordination process between the metal-polyphenol system and amine substances endows the system with color changes far beyond those of traditional dyes (reaching a detection limit of 300 ppb). By integrating convolutional neural network (CNN) technology, an online platform is developed for monitoring meat freshness, achieving an overall detection accuracy rate of 99.83%. This environmentally friendly, economically viable MPN-CSA that monitors the freshness of meat in complex storage environments can be incorporated into food packaging boxes, enabling consumers and suppliers to assess the freshness of meat in real-time, thus helping to reduce food waste and prevent foodborne illnesses.

Nature Materials, Published online: 01 April 2025; doi:10.1038/s41563-025-02217-7

Metamaterials are advancing with intricate structure designs and material combinations, with the support of computational methods and scalable fabrication techniques. These advancements enable the creation of multifunctional and smart devices, with growing presence in commercial devices.