Nanoscience News (<front>)

PhD Students' talks

Abstract not available

• Speaker: Speakers listed in abstract in due course
• Friday 30 May 2025, 14:00-17:00
• Venue: MR2.
• Series: Fluid Mechanics (DAMTP); organiser: Professor Grae Worster.

Add to your calendar or Include in your list

Translation Validation for LLVM's AArch64 Backend

Alive2 is a practical oracle for determining whether a transformation on LLVM IR is a refinement—that is, whether it is valid under the rules for LLVM optimizations. In this talk I’ll describe an analogous translation validation solution for LLVM ’s AArch64 backend that we’ve used to find 42 miscompilation bugs, many of which were in architecture-neutral code and hence could have also affected other backends. Our tool, arm-tv, reuses Alive2 as a source of LLVM semantics and offers a choice of two AArch64 semantics, one that we wrote by hand and the other derived from ARM ’s machine readable specification of their ISA .

John Regehr is a computer science professor at the University of Utah, USA . He liked to build tools for compiler developers to use, and then write papers about them.

If you want to attend the compiler social, please remember to sign up: https://grosser.science/compiler-social-2025-06-05/

• Speaker: John Regehr - University of Utah
• Thursday 05 June 2025, 15:00-16:00
• Venue: Computer Laboratory, William Gates Building, LT1.
• Series: compiler socials; organiser: Luisa Cicolini.

Add to your calendar or Include in your list

Explainable AI in Neuroscience: From Interpretability to Biomarker Discovery

Explainability plays a pivotal role in building trust and fostering the adoption of artificial intelligence (AI) in healthcare, particularly in high-stakes domains like neuroscience where decisions directly affect patient outcomes. While progress in AI interpretability has been substantial, there remains a lack of clear, domain-specific guidelines for constructing meaningful and clinically relevant explanations. In this talk, I will explore how explainable AI (XAI) can be effectively integrated into neuroscience applications. I will outline practical strategies for leveraging interpretability methods to uncover novel patterns in neural data, and discuss how these insights can inform the identification of emerging biomarkers. Drawing on recent developments, I will highlight adaptable XAI frameworks that enhance transparency and support data-driven discovery. To validate these concepts, I will present illustrative case studies involving large language models (LLMs) and vision transformers applied to neuroscience. These examples serve as proof of concept, showcasing how explainable AI can not only translate complex model behavior into human-understandable insights, but also support the discovery of novel patterns and potential biomarkers relevant to clinical and research applications.

• Speaker: Mike Mamalakis (University of Cambridge)
• Tuesday 13 May 2025, 13:00-14:00
• Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building.
• Series: Artificial Intelligence Research Group Talks (Computer Laboratory); organiser: Mateja Jamnik.

Add to your calendar or Include in your list

Thermally fueled, self-moving twisted rings made from liquid crystal elastomers are investigated. By studying single rings and linked knots, it is shown how autonomous locomotion emerges through coupling between the rings. The movement is programmable via handedness control at connection points, offering insights into collective behavior and programmable motion in soft materials.

Abstract

In nature, the interplay between individual organisms often leads to the emergence of complex belabours, of which sophistication has been refined through millions of years of evolution. Synthetic materials research has focused on mimicking the natural complexity, e.g., by harnessing non-equilibrium states to drive self-assembly processes. However, it is highly challenging to understand the interaction dynamics between non-equilibrium entities and to obtain collective behavior that can arise autonomously through interaction. In this study, thermally fueled, twisted rings exhibiting self-sustained movements are used as fundamental units and their interactive behaviors and emergent functions are investigated. The rings are fabricated from connected thermoresponsive liquid crystal elastomers (LCEs) strips that undergo zero-elastic-energy-mode, autonomous motions upon a heat gradient. Single-ring structures with various twisting numbers and nontrivial links, and connected knots where several LCE rings (N = 2,3,4,5) are studied and linked. The observations uncover that controlled locomotion of the structures can emerge when N ≥ 3. The locomotion can be programmed by controlling the handedness at the connection points between the individual rings. These findings illustrate how group activity emerges from individual responsive material components through mechanical coupling, offering a model for programming autonomous locomotion in soft matter constructs.

A high-performance anti-sintering catalyst with efficient and ultra-stable Ni-W catalytic layer on reductive WC (NiAWC) with stabilized Niδ+ sites for superior high-temperature RWGS reaction. Here the concept of design of high-performance catalysts through metal-substrate synergistic effects offers a promising path to engineering superior high-temperature thermal catalysis.

Abstract

The reverse water-gas shift (RWGS) reaction stands out as a promising approach for selectively converting CO2 into CO, which can then be upgraded into high-value-added products. While designing high selectivity and stability catalysts for RWGS reaction remains a significant challenge. In this study, an efficient and ultra-stable Ni-W catalytic layer on reductive WC (NiAWC) is designed as an anti-sintering catalyst for superior high-temperature RWGS reaction. Benefiting from the unique structures, the NiAWC catalyst exhibits exceptionally high performances with a CO production rate of 1.84 molCO gNi −1 h−1 and over 95% CO selectivity, maintaining stability for 120 h at 500 °C. Even after 300 h of continuous testing at 600 °C and five aging cycles at 800 °C, the activity loss is only 0.34% and 0.83%, respectively. Unlike the conventional mechanism in RWGS reaction, it is demonstrated that the Ni-W limited coordination can stabilize the Ni sites and allow a pre-oxidation of Niδ+ by CO, which produces an O* electronic reservoir and hinders the charge transfer from Ni to W-O, thereby avoiding the dissolution of Ni atoms. The design of new, efficient, and selective catalysts through metal-substrate synergistic effects is suggested to offer a promising path to engineering superior thermal catalysts.

3D-printed picoliter droplet networks have been fabricated that control gene expression in bacterial populations by releasing chemical signals with precise spatial definition and high temporal resolution. This system of effector release is widely applicable, offering diverse applications in biology and medicine.

Abstract

Synthetic cells, such as giant unilamellar vesicles, can be engineered to detect and release chemical signals to control target cell behavior. However, control over target-cell populations is limited due to poor spatial or temporal resolution and the inability of synthetic cells to deliver patterned signals. Here, 3D-printed picoliter droplet networks are described that direct gene expression in underlying bacterial populations by patterned release of a chemical signal with temporal control. Shrinkage of the droplet networks prior to use achieves spatial control over gene expression with ≈50 µm resolution. Ways to store chemical signals in the droplet networks and to activate release at controlled points in time are also demonstrated. Finally, it is shown that the spatially-controlled delivery system can regulate competition between bacteria by inducing the patterned expression of toxic bacteriocins. This system provides the groundwork for the use of picoliter droplet networks in fundamental biology and in medicine in applications that require the controlled formation of chemical gradients (i.e., for the purpose of local control of gene expression) within a target group of cells.

A macrophage-evading nano-lubricant is designed to enhance osteoarthritis (OA) treatment by suppressing complement C3 adsorption and subsequent macrophage phagocytosis. Through a zwitterionic PMPC brush layer, this strategy reduces inflammation, preserves joint lubrication, and prevents OA progression. This work highlights the pivotal role of complement C3 in nanoparticle clearance and offers a novel therapeutic approach for OA management.

Abstract

Administering a bio-lubricant is a promising therapeutic approach for the treatment of osteoarthritis (OA), in particular, if it can both manage symptoms and halt disease progression. However, the clearance of these bio-lubricants mediated by synovial macrophages leads to reduced therapeutic efficiency and adverse inflammatory responses. Herein, it is shown that this process is predominantly mediated by the specific binding of complement C3 (on nanoparticle) and CD11b (on macrophage). More importantly, through a systematic evaluation of various interface modifications, a macrophage-evading nanoparticle strategy is proposed, which not only minimizes friction, but also largely suppresses C3 adsorption. It involves employing a zwitterionic poly-2-methacryloyloxyethyl phosphorylcholine (PMPC) brush layer grafted from a crosslinked gelatin core. In vitro studies demonstrate that such a nanoparticle lubricant can evade macrophage phagocytosis and further prevent the pro-inflammatory M1 polarization and subsequent harmful release of cytokines. In vivo studies show that the designed PMPC brush layer effectively mitigates synovial inflammation, alleviates OA-associated pain, and protects cartilage from degeneration, thus preventing OA progression. These findings clarify the pivotal role of complement C3-mediated macrophage recognition in nanoparticles clearance and offer a promising nanoparticle design strategy to restore joint lubrication.

TBC

TBC

• Speaker: Anja Meyer, University of Loughborough
• Wednesday 21 May 2025, 16:30-17:30
• Venue: MR12.
• Series: Algebra and Representation Theory Seminar; organiser: Adam Jones.

Add to your calendar or Include in your list

A non-contact intelligent sensing system that can accurately recognize patterns formed by NIR light is reported. Black phosphate (BP)-based organogel with anti-freezing and water retention properties as non-contact sensor can be used in extremely cold or hot environments. The designed non-contact sensing system shows good robustness by maintaining high sensitivity over a wide temperature range, long working distances, different current intensities, and dark conditions.

Abstract

With the development of artificial intelligence and the Internet of Things, non-contact sensors are expected to realize complex human-computer interaction. However, current non-contact sensors are mainly limited by accuracy and stability. Herein, an intelligent infrared photothermal non-contact sensing system is developed that provides long-distance and high-accuracy non-contact sensing. A black phosphorus (BP)-based composite organogel is designed, which exhibits excellent photothermal properties and environmental stability, as the active material. This material can detect patterns created by near-infrared (NIR) light through various patterned masks monitored by an infrared thermal imager. The constructed non-contact sensing system is capable of accurately recognizing 26 letters with an impressive accuracy rate of 99.4%. Furthermore, even small size non-contact sensors can maintain high sensitivity and stability across a wide temperature range, at long working distances, and under different current intensities and dark conditions, demonstrating exceptional robustness. Combined with machine learning method, it is demonstrated that the non-contact sensing system excels in pattern recognition and human-computer interaction. These features highlight its potential applications in intelligent robotics and remote control systems.

Recreating the Physical Natural World from Images

For centuries, unraveling the mysteries of nature through the lens of physics has captivated countless scientists. Today, generative AI models excel at reproducing visual worlds in pixels, but still struggle with basic physical concepts such as 3D shape, motion, material, and lighting—-key elements that connect computer vision to a wide range of real-world engineering applications, including interactive VR, robotics, biology, and medical analysis. The main challenge arises from the difficulty of collecting large-scale physical measurements for training machine learning models.

In this talk, I will discuss an alternative unsupervised approach based on inverse rendering, which enables machine learning models to learn explicit physical representations from raw, unstructured image data, such as Internet photos and videos. This approach thus circumvents the need for any direct supervision, allowing us to model a wide variety of 3D objects in nature, including diverse wildlife, using only casually recorded imagery. The resulting model can generate physically-grounded 3D assets with controllable animations instantly, ready for downstream rendering and analysis. The papers presented can be found at: https://elliottwu.com/.

Link to join virtually: https://cam-ac-uk.zoom.us/j/87421957265

This talk is being recorded. If you do not wish to be seen in the recording, please avoid sitting in the front three rows of seats in the lecture theatre. Any questions asked will also be included in the recording. The recording will be made available on the Department’s webpage

• Speaker: Dr Elliott Wu - Department of Engineering, University of Cambridge
• Wednesday 21 May 2025, 15:05-15:55
• Venue: Lecture Theatre 1, Computer Laboratory, William Gates Building.
• Series: Wednesday Seminars - Department of Computer Science and Technology ; organiser: Ben Karniely.

Add to your calendar or Include in your list

Type-driven Development with Idris 2

Idris is a functional programming language with first-class types, which allow properties to be expressed in the type system, and with an interactive type-driven editor which allows programs to be developed as a formal conversation with the machine. In this talk I will introduce Idris and its type system, and cover recent developments in Idris 2. In particular, I will describe how the quantities in the type system give additional expressivity which allows us to implement state machines and communicating systems and verify their properties, interactively.

Link to join virtually: https://cam-ac-uk.zoom.us/j/87421957265

This talk is being recorded. If you do not wish to be seen in the recording, please avoid sitting in the front three rows of seats in the lecture theatre. Any questions asked will also be included in the recording. The recording will be made available on the Department’s webpage

• Speaker: Dr Edwin Brady - School of Computer Science, University of St Andrews
• Wednesday 14 May 2025, 15:05-15:55
• Venue: Lecture Theatre 1, Computer Laboratory, William Gates Building.
• Series: Wednesday Seminars - Department of Computer Science and Technology ; organiser: Ben Karniely.

Add to your calendar or Include in your list

Translation Validation for LLVM's AArch64 Backend

Alive2 is a practical oracle for determining whether a transformation on LLVM IR is a refinement—that is, whether it is valid under the rules for LLVM optimizations. In this talk I’ll describe an analogous translation validation solution for LLVM ’s AArch64 backend that we’ve used to find 42 miscompilation bugs, many of which were in architecture-neutral code and hence could have also affected other backends. Our tool, arm-tv, reuses Alive2 as a source of LLVM semantics and offers a choice of two AArch64 semantics, one that we wrote by hand and the other derived from ARM ’s machine readable specification of their ISA .

John Regehr is a computer science professor at the University of Utah, USA . He liked to build tools for compiler developers to use, and then write papers about them.

• Speaker: John Regehr - University of Utah
• Thursday 05 June 2025, 15:00-16:00
• Venue: Computer Laboratory, William Gates Building, LT1.
• Series: compiler socials; organiser: Luisa Cicolini.

Add to your calendar or Include in your list

Vertical mixing and associated biogeochemical fluxes via radium isotopes

Abstract not available

• Speaker: Tristan McKenzie, Uni of Gothenberg
• Thursday 15 May 2025, 11:30-12:30
• Venue: Open Plan Area, Institute for Energy and Environmental Flows, Madingley Rise CB3 0EZ.
• Series: Institute for Energy and Environmental Flows (IEEF); organiser: Catherine Pearson.

Add to your calendar or Include in your list

A copper cluster scintillator with excellent stability and scintillation performance, which shows high sensitivity response to α/β particles, is constructed. By coupling it with PMT and nuclear electronics system, a surface contamination monitor for precise detection of α/β particles is successfully fabricated.

Abstract

High-energy radiation is widely used in medicine, industry, and scientific research. Meanwhile, the detection of environmental ionizing radiation is essential to ensure the safe use of high-energy radiation. Among radiation detectors, scintillator detectors offer multiple advantages, including simple structure, high sensitivity, excellent environmental adaptability, and a favorable performance-to-price ratio. However, the development of high-performance scintillators that can provide highly sensitive responses to environmental radiation, especially α/β particles, remains a challenge. In this work, a copper cluster (Cu4I4(DPPPy)2 ) with excellent water-oxygen stability is prepared using a simple one-pot method at room temperature. Cu4I4(DPPPy)2 not only exhibits excellent X-ray excited luminescence (XEL) under X-ray irradiation but also demonstrates a highly sensitive scintillation response to α/β particles. By integrating Cu4I4(DPPPy)2 with a photomultiplier tube (PMT) and nuclear electronics, an α/β surface contamination monitor is successfully developed. This monitor enables the sensitive detection of excessive α/β particles in real-world environments. The detection frequency and signal intensity of Cu4I4(DPPPy)2 significantly surpass those of commercial scintillator of YAP:Ce, BGO, PbWO4, and anthracene under identical conditions, highlighting the promising application of metal clusters in low-dose environmental radiation detection.

Conditional Expectation and Machine Learning

The problem addressed by machine learning can also be formulated as one of computing conditional expectation, an approach little explored because of the success of machine learning. The focus of this presentation is to view conditional expectation, not as an alternative, but as a means of performance enhancement for machine learning. In particular, we show that conventional machine learning is itself a vehicle for computing conditional expectation, both post training and during training. A neural network architecture that combine conventional machine learning and computing conditional expectation will be presented.

• Speaker: Eugene Wong(UC Berkeley and Clare Hall, Cambridge)
• Wednesday 28 May 2025, 11:00-12:30
• Venue: Cambridge University Engineering Department, CBL Seminar room BE4-38..
• Series: Machine Learning Reading Group @ CUED; organiser: Xianda Sun.

Add to your calendar or Include in your list

Conditional Expectation and Machine Learning

The problem addressed by machine learning can also be formulated as one of computing conditional expectation, an approach little explored because of the success of machine learning. The focus of this presentation is to view conditional expectation, not as an alternative, but as a means of performance enhancement for machine learning. In particular, we show that conventional machine learning is itself a vehicle for computing conditional expectation, both post training and during training. A neural network architecture that combine conventional machine learning and computing conditional expectation will be presented.

• Speaker: Eugene Wong(UC Berkeley and Clare Hall, Cambridge)
• Wednesday 28 May 2025, 11:00-12:30
• Venue: Cambridge University Engineering Department, CBL Seminar room BE4-38..
• Series: Machine Learning Reading Group @ CUED; organiser: Xianda Sun.

Add to your calendar or Include in your list

Detection of antiferromagnetic spin texture in a 2D magnetic crystal is achieved through nanomechanical resonators at radio frequencies. Sharp magnetic transitions that lead to abrupt changes in mechanical linear and nonlinear responses are assigned to antiferromagnetic domain motions. The results indicate rich and fluid-like dynamics between the coupled spin and lattice at the transition field.

Abstract

The coupling between the spin degrees of freedom and macroscopic mechanical motions, including striction, shearing, and rotation, has attracted wide interest with applications in actuation, transduction, and information processing. Experiments so far have established the mechanical responses to the long-range ordered or isolated single spin states. However, it remains elusive whether mechanical motions can couple to a different type of magnetic structure, the non-collinear spin textures, which exhibit nanoscale spatial variations of spin (domain walls, skyrmions, etc.) and are promising candidates to realize high-speed computing devices. Here, collective spin texture dynamics is detected with nanoelectromechanical resonators fabricated from 2D antiferromagnetic (AFM) MnPS3 with 10−9 strain sensitivity. By examining radio frequency mechanical oscillations under magnetic fields, new magnetic transitions are identified with sharp dips in resonant frequency. They are attributed to collective AFM domain wall motions as supported by the analytical modeling of magnetostriction and large-scale spin-dynamics simulations. Additionally, an abnormally large modulation in the mechanical nonlinearity at the transition field infers a fluid-like response due to ultrafast domain motion. The work establishes a strong coupling between spin texture and mechanical dynamics, laying the foundation for electromechanical manipulation of spin texture and developing quantum hybrid devices.

This work uncovers the Li0 reutilization behavior of AFLMBs at different discharge rates, which exhibits a “volcano-type” variation. The opposite effects of the distribution relationship between fresh Li and residue Li0 and concentration polarization at specific discharge rate dominate Li0 reutilization. This cognition provides guidance toward high-power density AFLMBs under practical conditions.

Abstract

Anode-free lithium metal battery (AFLMB) has become an excellent candidate for long endurance electric vehicles and electric low altitude aircraft, profiting from its high energy density as well as outstanding manufacturing safety. However, the limitation at high discharge rates of AFLMBs is shrouded in mystery, yet to achieve more attention. Herein, the limitation of fast discharge for AFLMBs is dissected exhaustively, and a symptomatic strategy to break the limit is put forward, in order to eliminate the inevitable mismatch that lies in the inferior performance of AFLMBs. A “volcano-type” curve of capacity retention of AFLMBs is discovered with the discharge rate increased. Systematic investigation revealed that the overlapped spatial relationship between fresh deposited Li and residue Li0 facilitated the utilization of “recoverable Li0” (Li0) at the prophase of discharge rate increase. However, further enhanced discharge rate induced large concentration polarization (η conc), reflecting limited Li+ diffusion. Enabling the electrolyte to rapidly transport Li+ by lowering η conc increased the optimal discharge rate as well as the cycling stability of AFLMBs. This work reveals the rate-determining step for high-rate discharge and expands the employment boundary of AFLMBs under harsh conditions, providing a significant complement of present knowledge with respect to the power performance of AFLMBs.

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D5EE00266D, Review ArticleZaiping Guo, Divyani Gupta, Jinshuo Zou, Jianfeng Mao
Rechargeable metal-CO2 batteries (RMCBs) are highly promising for renewable energy storage and simultaneous reduction of carbon footprint from the environment, making it very attractive for next-generation battery development. An electrolyte...
The content of this RSS Feed (c) The Royal Society of Chemistry

Nature Nanotechnology, Published online: 30 April 2025; doi:10.1038/s41565-025-01896-2

An achromatic metagrating waveguide with a tailored periodic structure designed using a stochastic topology optimization algorithm efficiently guides red, green and blue light at the same angle. This structure provides a compact, lightweight architecture for waveguide-based full-colour augmented reality displays.