London Centre for Nanotechnology

Abstract Loneliness in later life is a public‑health concern, and large language model (LLM)-enabled conversational robots have potential to supplement social connection at scale by facilitating everyday conversations. However, evidence on the development and acceptability of such systems for older people remains scarce. We conducted a two‑round mixed methods study with older community‑dwelling participants, to test whether pipeline refinements—streaming automatic speech recognition, text‑to‑speech, and simplified turn‑taking cues—improved conversational experience. A tabletop conversational robot (Sota, Vstone), powered by Generative Pre‑trained Transformer (GPT)‑4o was used. Eighteen individuals participated in round one; paired analyses were conducted on the 17 eligible participants remaining in round two. We collected interaction logs - (response latency, LLM‑level latency, and deletion rate (proportion of words in the manual transcript omitted in the ASR output), technical failure events, and participant questionnaires (Godspeed Questionnaire Series, GQS; System Usability Scale, SUS). Qualitative data were analysed using a framework approach. Median response latency decreased from 9.8 to 4.7s, deletion rate from 0.37 to 0.07, and usability (SUS) improved from 56.2 to 66.9. GQS subscales changed little and perceived companionship remained only modest. Qualitative accounts converged: interactions felt smoother and more in‑context, yet interruptions, perceived superficiality of conversation, and only a modest sense of companionship was still experienced. Findings indicate that improving basic conversational fluency enhances naturalness and usability but is insufficient for delivering truly relational experience outcomes. Progress toward better companionship requires an additional relational layer—shared memories, age‑attuned personas, and calibrated reciprocity.

BackgroundParticipatory systems mapping (PSM) methods are increasingly applied in population health research to understand and address complex challenges. Despite their growing use, there remains limited understanding of how these approaches are implemented in practice. This systematic scoping review aimed to explore the application of PSM in population health research, identify methodological gaps and highlight opportunities for advancing methods development and reporting standards, with particular attention to participatory approaches.MethodsA systematic search of OVID MEDLINE and Scopus identified peer-reviewed papers published in English between January 2000 and September 2023 that: (1) applied and presented the results of PSM related to population health or health improvement questions and (2) incorporated a participatory design. Two reviewers screened and assessed papers, extracting data on study characteristics, participatory approaches, map features and integration of conceptual frameworks and methods not directly related to PSM.ResultsIn the 123 included studies, involving stakeholders in building causal loop diagrams was the most commonly used approach. Variability was evident in geographical focus, study design, application and reporting. Participant involvement was mostly limited to map building, with less engagement in map validation. Significant gaps in reporting study samples and procedures were identified. A small number of studies involved end users or people with lived experiences in mapping processes. Only a few studies evaluated stakeholders’ experience with participatory processes. Lessons learnt on participatory processes include: PSM in population health benefits from cross-disciplinary, inclusive collaboration and capacity-building efforts that support meaningful involvement, shared ownership and trust among diverse stakeholders. Adaptability in the design of PSM approaches, continuous reflection and long-term partnerships are essential to maintaining relevance, enhancing impact and fostering systemic change over time.ConclusionsTo advance participatory systems mapping in population health, there is a need for further methodological innovation, stronger stakeholder engagement and more transparent, reflexive reporting practices. Building capacity through training, practical guidance and cross-disciplinary communities of practice will also be essential to support rigorous and inclusive application of these methods.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12961-026-01457-6.

Abstract Using a surface force balance, we measure the force between two crossed-cylinder electrodes for different electrolyte systems and different geometrical configurations under the influence of an AC field. When an AC field is applied to concentrated mixtures of an ionic liquid (IL) in a polar solvent, a force arises that is at least an order of magnitude larger and slower to reach steady state than the static response of the electrolyte. We demonstrate that this AC field effect persists in concentrated electrolytes beyond what has already been observed in pure ILs or aqueous electrolytes. Furthermore, we show that the magnitude of the electric field force response is dependent on the electrode radius of curvature. In experiments with pure ILs, a stronger force magnitude is observed for a smaller radius of curvature, implying that these interactions can be fine-tuned through geometrical design. These observations should guide the application of AC electric fields for the control of colloids and the design of passive control mechanisms in microfluidics and nanotribology.

تستكشف هذه الدراسة حلاً مستداماً لمعالجة عدم التطابق الومني بين العرض والطلب على الكهرباء في منطقة براك الشاطئ من خلال استغلال فائض الكهرباء خلال ساعات انخفاض الطلب لإنتاج الهيدروجين عبر التحليل الكهربائي، ثم استخدام خلايا الوقود لتوليد الكهرباء خلال فترات الذروة. تقيم الدراسة جدوى استخدام نظام يعتمد على الهيدروجين لتحقيق استقرار الشبكة الفرعية دون الحاجة إلى وحدات توليد كهرباء إضافية. ووفقاً للنتائج، يمكن للمحلل الكهربائي إنتاج ما يصل إلى 271.8 متر مكعب من الهيدروجين في الساعة، ويتطلب ذلك حوالي 9.1 ميغاواط من الكهرباء، وهو ما يكفي لتغطية العجز في الطاقة البالغ 6.1 ميغاواط في الشبكة. وبسعة تخزين للهيدروجين تبلغ 421,255 متر مكعب، يمكن لهذا النظام تخزين الهيدروجين لاستخدامه لاحقاً في خلايا الوقود لتغطية العجز البالغ 1689 ميغاواط في الشبكة الفرعية خلال أوقات الذروة. وتشير النتائج إلى أنه من خلال تحسين إنتاج الهيدروجين بالتحليل الكهربائي واستخدام خلايا الوقود لتوليد الكهرباء، يمكن للنظام المقترح توفير 8,424 ميغاواط من الطاقة، مما يغطي فعلياً عجز الشبكة. ولا يقتصر هذا الأسلوب على سد الفجوة بين العرض والطلب فحسب، بل يثبت أيضاً أنه حل فعال من حيث التكلفة والكفاءة لتحقيق استقرار الشبكة دون الحاجة إلى محطات توليد كهرباء إضافية.

تقدم هذه الورقة تحليلاً لنظام طاقة شمسية عائم بقدرة 600 ميجاوات مقترح تركيبه في الخمس، ليبيا. تمت محاكاة النظام باستخدام برنامج PVsyst الإصدار 7.4 مع طريقة التحويل القائمة على طريقة HDKR باستخدام بيانات الطقس من Meteonorm الإصدار 8.1 ومصادر بيانات الطقس Solargis. تظهر نتائج المحاكاة أن النظام سيحقق إنتاجًا سنويًا للطاقة يبلغ 1,346 جيجاوات ساعة/سنة، ونسبة أداء للنظام تبلغ 81.7٪، وإنتاجًا محددًا للطاقة يتراوح بين 1,850 و 2,050 كيلووات ساعة/كيلوواط/سنة مع عامل قدرة يبلغ 25.6٪. سيحقق نظام FPV مكاسب إشعاع خلفي ثنائي الوجه تتراوح بين 3.0 و4.5٪ وسيوفر 1.25 مليون متر مكعب من المياه السطحية سنويًا، ما يعادل قيمة نقدية تتراوح بين 250,000 و625,000 يورو سنويًا. يبلغ إجمالي الاستثمار المطلوب للمشروع ما بين 480 و540 مليون يورو، أي ما يعادل 0.80 إلى 0.90 يورو لكل واط ذروة من التيار المباشر، وينتج عنه تكلفة طاقة متوسطة (LCOE) تتراوح بين 0.055 و 0.065 يورو لكل كيلوواط/ساعة على مدى 25 عامًا ومعدل خصم 5٪. وهذا يتماشى مع LCOE لمحطات الطاقة التقليدية في ليبيا، والتي تتراوح بين 0.07 و 0.10 يورو لكل كيلوواط/ساعة. عند إضافة التكلفة الكربون (SC_CO2 ~70 دولارًا أمريكيًا/طن CO2)، تتراوح تكلفة الطاقة المعدلة بيئيًا بين 0.028 و0.035 يورو/كيلوواط ساعة. تثبت النتائج التي تم الحصول عليها جدوى أنظمة الطاقة الكهروضوئية العائمة (FPV) على نطاق المرافق العامة في ليبيا.

Heterostructure engineering provides a versatile route for tailoring emergent functionalities that are often difficult to realize in single-phase materials. In this work, the focus is on superlattices composed of the insulating and ferromagnetic double perovskites La2NiMnO6 and Sm2NiMnO6, which constitute an intriguing model system. These layered structures are predicted to feature unequal antipolar displacements of the La and Sm ions; when combined with odd periodicity stacking, this structural motif is expected to give rise to polar behavior. The respective superlattices are grown with atomic precision and display robust ferromagnetism, as confirmed by in-house magnetometry and synchrotron measurements. Scanning transmission electron microscopy combined with first-principles calculations confirms the presence of the predicted unequal antipolar displacements, paving the way for the realization of hybrid improper ferroelectricity in such oxideheterostructures.

Abstract Training machine learning interatomic potentials (MLIPs) on total energies of molecular clusters using differential or transfer learning is becoming a popular route to extend the accuracy of correlated wave-function theory to condensed phases. A key challenge, however, lies in validation, as reference observables in finite-temperature ensembles are not available at the reference level. Here, we construct synthetic reference data from pretrained MLIPs and evaluate the generalizability of cluster-trained models on ice-Ih, considering scenarios where both energies and forces and where only energies are available for training. We study the accuracy and data-efficiency of differential, single-fidelity transfer, and multi-fidelity transfer learning against ground-truth thermodynamic observables. We find that transferring accuracy from clusters to bulk requires regularization, which is best achieved through multi-fidelity transfer learning when training on both energies and forces. By contrast, training only on energies introduces artefacts: stable trajectories and low energy errors conceal large force errors, leading to inaccurate microscopic observables. More broadly, we show that accurate reproduction of microscopic structure correlates strongly with low force errors but only weakly with energy errors, whereas global properties such as energies and densities correlate with low energy errors. This
highlights the need to incorporate forces during training or to apply careful validation before production. Our results highlight the promise and pitfalls of cluster-trained MLIPs for condensed phases and provide guidelines for developing - and critically, validating - robust and data-efficient MLIPs.

ABSTRACTBody movements like walking can synchronize with auditory and visual inputs presented within a periodic frequency range, peaking around 2 Hz. Some evidence has shown that the spontaneous tempo of human locomotion is around 2 Hz. The EEG frequency‐tagging approach allows us to capture the coupling of beat perception with neural brain oscillations at beat frequency. This study used EEG frequency tagging to explore brain dynamics during the perception of walking‐related sensory information in the auditory (footstep sounds) and visual (point‐light figure) modalities. Sensory inputs were delivered at different rates (1, 2, and 3.6 Hz) in rhythmic or random sequences while recording EEG activity. The experiment included three conditions: (i) auditory, (ii) visual, and (iii) audiovisual, including data from 22 participants. Results showed a main effect of rhythmic sequences compared with random sequences across all frequencies in all three auditory, visual, or audiovisual conditions. Specifically, at 2 Hz, rhythmic sequences enhanced neural entrainment in the sensorimotor cortex for auditory and audiovisual conditions. This effect was absent in the visual condition alone. Notably, 2 Hz rhythmic sequences in the audiovisual condition led to coupling with temporal, sensorimotor, and occipital regions. The study suggests that sensory auditory input related to walking movement presented at 2 Hz can mediate neural entrainment with sensorimotor areas. The findings of this study can have an impact on the spontaneous rhythmic integration of body movements using sensory inputs for walking rehabilitation.

This work presents a single-step method for producing cost-efficient copper-modified zinc oxide photoanodes through scalable chemical vapor deposition. The role of Cu incorporation is thoroughly investigated, with the identification of an optimized loading of the metal in these films. The optimally Cu-modified ZnO sample (CZO-5.6) achieved a stable photocurrent of approximately 1.22 mA cm-2 at 1.23 VRHE, along with a Faradaic efficiency of 89%. This enhanced performance was attributed to surface plasmon resonance (SPR) effects induced by copper nanoparticles, as evidenced by photoluminescence spectroscopy results. To promote stability under the experimental conditions of the PEC cell, the best-performing photoanode was further protected using amorphous TiO2 deposited by atomic layer deposition. Amorphous TiO2 coatings have been found to be exceptionally stable in alkaline solutions and highly conductive for photogenerated holes, offering a promising solution for PEC electrode protection. This work not only describes a method for fabricating photoanodes with high photocatalytic activity but also suggests a low-cost route toward the development of photocatalysts for hydrogen production.

Manipulation of excitonic emission properties is important for numerous photonic applications. Of particular interest are developing easy-to-implement yet effective approaches for controlling the radiation dynamics and directionality of spin-forbidden dark excitons (XD) in two-dimensional semiconductors. Here, we investigate the spectral, temporal, and directional characteristics of room-temperature XD emission from a tungsten diselenide monolayer coupled to a dissipative plasmonic nanocavity. Under resonant plasmon-exciton coupling, the radiative decay rate of XD is accelerated by nearly four orders of magnitude, and correspondingly, the XD lifetime is shortened to a subnanosecond level, making it comparable to that of bright excitons. Fitting the measured lifetimes with a Purcell-formalism–based cavity quantum electrodynamics model allows estimating of the intrinsic room-temperature XD lifetime to be about 24 ± 2.3 microseconds. Furthermore, the measured radiation patterns of the dark excitons show that subtle variations in the nanocavity orientation can effectively tailor the XD emission directionality, important for quantum technologies and optoelectronics applications.