arXiv subjects

Rosie Wood

Publications and source records attributed to Rosie Wood.

4 recordsLinked to original sources

arXiv2025

Unravelling 2,4-D -- biochar interactions by molecular dynamics: adsorption modes and surface functionalities

We report a molecular dynamics investigation of 2,4-dichlorophenoxyacetic acid (2,4-D) adsorption at the aqueous-biochar interface using experimentally constrained woody biochar models representative of softwood-derived biochars produced at 400, 600 and 800 $°$C. The models reproduce experimental descriptors (H/C, O/C, aromaticity, true density, and surface functionality) of their experimental counterparts, and simulations enable calculation of adsorption isotherms that align with available experimental measurements. Our results reveal that 2,4-D$^{-}$ uptake is governed by a synergy of three interaction classes: (i) $π$-$π$ and $π$-Cl contacts with graphitic domains with either parallel or perpendicular alignments, (ii) polar interactions including H-bonding to surface -OH and other oxygen-containing groups, and (iii) Na$^{+}$-mediated cation bridging that links 2,4-D$^{-}$ anion to surface oxygens, that would have an increasing relevance for biochars near or above the pH at point of zero charge. Notably, we found that low-temperature produced biochars, which retain higher densities of surface O functionalities, exhibit higher adsorption per unit surface area due to cooperative polar interactions alongside $π$-$π$ binding, whereas medium-to-high temperature biochars rely more on $π$-$π$ and cation-bridging mechanisms. The distinct adsorption distances measured emphasize surface heterogeneity and porosity. Taken together, these atomistic insights corroborate experimental observations and yield actionable guidance for the rational design of biochars for remediation of anionic herbicides, highlighting how surface functionality and solution chemistry can be tuned to optimize sorption. Our approach provides a general framework to interrogate pollutant-biochar interactions and to inform remediation strategies.

cond-mat.mtrl-sci↗

arXiv2025

Retrieval-augmented reasoning with lean language models

This technical report details a novel approach to combining reasoning and retrieval augmented generation (RAG) within a single, lean language model architecture. While existing RAG systems typically rely on large-scale models and external APIs, our work addresses the increasing demand for performant and privacy-preserving solutions deployable in resource-constrained or secure environments. Building on recent developments in test-time scaling and small-scale reasoning models, we develop a retrieval augmented conversational agent capable of interpreting complex, domain-specific queries using a lightweight backbone model. Our system integrates a dense retriever with fine-tuned Qwen2.5-Instruct models, using synthetic query generation and reasoning traces derived from frontier models (e.g., DeepSeek-R1) over a curated corpus, in this case, the NHS A-to-Z condition pages. We explore the impact of summarisation-based document compression, synthetic data design, and reasoning-aware fine-tuning on model performance. Evaluation against both non-reasoning and general-purpose lean models demonstrates that our domain-specific fine-tuning approach yields substantial gains in answer accuracy and consistency, approaching frontier-level performance while remaining feasible for local deployment. All implementation details and code are publicly released to support reproducibility and adaptation across domains.

cs.CL↗

arXiv2023

Biochars at the molecular level. Part 2 -- Development of realistic molecular models of biochars

Biochars have been attracting renewed attention as economical and environmentally friendly carbon sequestration materials with a diverse range of applications. However, experimental developments may be limited by the lack of molecular-level knowledge of the key interactions driving these applications. Molecular modelling techniques, such as molecular dynamics simulations, offer a systematic and reproducible alternative and yield atomistic insights into physicochemical processes, allowing the identification of adsorption mechanisms and, through this, informing and guiding experimental development. In this work, on the basis of the critical assessment of the analytical techniques for characterisation of biochars and collation of a large volume of experimental data, we develop molecular models of three woody biochar materials, representative of those produced under low-, medium-, and high-temperature treatments. We characterise these models, validating them against experimental data, and share them with the research community. Furthermore, we detail our iterative approach to the design of these biochar models, discuss what we have learned about the relationship between biochar composition and its morphology, and finally share all of the building blocks used to create these biochar models. With this work, we hope to speed up the uptake of molecular dynamics simulations for the study and development of biochar materials and, to this end, we distribute our easy-to-use surface-exposed biochar models ready for the adsorption studies.

cond-mat.mtrl-sci↗

arXiv2023

Biochars at the molecular level. Part 1 -- Insights into the molecular structures within biochars

Biochars are black carbonaceous solids produced through biomass pyrolysis under conditions of little or no oxygen. Whilst their properties are well studied, and their applications numerous, the underlying molecular structures within biochars still need to be defined. This raises a substantial barrier to the molecular modelling of biochars and has limited computational study of these materials, despite the advantages of such techniques. In this work, we critically assess the analytical techniques used to characterise biochars and use this information to gain molecular-level insights into biochars' molecular compositions and nanostructures. We focus on properties present at the nanoscale and which provide atomic-resolution insights into the molecular structures within these materials. Our goal is to create a holistic understanding of biochars' chemical, physical and molecular properties and to lay the foundation for future work focused on developing realistic molecular models of these materials.

cond-mat.mtrl-sci↗