巴斯夫

Polyelectrolyte complexation is an entropically driven, associative phase separation that has been leveraged to produce aqueously processed plastics known as polyelectrolyte complexes (PECs). Previously, we showed that their affinity to water and their chain mobility are important aspects to consider when designing PEC materials. To establish a more complete picture of influencing parameters, we examined the effect of polymer chemistry, specifically chain length and the side chain and backbone chemistry, on both the phase behavior and mechanical properties of homopolymer PECs. We combined compositional studies of PEC phase behavior with analyses of PEC dynamics and mechanics to understand how these aspects of polymer chemistry affect material performance. We observed that the identity of the ionizable groups heavily affected ion solvation, where PECs with lower water affinities had higher glass transition humidities and were generally more brittle, compared to PECs with higher water affinities. In contrast, backbone chemistry affected chain mobility, allowing acryloyl chemistries to have lower glass transition humidities compared to methacryloyl. Finally, chain length effects depended on the degree of match/mismatch of the polymer's lengths, with matched PEC systems having higher glass transition humidities than mismatched. Comparisons of the phase behavior and glass transitions revealed that side chain and backbone chemistry effects are universal across different mediums, while length effects are medium specific. These results establish fundamental structure-property relationships for the rational design of functional PEC materials.

Plastic additives are used to strengthen the mechanical properties of polymers, improve processing efficiency, and enhance product durability, thereby enabling their use for diverse applications across many chemical industries. These additives are typically produced in powder form; however, the handling and storage of fine powders in industrial environments present significant challenges, making it necessary to convert them into pellets. The process of pellet formation involves compression of the powder under controlled pressure and temperature. Several models have been developed to explain pelletization in various fields, including biomass, metals, pharmaceuticals, and ceramic production. A common theme of these studies is the use of a continuum approximation for the powder using the Drucker-Prager Cap (DPC) model. These studies relied on an instrumented die to measure the material model parameters. Furthermore, the identification of DPC model cap surface and hardening parameters typically relied on preparing multiple pellets at different target densities, making the parameter calibration process time-consuming and experimentally intensive. In contrast, the present work extracts material parameters from load-displacement data, avoiding extensive pellet preparation without relying on a specialized instrumented die. Additionally, we have presented a modified approach to calculate cap hardening parameters based on global optimization of stress-strain values. Finally, the framework was applied to plastic additive powders and validated against experimental results. An additional sensitivity analysis of the model, based on a full factorial design of experiments (DoE), was performed to evaluate the influence of key parameters on the predicted compaction response. Overall, our findings may help reduce pre-production iterations and improve pellet quality, given the constraints at the factory.

Higher-order graph neural networks (HOGNNs) and the related architectures from Topological Deep Learning are an important class of GNN models that harness polyadic relations between vertices beyond plain edges. They have been used to eliminate issues such as over-smoothing or over-squashing, to significantly enhance the accuracy of GNN predictions, to improve the expressiveness of GNN architectures, and for numerous other goals. A plethora of HOGNN models have been introduced, and they come with diverse neural architectures, and even with different notions of what the "higher-order" means. This richness makes it very challenging to appropriately analyze and compare HOGNN models, and to decide in what scenario to use specific ones. To alleviate this, we first design an in-depth taxonomy and a blueprint for HOGNNs. This facilitates designing models that maximize performance. Then, we use our taxonomy to analyze and compare the available HOGNN models. The outcomes of our analysis are synthesized in a set of insights that help to select the most beneficial GNN model in a given scenario, and a comprehensive list of challenges and opportunities for further research into more powerful HOGNNs.

In August 2023, European Food Safety Authority (EFSA) and European Chemicals Agency (ECHA) published guidance (applicable from April 2026) to assess the impact of drinking water treatment (DWT) processes on residues of plant protection products and biocides (hereafter referred to as EFSA/ECHA guidance). This guidance addresses an important regulatory need by introducing new experimental approaches to evaluate transformations of active substances and their environmental transformation products during drinking water treatment. However, while the regulatory intent of the EFSA/ECHA guidance is clear, its practical implementation presents significant challenges. The experimental concepts are described at a high level, but key testing parameters, operational conditions and analytical considerations are insufficiently specified, and no recognized or validated protocols were available at the time of publication. As a result, implementation of the guidance within such a short timeline risks inconsistent experimental design, poor reproducibility and limited comparability of data across laboratories and studies. To address these challenges, CropLife Europe (CLE) has undertaken a coordinated effort to establish the scientific and technical foundations required to operationalize the EFSA/ECHA guidance. This work focuses on the development of harmonized experimental and analytical frameworks for DWT simulation, including definition of representative treatment conditions, standardized sampling and sample-preparation procedures, and fit-for-purpose analytical workflows. The CLE approach combines targeted quantification with non-target and suspect screening by high-resolution mass spectrometry (HRMS) to detect and elucidate transformation products, enabling reproducible, robust data generation that aligns with the EFSA/ECHA guidance. The overarching objective of these efforts is not to reinterpret the regulatory intent of the EFSA/ECHA guidance, but to enable its consistent and scientifically robust application. By establishing practical testing protocols and best-practice recommendations, the CLE framework aims to improve repeatability and reproducibility of DWT studies across industry, facilitate meaningful comparison of results, and provide a reliable evidence base for regulatory evaluation and decision-making. This policy brief outlines the rationale, methodology and scope of the CLE approach and advocates for its adoption and further validation to support harmonized implementation of the EFSA/ECHA guidance.

Visual counterfactual explainers (VCEs) are a straightforward and promising approach to enhancing the transparency of image classifiers. VCEs complement other types of explanations, such as feature attribution, by revealing the specific data transformations to which a machine learning model responds most strongly. In this paper, we argue that existing VCEs focus too narrowly on optimizing sample quality or change minimality; they fail to consider the more holistic desiderata for an explanation, such as fidelity, understandability, and sufficiency. To address this shortcoming, we explore new mechanisms for counterfactual generation and investigate how they can help fulfill these desiderata. We combine these mechanisms into a novel 'smooth counterfactual explorer' (SCE) algorithm and demonstrate its effectiveness through systematic evaluations on synthetic and real data.