119092 research outputs found
Sort by
ETA Factory – Sustainability Report 2024
No full textThis condensed sustainability report has been prepared in accordance with the Corporate Sustainability Reporting Directive (CSRD) and the European Sustainability Reporting Standards (ESRS). It reflects our activities and impacts for the year 2024, which will be used to benchmark them for subsequent years. The report is voluntary and compiles only the most relevant sustainability aspects
Shining New Light on the Mechanism of Selective Oxidation Catalysis Using Method Development in Transient Spectroscopy
No full textSelective oxidation reactions are some of the most important chemical processes, with enormous economic and environmental contributions. Controlling the selectivity remains the greatest challenge, owing to their complexity, with both parallel and consecutive reaction pathways leading to by-products. Their mode of operation over supported and bulk oxide catalysts has been the subject of debate for decades, largely influenced by phenomenological principles. Recently, direct evidence from transient spectroscopy has provided insight into the dynamical nature of selective oxidation catalysts as well as the actively participating surface species and sites of the catalysts, while highlighting important functions of the supporting structure. This perspective presents the implications of these findings for a scientific understanding of the characteristics of selective oxidation reactions as a basis for rational catalyst design. First, the potential of the transient spectroscopic approach is illustrated based on the available literature on selective oxidation reactions. When moving from supported catalysts to bulk oxide systems with their increased level of structural complexity, additional challenges concerning the determination of structure–performance relationships emerge, but these may be tackled successfully in the future in view of current method development
Methodological approaches for numerical integration of culverts in 2D surface models in urban flood analysis
No full textUrban flood response at road crossings hinges on how short conveyance structures are represented. In our study, each location in the watercourse is not modelled as a bridge per se but alternatively represented as (i) oDEM - a terrain opening that preserves open-channel through-routing; (ii) Culvert - an explicit 1D element transferring flow between inlet and outlet via head-loss formulations; or (iii) Sewer - a short 1D sewer network element (using SWMM) coupled to the surface. As a control, (iv) cDEM enforces overtopping-only exchange to emulate hydraulically ineffective bridge passages. We provide a scenario-controlled, side-by-side benchmark of these four approaches across five sites and three precipitation types (Euler-II 30-year, Euler-II 100-year, and a duration-matched 100-year block event), comparing discharge timing, maximum water depths, flooded area, and end-of-run water balances. Across scenarios, oDEM and Culvert preserve longitudinal connectivity, yielding broader-but-shallower upstream extents and higher boundary outflow, whereas cDEM and Sewer promote upstream storage and compact, deeper ponds. Our contribution is a controlled comparison that clarifies when each representation is fit-for-purpose: for short crossings, the Culvert representation emerges as a practical baseline that captures inlet/outlet losses and pressurised/free-surface transitions, while Sewer may over-retain heads where surface–sewer coupling footprints are restrictive. oDEM and cDEM can be read as bounding models - oDEM as an optimistic high-connectivity envelope, cDEM as a conservative overtopping-only envelope—to bracket plausible responses where calibration data are scarce
The effects of oxygen carrier, CO₂/H₂O ratio, temperature, and CO₂ content in chemical looping reforming of biogas: a parametric experimental study
No full textEgg white assisted synthesis of Fe-Mn spinel oxides: effects of egg white ratio, oxygen partial pressure, and life cycle impacts
No full textEgg white was chosen as a renewable, non-toxic agent for the synthesis of FeMn2O4 spinel pre-catalysts to avoid the use of critical transition metals such as Ni and Co. However, synthesizing phase-pure FeMn2O4 remains challenging due to (i) the requirement of low oxygen partial pressures to counter rapid reoxidation of Mn3O4 in the presence of iron oxides, which can be achieved by the preferred oxidation of the egg white during the calcination, and (ii) the probable formation of Fe3O4 and Mn3O4 during intermediate steps in the reaction, leading to multiphase spinel formation caused by a miscibility gap between the spinels. In contrast, spinels with Ni, Co, Zn, or Al are phase-pure. Egg white has significant environmental impacts in the synthesis of all spinel manganites, as assessed from a life-cycle perspective, which can exceed those of petroleum-based agents such as ethylenediaminetetraacetic acid (EDTA) in most impact categories. Therefore, our results show that the investigated synthesis route is not more sustainable, and we demonstrate that implementing quantitative evaluation of environmental impacts already at an early stage is essential to determine whether a synthesis is truly sustainable
The moderating effects of individual differences in baseline episodic memory on acute exercise benefits in memory
No full textAuditing Language Model Unlearning via Information Decomposition
No full textWe expose a critical limitation in current approaches to machine unlearning in language models: despite the apparent success of unlearning algorithms, information about the forgotten data remains linearly decodable from internal representations. To systematically assess this discrepancy, we introduce an interpretable, information-theoretic framework for auditing unlearning using Partial Information Decomposition (PID). By comparing model representations before and after unlearning, we decompose the mutual information with the forgotten data into distinct components, formalizing the notions of unlearned and residual knowledge. Our analysis reveals that redundant information, shared across both models, constitutes residual knowledge that persists post-unlearning and correlates with susceptibility to known adversarial reconstruction attacks. Leveraging these insights, we propose a representation-based risk score that can guide abstention on sensitive inputs at inference time, providing a practical mechanism to mitigate privacy leakage. Our work introduces a principled, representation-level audit for unlearning, offering theoretical insight and actionable tools for safer deployment of language models
CORE-T: COherent REtrieval of Tables for Text-to-SQL
No full textRealistic text-to-SQL workflows often require joining multiple tables. As a result, accurately retrieving the relevant set of tables becomes a key bottleneck for end-to-end performance. We study an open-book setting where queries must be answered over large, heterogeneous table collections pooled from many sources, without clean scoping signals such as database identifiers. Here, dense retrieval (DR) achieves high recall but returns many distractors, while join-aware alternatives often rely on extra assumptions and/or incur high inference overhead. We propose CORE-T, a scalable, training-free framework that enriches tables with LLM-generated purpose metadata and pre-computes a lightweight table-compatibility cache. At inference time, DR returns top-K candidates; a single LLM call selects a coherent, joinable subset, and a simple additive adjustment step restores strongly compatible tables. Across Bird, Spider, and MMQA, CORE-T improves table-selection F1 by up to 22.7 points while retrieving up to 42% fewer tables, improving multi-table execution accuracy by up to 5.0 points on Bird and 6.9 points on MMQA, and using 4-5x fewer tokens than LLM-intensive baselines
Ad Personalization and Transparency in Mobile Ecosystems: A Comparative Analysis of Google's and Apple's EU App Stores
No full textSmartphones have become the primary interface to the Internet for many users, making app stores an essential part of the mobile ecosystem. Apple's App Store and Google's Play Store form a duopoly of the two largest app stores, both of which offer targeted advertisements in their store ecosystems. Consequently, their need for user data to improve the targeting of ads conflicts with users' desire for privacy. Users have to trust the statements given in privacy policies that are often scattered over multiple places and have no way of overseeing how their data is used for ad targeting. The European Union passed several regulations, most notably the DSA and DMA, addressing this transparency issue. The implementation of these laws, however, must be audited to ensure their effectiveness. Unfortunately, the transparency measures implemented in the context of advertising and the ad-targeting mechanisms in app stores have received little attention so far. In this work, we analyze the first-party ad tracking ecosystem on Apple's and Google's app stores. We measure the effects of different account parameters and interest patterns on the ads these accounts receive. Furthermore, we study the transparency measures implemented by the platforms. While we only detect rare occurrences of targeted advertising, we find Google's recommendations to be highly personalized. We notice multiple issues with the realization of transparency measures that affect their effectiveness and, in our opinion, contradict corresponding EU laws
