12 Search Results

Large language models (LLMs) show remarkable potential for scientific reasoning but often produce unreliable or scientifically unactionable outputs when faced with multi-step logic, domain grounding, and interpretability challenges, especially in complex fields like chemistry and materials science. Here, we introduce a framework of philosophical reasoning agents, inspired by canonical thinkers such as Socrates, Descartes, Kant, and Hume, to guide LLM behavior via structured prompt engineering. These agents embody distinct reasoning paradigms (dialectical inquiry, deductive logic, rule-based judgment, and empirical validation) and are evaluated across multiple chemistry subdomains, physical, analytical, general, inorganic, and organic chemistry, using the ChemBench benchmark. Our agenticmore » prompting approach yields substantial accuracy gains on open-ended numerical chemistry questions, with gains of +11.5 percentage points for GPT-4o with Hume, +4.5 percentage points for GPT-5 with Kant, and +21.8 percentage points for GPT-5.1 with Socrates at the strict 1% error threshold, relative to the corresponding base models. Beyond accuracy, we observe benchmark-level model–agent performance patterns, suggesting that different prompting styles interact differently with each base model. These findings demonstrate that embedding philosophy-of-science principles into multi-agent frameworks can improve and produce interpretable, adaptive, and domain-aligned scientific LLMs.« less

Accelerating the discovery of new molecules with targeted properties is a central challenge in molecular design. In this contribution, we present an AI-driven molecular discovery framework that integrates Large Language Models (LLMs) for generative molecular design with Machine Learning (ML)-based screening to identify novel Liquid Organic Hydrogen Carrier (LOHC) candidates. Using the developed framework, LOHC molecules were systematically generated, evaluated, and refined iteratively, combining LLM-guided molecular generation and ML-predicted hydrogenation enthalpies (ΔH), under physicochemical property constraints such as optimal melting points (MP), desired hydrogen storage capacity (wt% H₂), and synthetic accessibility (SA) scores. This approach enabled the discovery of 42more » new LOHC candidates in two distinct campaigns, one seeded with experimentally known and another with previously computationally identified LOHCs, respectively. Although we began with different numbers of starting molecules (31 vs. 7 seed molecules), both runs yielded a comparable number of viable candidates, suggesting an influence of chemically intuitive seed molecule selection for success. Selected LOHC molecules, such as 3-methyl pyridine, 1-ethylnapthalene, 1,1-diphenylethane, and benzofuran, were experimentally tested and compared with benchmark LOHCs (toluene and 9-ethylcarbazole) for hydrogenation using a series of commercial supported metal catalysts. The order of conversion into fully hydrogenated products at 200 °C was 3-methyl pyridine (100%) > 9-ethyl carbazole (86.4%) > 2,3-benzofuran (74%) > 1,1-diphenylethane (66.9%) > 1-ethylnapthalene (66.7%) > toluene (57%), further validating the AI-guided molecular design. This study demonstrates promise of LLM-driven molecular design in conjunction with ML-based screening for accelerated discovery and design of molecules.« less

As large language models (LLMs) become central tools in science, improving their reasoning capabilities is critical for meaningful and trustworthy applications. We introduce a Socratic agent for scientific reasoning, implemented through a structured system prompt that guides LLMs via classical principles of inquiry. Unlike typical prompt engineering or retrieval-based methods, our approach leverages definition, analogy, hypothesis elimination, and other Socratic techniques to generate more coherent, critical, and domain-aware responses. We evaluate the agent across diverse scientific domains and benchmark it on the abstraction and reasoning corpus challenge dataset, achieving 97.15% under a fixed prompting protocol and without fine-tuning or externalmore » tools. Expert evaluation shows improved reasoning depth, clarity, and adaptability over conventional LLM outputs, suggesting that structured prompting rooted in philosophical reasoning can improve the scientific utility of language models.« less

The development of novel energy materials and fuels is required to expand current available energy sources. Aiming to reach this goal, there is growing interest in using molecular hydrogen as an energy carrier due to its abundance and high energy density. Liquid organic hydrogen carriers (LOHCs) are a promising route to the large-scale storage and transport of hydrogen for use in the energy economy. The search for thermodynamically viable LOHC molecules for real world use has led to a set of constraints on the dehydrogenation enthalpy and the minimum gravimetric hydrogen capacity. These constraints allow one to formulate the searchmore » for an ideal LOHC candidate molecule as an optimization problem well suited to the strengths of machine learning and artificial intelligence computational approaches. A critical barrier to a large-scale, high-throughput screening of LOHC candidate molecules is the lack of reliable training data. Computational electronic structure methods including density functional theory, coupled cluster approximations, and diffusion Monte Carlo can be used to provide training data where experimental data are either unreliable or do not exist. In this work, we use these methods to calculate the dehydrogenation energies and enthalpies of candidate LOHC molecules.« less

The development of efficient methods for metal ion recovery and water remediation is critical for addressing the nation’s urgent need for secure domestic supply chains, overcoming critical materials challenges, and ensuring resilient manufacturing. This work presents a novel approach for synthesizing stable Alizarin (Alz, 1,2-dihydroxy anthraquinone) multilayers on glassy carbon electrodes (GCEs) for electrochemical sequestration of manganese, sodium, and lithium cations. Alz-GCEs exhibit negatively shifted reduction potentials, indicating strong interactions between Alz and metal cations through metal-coupled electron transfer (MCET) mechanisms. The cation binding interactions and redox behavior of these electrodes were investigated using cyclic voltammetry (CV) and density functionalmore » theory (DFT) calculations. Our results demonstrate that Alz forms multilayer structures on GCEs with redox properties that are modulated by the presence of metal cations in the electrolyte. DFT calculations provide insights into the electrochemical mechanism, indicating both stepwise and concerted pathways for metal binding. The findings highlight the potential of Alz-GCEs for efficient and selective metal ion capture, which could be useful for developing sustainable materials for critical metal recovery and water remediation. This work suggests that redox-mediated organic adlayers are promising candidates for advancing electrochemical separation technologies of metal ions.« less

Electrochemical transformations of metal ions in aqueous media are challenging to model accurately due to the dynamic solvation structure surrounding ions at different charge states. Predictive modeling at the atomistic scale is essential for understanding these solvation architectures but is often computationally prohibitive. In this contribution, we present a simple, fast, and accurate three-layer micro-solvation model to evaluate the redox potential of metal ions in aqueous solutions. Our model, developed and validated for Fe³⁺/Fe²⁺ redox potentials, combines the DFT-based geometry optimizations of the octahedral Fe complex with two layers of explicit water molecules to capture solute–solvent interactions and an implicitmore » solvation model to account for bulk solvent effects. This approach yields accurate predictions for Fe³⁺/Fe²⁺ redox potentials in water, achieving errors of 0.02 V with ωB97X-V, 0.01 V with ωB97X-D3, 0.04 V with ωB97M-V, and 0.02 V with B3LYP-D3 functionals. We further demonstrate the generality of our model by applying it to additional metal complexes, including the challenging Fe(CN)₆^3−/4− system, where our model successfully achieves close agreement with experimental values, with an error of 0.07 V and an average error of 0.21 V for all five systems. In summary, the presented simple solvation model has broad applicability and potential for enhancing computational efficiency in redox potential predictions across various chemical and industrial processes of metal ions.« less

This contribution presents a comprehensive extension of the QM9 dataset (originally at 133 K molecules) with the calculation of G4MP2 enthalpies for 9,841 molecules, featuring up to nine heavy atoms. We present QM9-LOHC, a (de)hydrogenation dataset of 10,373 reactions, including a minimum of 5.5% weight hydrogen storage capacity in line with the Department of Energy standards for Liquid Organic Hydrogen Carriers (LOHC). By utilizing the accurate quantum chemical method G4MP2 we expand the QM9 database and explore new avenues for the exploration of hydrogen storage technologies (electrochemical LOHCs, alkali metal-LOHCs, and mixtures of LOHCs). The QM9-LOHC dataset, with its focusmore » on reactions that vary only by hydrogen saturation levels, provides a needed data resource for advancing the design and optimization of both conventional and innovative LOHC systems, and high-fidelity data for molecular discovery.« less

Molecular electronic devices require precise control over the flow of current in single molecules. However, the electron transport properties of single molecules critically depend on dynamic molecular conformations in nanoscale junctions. Here, in this work, we report a unique strategy for controlling molecular conductance using shape-persistent molecules. Chemically diverse, charged ladder molecules, synthesized via a one-pot multicomponent ladderization strategy, show a molecular conductance (d[log(G/G₀)]/dx ≈ -0.1 nm^-1) that is nearly independent of junction displacement, in stark contrast to the nanogap-dependent conductance (d[log(G/G₀)]/dx ≈ -7 nm^-1) observed for non-ladder analogues. Ladder molecules show an unusually narrow distribution of molecular conductance duringmore » dynamic junction displacement, which is attributed to the shape-persistent backbone and restricted rotation of terminal anchor groups. These principles are further extended to a butterfly-like molecule, thereby demonstrating the strategy's generality for achieving gap-independent conductance. Overall, our work provides important avenues for controlling molecular conductance using shape-persistent molecules. Achieving robust and controllable conductance in single-molecule junctions is challenging due to the dynamic nature of molecular conformations that fluctuate over operational timescales. A strategy using shape-persistent molecules has now been developed that demonstrates nearly junction-displacement-independent conductance, providing a stable solution for single-molecule electronic properties.« less

Predicting the synthesizability of a new molecule remains an unsolved challenge that chemists have long tackled with heuristic approaches. Here, in this study, we report a new method for predicting synthesizability using a simple yet accurate thermochemical descriptor. We introduce E_min, the energy difference between a molecule and its lowest energy constitutional isomer, as a synthesizability predictor that is accurate, physically meaningful, and first-principles based. We apply E_min to 134,000 molecules in the QM9 data set and find that E_min is accurate when used alone and reduces incorrect predictions of "synthesizable" by up to 52% when used to augment commonlymore » used prediction methods. Our work illustrates how first-principles thermochemistry and heuristic approximations for molecular stability are complementary, opening a new direction for synthesizability prediction methods.« less

We present a comprehensive, in silico-based discovery approach to identifying novel liquid organic hydrogen carrier (LOHC) candidates using cheminformatics methods and quantum chemical calculations. We screened over 160 billion molecules from ZINC15 and GDB-17 chemical databases for structural similarity to known LOHCs and employed a data-driven selection criterion connecting molecular features with dehydrogenation enthalpy. This scoring criterion effectively predicts dehydrogenation enthalpies from SMILES strings, streamlining the LOHC screening process. After rigorous screening and down-selection, we compiled a database of 3000 dehydrogenation reactions for the most promising LOHC candidates, setting the stage for future selection based on kinetics and catalysis. Thismore » work demonstrates the significant impact of integrating quantum chemistry and cheminformatics in materials discovery, accelerating the selection process while reducing experimental efforts and time. By proposing new molecules as prospective LOHC candidates, our study provides a valuable resource for researchers and engineers in the development of advanced LOHC systems and showcases a successful approach for high-throughput discovery, contributing to more efficient and sustainable energy storage solutions.« less