Samuel John Cooper

Battery research increasingly relies on advanced imaging, yet open access to such data remains rare, scattered across various sources, and difficult to find.

Scientific optimization problems are often characterized by slow, expensive, and non‐differentiable evaluations, which severely limit efficient exploration of rich design spaces. In these settings, performance depends not only on the optimizer itself, but also on whether the chosen search space can represent good solutions and whether important design requirements are effectively incorporated into the search. Here, we introduce a gradient‐free optimization framework based on large language models (LLMs) that enables the generation and refinement of executable control policies directly from natural language. We propose two complementary paradigms: Prompt‐to‐Optimizer (P2O), which combines LLM‐based structural generation with gradient‐free parameter optimization, and Prompt‐to‐Protocol (P2P), which directly synthesizes fully specified control policies. We apply this framework to battery fast‐charging protocol design, a representative slow‐feedback optimization problem. Across multiple case studies, our methods outperform conventional approaches restricted to predefined functional forms. In a realistic fast‐charging scenario, both approaches improve battery state of health by approximately 4.2% over a representative baseline, with P2P achieving this under matched evaluation budgets. More broadly, our results suggest that LLMs can serve as representation generators that expand the hypothesis space of scientific optimization, enabling efficient search in domains where evaluations are costly, and gradients are unavailable.

Ragone plots provide a compact representation of energy storage performance by relating accessible energy to power. Despite this, battery degradation is often described using simplified metrics such as “energy fade” and “power fade”, which obscure changes in the overall performance envelope. Here, we propose a shift towards a Ragone-based analysis of degradation, enabling a more comprehensive assessment of battery health. We use physics-based models to simulate ageing under a range of operating conditions and degradation modes, and evaluate Ragone plots over the course of ageing. Each degradation mechanism leaves a distinct signature in the Ragone plots, which can be directly associated with energy and power fade, as well as with specific internal states of the battery. We also show that the widely reported “knee” in reference performance tests is an artefact of testing protocols, rather than an intrinsic feature of battery behaviour, underscoring the advantages of Ragone-based characterisation. Finally, we introduce a systematic empirical method to extract energy and power fade from Ragone plots. These results demonstrate that Ragone analysis provides a more powerful framework for diagnosing, interpreting, and ultimately mitigating battery degradation in practical applications.

Microstructure design of battery electrodes is key to enable fast charge and higher capacity in next-generation lithium and sodium ion batteries. While porous-electrode (Doyle–Fuller–Newman-type) models capture the coupled transport–reaction physics governing cell performance, they typically treat the electrode as a homogenized medium and encode microstructure only through fitted, scalar “effective” parameters (e.g., Bruggeman-type tortuosity). This assumption is often adequate for uniform electrodes, but it breaks down for modern architectures with graded porosity, binder migration, bimodal particle distributions, or bilayer designs, where transport resistance is heterogeneous and reaction rates localize, leading to non-uniform utilization and premature rate limitations.

Ragone plots provide a compact representation of energy storage performance by relating accessible energy to power. Despite this, battery degradation is often described using simplified metrics such as “energy fade” and “power fade”, which obscure changes in the overall performance envelope. Here, we propose a shift towards a Ragone-based analysis of degradation, enabling a more comprehensive assessment of battery health. We use physics-based models to simulate ageing under a range of operating conditions and degradation modes, and evaluate Ragone plots over the course of ageing. Each degradation mechanism leaves a distinct signature in the Ragone plots, which can be directly associated with energy and power fade, as well as with specific internal states of the battery. We also show that the widely reported “knee” is an artefact of testing protocols, rather than an intrinsic feature of battery behaviour, underscoring the advantages of Ragone-based characterisation. Finally, we introduce a systematic empirical method to extract energy and power fade from Ragone plots. These results demonstrate that Ragone analysis provides a more powerful framework for diagnosing, interpreting, and ultimately mitigating battery degradation in practical applications.

Establishing optimised lithium-ion battery manufacturing processes is a multi-factored, application dependent, and typically resource-intensive challenge, involving the assembly and testing of large numbers of cells over a long period of time. This effectively slows the development and adoption of new technologies. In this work, we adopt and evaluate an accelerated artificial intelligence (AI) assisted method for optimisation of cell manufacturing processes, utilising an LiMn _x Fe _1-x PO ₄ (LMFP) cathode as a model system. Electrode microstructure (controlled via the calendaring process) was optimised for 4680 cylindrical cell performance via AI analysis of microscopy images and electrochemical data from only a relatively small number of manufactured coin cells, dramatically reducing needed development time and cost. Through this analysis, an optimised electrode density for high power performance (2.8 g cm ^-3 ) was identified and experimentally verified. The optimised density, balancing electronic and ionic transport, provided enhanced volumetric energy density under fast-charging conditions (1 C – 3 C, geared for electric vehicles). This study quantitatively outlines the role of LMFP electrode density on performance and highlights the growing relevance of AI-based approaches as an effective tool in advancing lithium-ion battery research and development.

The features of self‐supervised vision transformers (ViTs) contain strong semantic and positional information relevant to downstream tasks like object localization and segmentation. Recent works combine these features with traditional methods like clustering, graph partitioning or region correlations to achieve impressive baselines without finetuning or training additional networks. Upsampled features are leveraged from ViT networks (e.g., DINOv2) in two workflows: in a clustering‐based approach for object localization and segmentation and paired with standard classifiers in weakly supervised materials segmentation. Both show strong performance on benchmarks, especially in weakly supervised segmentation where the ViT features capture complex relationships inaccessible to classical approaches. It is expected that the flexibility and generalizability of these features will both speed up and strengthen materials characterization, from segmentation to property‐prediction.

Efficiently optimizing battery charging protocols is challenging because each evaluation is slow, costly, and non-differentiable. Many existing approaches address this difficulty by heavily constraining the protocol search space, which limits the diversity of protocols that can be explored, preventing the discovery of higher-performing solutions. We introduce two gradient-free, LLM-driven closed-loop methods: Prompt-to-Optimizer (P2O), which uses an LLM to propose the code for small neural-network-based protocols, which are then trained by an inner loop, and Prompt-to-Protocol (P2P), which simply writes an explicit function for the current and its scalar parameters. Across our case studies, LLM-guided P2O outperforms neural networks designed by Bayesian optimization, evolutionary algorithms, and random search. In a realistic fast charging scenario, both P2O and P2P yield around a 4.2 percent improvement in state of health (capacity retention based health metric under fast charging cycling) over a state-of-the-art multi-step constant current (CC) baseline, with P2P achieving this under matched evaluation budgets (same number of protocol evaluations). These results demonstrate that LLMs can expand the space of protocol functional forms, incorporate language-based constraints, and enable efficient optimization in high cost experimental settings.

Ragone plots (a) offer a superior representation of the degradation state of a battery, clearly showing energy and power fade, which is not captured in standard reference performance tests (b), which only probe the battery at a single power.