Technological progress and sustainable development fundamentally rely on the discovery of advanced materials that enable cleaner energy, smarter devices, resilient infrastructure, and efficient resource use. From energy storage and conversion to electronics, catalysis, and structural applications, new materials are the foundation of transformative innovation. Yet, conventional trial-and-error approaches to materials discovery remain slow, expensive, and increasingly inadequate to address the growing complexity and urgency of today’s scientific and technological challenges. The decades-long development cycles observed in areas such as lithium-ion batteries clearly demonstrate the limitations of traditional methodologies. Across diverse application domains, there is an urgent need to accelerate the discovery of high-performance, cost-effective, non-toxic, earth-abundant, and environmentally sustainable functional materials. In this talk, I will present FULL-MAP, a pioneering European initiative designed to transform the way materials and interfaces are discovered. By integrating laboratory automation, high-throughput experimentation, and artificial intelligence and machine learning–driven multiscale, multiphysics modeling, FULL-MAP establishes a closed-loop, data-driven framework for systematic and adaptive experimentation. This approach accelerates discovery, reduces cost and risk, and demonstrates a new paradigm for materials innovation, one that can be extended beyond batteries to a wide range of advanced material systems critical to future technologies.

Carbon electrodes offer a compelling combination of low cost, chemical stability, and metal-free processing for photovoltaics and printed electronic devices, but realizing their potential requires fabrication methods that can match the pace of modern materials discovery. In this talk, I will present our work on printing as a platform for high throughput fabrication of carbon electrodes, tracing a progression from rapid screening of printed films through conformal printing on complex geometries to fully autonomous roll-to-roll manufacturing.

First, I will show how printing enables high throughput fabrication in its own right. By exploiting the speed and programmability of printing processes, we generate large libraries of carbon electrode films with systematically varied compositions and deposition parameters, allowing structure, processing, and property relationships to be mapped far faster than conventional one-at-a-time fabrication permits.

Second, I will present our development of conformally printed carbon electrodes. Moving beyond flat substrates, we demonstrate printing strategies that deposit uniform, well-adhered carbon films onto curved and textured surfaces, opening a route to electrodes for non-planar device architectures while preserving the throughput advantages of printing.

Finally, I will describe how we have coupled self-driving laboratories with roll-to-roll processing. Closed-loop optimization, combining automated experimentation with machine learning driven experiment selection, is used to navigate the coupled formulation and processing parameter space of continuous coating, and I will share carbon electrode demonstrations in which this approach substantially compressed the optimization timeline while producing manufacturable films.

Together, these results position printing not merely as a scalable deposition method but as an engine for accelerated development, and point towards a workflow in which discovery, optimization, and manufacturing of functional electrodes proceed as a single continuous process.

Autonomous laboratories aim to close the gap between the rates of computational screening and experimental realization of novel materials. In this talk, we present an overview of A-Lab, an autonomous laboratory for the accelerated synthesis of battery materials. A-Lab combines autonomous solid-state synthesis with multimodal characterization techniques, such as X-ray diffraction and energy-dispersive X-ray spectroscopy. We also discuss expanding these capabilities to include non-equilibrium synthesis, enabling tunable heating and cooling rates and opening avenues to investigate synthesis pathways far from equilibrium. Finally, we showcase an example of using this framework to achieve controllable particle size distribution in the one-step solid-state synthesis of single crystal (SC) Li Cathode Active Materials (CAM) with moderate contents of Ni. In particular, we investigate the effect of various Ni and Li precursor combinations on the reaction pathway and optimal reaction and growth temperature boundaries using an in-situ X-ray diffractometer (XRD) with heating capability. In parallel, we investigate the crystallite and particle size distribution and elemental homogeneity using automated multimodal characterizations composed of Scanning Electron Microscopy (SEM) imaging, Energy Dispersive X-ray Spectroscopy (EDS), and ex-situ XRD integrated with the robotic synthesis.

Energy storage is high in demand nowadays, for example to assist in avoiding peak prices of electricity (Kühlkraftkrise) or for sustaining GPU data centers.  Two of the most important battery energy storage technologies today are lithium and sodium ion batteries.  However, the criticality of battery raw materials, e.g. Li, Co,… and their health risks during processing and recycling are a persistent issue.  Therefore, new materials consisting of abundant and (compositionally) non-toxic elements are important.  Though novel materials compositions can be predicted in silico, development of their fabrication processes is largely still done by slow manual experimentation, because predictive synthesis and high throughput screening are comparatively underdeveloped. 

An interesting synthesis method for screening purposes is the aqueous solution-gel route, which is based on citrate complexes as precursors.  Such complexes exist for almost all relevant metal ions in the periodic system, making any thinkable metal oxide in theory easily accessible.  Of course, the precursors need to be compatible: precipitation may readily occur in case of a pH mismatch or a difference in citrate:metal ion ratios for example.  Also, the crystallization of the desired oxide at the end of this synthesis route will require thermal treatment, during which phase segregation of a homogeneous precursor is risked.  Thermodynamics and kinetics of the thermal decomposition and phase formation will be determinant of which crystal phase is actually obtained and more insight into these aspects is required in order to achieve predictive power.  These aspects will be illustrated by our work on various materials such as LiFePO₄, LNMO, Sn substituted Li_1.2Ni_0.13Co_0.13Mn_0.54−xSn_xO₂, LMTO, etc. which span commercially impactful materials up to abundant element containing materials holding high potential for future application.

Battery innovation is constrained not only by the size of the chemical design space, but also by the complex and often poorly connected chain linking material composition, synthesis history, electrode manufacturing, microstructure, interfaces and cell performance. Overcoming this fragmentation requires more than faster individual experiments or simulations: it requires an integrated discovery system capable of reasoning across scales, selecting the next most informative action and translating target performance into experimentally realizable materials and processes.

This contribution presents the emerging FULL-MAP concept for an AI-driven, chemistry-agnostic materials acceleration platform for sustainable batteries. The framework connects high-throughput synthesis and characterization with multiscale and multiphysics modelling, manufacturing simulation, cell-level performance prediction and sustainability assessment. Agentic AI provides the orchestration layer: it decomposes design objectives into linked computational and experimental tasks, selects suitable models and data sources, coordinates autonomous workflows, evaluates uncertainty and provenance, and iteratively proposes the next simulations or experiments. In this way, the platform advances from conventional forward prediction towards multi-objective inverse design—identifying combinations of chemistry, synthesis route, processing conditions and electrode architecture that can deliver required performance, lifetime, manufacturability and sustainability.

A first methodological frontier is the use of machine-learned interatomic potentials (MLIPs) to model synthesis-relevant phenomena at atomistic resolution and at time and length scales inaccessible to conventional first-principles calculations. Beyond predicting equilibrium structures and properties, emerging reactive, charge-aware and chemically transferable MLIPs can address nucleation, phase formation, surface and interphase evolution, defect generation and temperature- or environment-dependent transformations. They also provide physically grounded descriptors and parameter surfaces for coarse-grained mesoscale models, enabling information to propagate from atomistic synthesis mechanisms towards microstructure and electrode behaviour. A second frontier concerns electrode processing. Particle-based and other microstructure-resolved methods can represent slurry constituents, aggregation, drying, binder migration, compaction and particle rearrangement, while continuum manufacturing and electrochemical models connect the resulting structure to transport and performance. Because direct high-fidelity simulation is computationally demanding, AI surrogates and multifidelity learning are required to explore processing windows, quantify trade-offs and enable rapid inverse mapping from a desired electrode structure or performance to feasible manufacturing parameters.

Together, these developments redefine modelling as an active component of autonomous discovery. By coupling novel physics-based and ML-assisted models with agentic decision-making and automated laboratories, FULL-MAP aims to establish closed-loop workflows spanning materials discovery, synthesis, manufacturing and validation—and thereby open new routes for faster, more systematic and more sustainable battery innovation.

Next-generation batteries, such as sodium-ion batteries (SIBs), offer a potentially cheaper and safer alternative to lithium-ion batteries as demand for energy storage continues to increase with the growth of renewable energy. However, current SIB technologies do not yet match the performance of lithium-ion batteries in key metrics such as energy density, cycle life and charging rates. Accelerating the pace of discovery is therefore essential for developing competitive technologies, but traditional research methodologies are often too slow to meet this need.  

Self-driving laboratories combine automation and artificial intelligence to generate reproducible, statistically significant battery datasets that can guide future battery development. Factors such as operator experience, cell assembly procedures and attention to detail can strongly influence cell performance [1]. Although extensive datasets have been developed and utilised by researchers, these systems are typically designed to address highly specific problems, limiting standardisation and reducing the reusability of the resulting data [2, 3].  

In this study, sodium-ion full cells were assembled and evaluated using three commercially available Na3V2(PO4)3 (NVP) cathode materials. These systems were systematically investigated using electrolyte formulations containing fluoroethylene carbonate (FEC) additive concentrations of 0.5, 1.25, 2.5 and 5 wt.%. The resulting dataset demonstrated that the commercial source of NVP had a substantial impact on cell performance. The poorest-performing NVP material showed the greatest improvement with the addition of FEC, whereas the highest-performing cells exhibited no measurable benefit from FEC addition in either rate capability or long-term cycling performance.  

Lithium (Li) metal is widely considered one of the most promising anode candidates for next-generation batteries. However, its commercialization is hindered by interphase instability. The presence of a native passivation layer and the formation of an inhomogeneous, electrochemically derived solid electrolyte interphase (SEI) lead to uneven Li deposition during cycling. This results in dendrites formation and continuous consumption of the Li reservoir, ultimately causing rapid capacity fading.

To address these challenges, we explore vacuum-based approaches to validate machine learning derived nucleation simulations and develop advanced Li metal anodes. First, vacuum deposited metallic interlayers are employed as well-defined platforms to validate machine learning-derived interatomic potential (MLIP) simulations, enabling improved understanding of Li nucleation behavior [1].

In a sperate approach, vacuum thermal evaporation is used to fabricate ultra-pure Li metal anodes [2], followed by the deposition of protective coatings within the same chamber. This process enables precise interface control and the formation of a composite halide artificial SEI. Through a co-evaporation method, controlled doping of a LiF-based artificial SEI is achieved and confirmed by cryogenic transmission electron microscopy (cryo-TEM). When implemented in practical pouch cells with a LFP cathode (2 mAh cm^-2), the coated Li anodes exhibit excellent electrochemical stability, maintaining over 350 cycles at 0.5C and 1C.