Periodic Reporting for period 1 - DECODE (DE-centralised Cloud labs fOr inDustrialisation of Energy materials)
Reporting period: 2023-12-01 to 2025-05-31
The clean energy technology sector faces a major challenge, with development lagging behind the European Green Deal and other strategies toward a climate-neutral, resource-efficient economy. A key cause is that research labs and developers still operate in separation, lacking coordination. Green hydrogen will be essential for future energy systems, with applications in industry, transport, and storage, but progress remains slow. Innovation in materials like electrocatalysts and ionic polymers still follows an uncoordinated, paper-by-paper mode. The barrier is one of integration—transitioning from lab discovery to device deployment.
The EU-funded DECODE project (DE-centralised ClOud Labs for inDustrialisation of Energy Materials) aims to break these barriers with a decentralized, adaptive cloud-connected labs concept. It digitally connects labs advancing methods in materials characterization, modelling, simulation, fabrication, and testing, focused on sustainable hydrogen technologies. DECODE’s modular platform connects universities and institutions to accelerate clean energy R&D. From December 2023, DECODE has worked to bridge gaps between lab-scale tests and real-world conditions, addressing data fragmentation, missing metadata, and ecosystem silos. Integrated workflows are being developed to accelerate R&D timelines and support climate goals. Platform elements include DECODE FABRIC (collaboration matrix), IRL (tool integration scores), FOUNDRY (a knowledge graph using reinforcement learning), and an AI-enabled CPU to orchestrate workflows. In its first phase, DECODE assembled tools to study the local reaction environment (LRE), identified knowledge gaps, and defined Fact Sheets and Property Correlation Trees. Initial multiparametric activity-durability descriptors were generated to guide materials design and integration. Ongoing efforts build correlative workflows linking modelling and testing to device metrics. The DECODE CPU will adaptively coordinate modular R&D tasks in real time. Together, the platform’s tools, knowledge structures, and AI orchestration will accelerate clean energy innovation through coordinated, cloud-connected research.
The EU-funded DECODE project (DE-centralised ClOud Labs for inDustrialisation of Energy Materials) aims to break these barriers with a decentralized, adaptive cloud-connected labs concept. It digitally connects labs advancing methods in materials characterization, modelling, simulation, fabrication, and testing, focused on sustainable hydrogen technologies. DECODE’s modular platform connects universities and institutions to accelerate clean energy R&D. From December 2023, DECODE has worked to bridge gaps between lab-scale tests and real-world conditions, addressing data fragmentation, missing metadata, and ecosystem silos. Integrated workflows are being developed to accelerate R&D timelines and support climate goals. Platform elements include DECODE FABRIC (collaboration matrix), IRL (tool integration scores), FOUNDRY (a knowledge graph using reinforcement learning), and an AI-enabled CPU to orchestrate workflows. In its first phase, DECODE assembled tools to study the local reaction environment (LRE), identified knowledge gaps, and defined Fact Sheets and Property Correlation Trees. Initial multiparametric activity-durability descriptors were generated to guide materials design and integration. Ongoing efforts build correlative workflows linking modelling and testing to device metrics. The DECODE CPU will adaptively coordinate modular R&D tasks in real time. Together, the platform’s tools, knowledge structures, and AI orchestration will accelerate clean energy innovation through coordinated, cloud-connected research.
Over the first reporting period, 20+ novel tools were developed in theory, modelling and experiment to address gaps in understanding the local reaction environment and distribution of reaction rates, from idealized to realistic interfaces. In addition, 140+ methods and tools that existed prior to the project were formalized and adapted or improved upon, or were newly developed within the DECODE project, as Fact Sheets and development of a connective framework, viz. the Property Correlation Tree, linking all methods to key performance indicators. 13 new knowledge gaps as well as 6 new, multiparametric, activity-durability descriptors were identified. Materials and testing conditions of PEFC base case and variation were defined and characterization methods for this base case were selected and assembled into a workflow. Models involved in base case were selected and a variant of the workflow relevant for this case was assembled, with implementation thereof ongoing. The base case samples were distributed and the analysis was started. The data platform for base case data handling was developed and made ready for testing, with on-boarding conducted with all partners. The first working prototype of a multi-agent recommendation system was completed and designed to suggest tailored modeling and experimental toolchains based on user-defined use cases. Advanced metadata standardization and cloud data management were established and the first DECODE platform user workshop was held to introduce participants to the platform. The work on defining the criteria for an industry-grade virtual cloud environment was started, informed by diverse lab workflows and emphasizing the need for standardized metadata to support seamless data integration. Steps toward industrial deployment were initiated, engaging closely with industry partners (e.g. BOSCH, NEL) to define interface, confidentiality, and IT security requirements.
Significant progress was made in the development of adequate tools and methods for generating new knowledge on the vital Local Reaction Environment (LRE). Tools were also developed to seamlessly connect LRE assessment with the distribution of reaction rates in real electrode structures under realistic operating conditions. New insights were gained into key interfacial phenomena, such as the influence of cations on electrocatalytic activity and the pH at the electrode/electrolyte interface, which provide a foundation for the informed design of future generations of electrocatalysts both within and beyond the DECODE project. A formalized approach for defining methods and tools was established through the creation of structured Fact Sheets, enhancing their understanding and integration into a connected framework. These Fact Sheets were aligned with the EMMO ontology, enabling a standardized data collection framework across all experiments and simulations. Furthermore, the DECODE Knowledge Graph was developed to integrate all Fact Sheets and harmonize terminology through ontology-based mappings, thereby enabling advanced, cross-lab semantic search capabilities using tools such as GraphRAG. Complementing these developments, the DECODE FOUNDRY, IRL, and CPU modules were created, and an agentic workflow was established to connect the DECODE FOUNDRY and IRL via AI agents built within the DECODE platform. Together, these innovations form a robust digital and semantic infrastructure to support and accelerate tool interoperability, knowledge generation, and AI-driven research workflows.