Accelerating the discovery of high-performance electrocatalysts through artificial intelligence and robotics technology

Projektinformationen

Dunia.ai

ID Finanzhilfevereinbarung: 101188864

DOI

10.3030/101188864

EK-Unterschriftsdatum 22 November 2024

Startdatum 1 Dezember 2024

Enddatum 30 November 2026

Finanziert unter

• The European Innovation Council (EIC)

Gesamtkosten

Keine Daten

EU-Beitrag

€ 2 499 525,00

Koordiniert durch

DUNIA INNOVATIONS UG
Germany

Periodic Reporting for period 1 - Dunia.ai (Accelerating the discovery of high-performance electrocatalysts through artificial intelligence and robotics technology)

Berichtszeitraum: 2024-12-01 bis 2025-11-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

Catalysts are fundamental to almost all industrial products, as they accelerate chemical reactions and control selectivity, efficiency, and yield across the chemical, energy, and manufacturing sectors. They are therefore critical enablers of the transition toward a climate-neutral industry, particularly for replacing fossil-based processes with electrochemical pathways that utilize captured CO2 to synthesize carbon-neutral fuels and chemicals. Despite their importance, catalyst innovation remains one of the slowest and most expensive components of industrial R&D: discovering and validating a new high-performance catalyst typically requires 15–20 years and multi-million-euro investments, with high technical risk and low reproducibility. This structural bottleneck directly limits the pace at which the EU can deploy clean technologies aligned with the Green Deal, Net Zero targets, and industrial decarbonization strategies.

To address this challenge, Dunia is revolutionizing electrocatalyst discovery through IRIS (Intelligent Research and Innovation System), an autonomous, AI- and robotics-driven laboratory that transforms catalyst R&D into a scalable, data-driven industrial process. Combining physics-informed AI models, robotic synthesis, and device-level electrochemical testing in a closed-loop system enables IRIS to reduce discovery timelines by up to 90% and cuts R&D costs by 60–70% compared to conventional approaches, while simultaneously generating the largest and most reliable experimental electrocatalyst datasets available. This enables Dunia to rapidly identify high-performance catalysts capable of converting captured CO2 into value-added chemicals and green fuels under realistic operating conditions. The project’s expected impact includes accelerated deployment of carbon-neutral industrial processes, reduced innovation risk for European industry, and the creation of a strategic data and technology asset that strengthens Europe’s leadership in AI-enabled clean-energy innovation.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

During the project, Dunia progressed the core technical development of Dunia’s autonomous catalyst discovery platform, IRIS, with the goal of delivering an integrated, reliable system that can design, run, and learn from experiments with minimal human intervention. The main focus was on building the full end-to-end workflow: laboratory automation, experiment orchestration, data capture and storage, and machine-learning models operating together as one operational platform.

A major achievement was the development and validation of IRIS Platform V2. We integrated robotics and automated lab workflows with software orchestration, data management, and ML components into a single system and ran extended trials to test stability and robustness. In total, 2,533 automated platform tasks were executed during the first half of the project only. System reliability improved steadily, and by the end of the first half of the project the overall failure rate decreased to 0.19%, demonstrating strong progress toward stable and repeatable autonomous operation.

In parallel, we built a unified data infrastructure that allows structured ingestion, tracking, and versioning of experimental results, performance metrics, degradation/stability data, and analytical outputs. This enables consistent descriptor generation, model training, and closed-loop optimisation, and establishes the foundation for reproducible, large-scale CO2 reduction research. Building on these capabilities, we initiated AI-driven catalyst discovery campaigns in closed-loop mode, where algorithms select candidates, the platform runs automated experiments, and models are updated based on results. This demonstrates meaningful progress toward identifying high-performance and stability-aware CO2 reduction catalysts and toward maturing the overall system for industrially relevant operation.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

Dunia is advancing beyond the current state of the art by moving from “partial automation” to a fully integrated, closed-loop autonomous discovery system. Most existing approaches automate isolated steps (e.g. synthesis, screening, or analytics) and still rely on manual handoffs, fragmented data, and low reproducibility. In contrast, IRIS integrates algorithm-driven candidate generation, automated catalyst synthesis and preparation, device-level electrochemical testing, and multimodal characterization data capture into one operational platform with unified data lineage. A key differentiator is that IRIS continuously generates high-fidelity experimental data under controlled and repeatable conditions, enabling models to learn from real outcomes rather than sparse, inconsistent, or simulation-heavy datasets. This addresses a core failure mode of AI in materials: lack of reliable industrial-grade training data and the inability to iterate fast enough for AI to converge on scalable solutions.

The expected impact is a step change in the speed, cost, and reliability of catalyst innovation, with the potential to reduce discovery timelines by up to ~90% and cut R&D cost by >50%, while lowering scale-up risk through stability-aware optimization and reproducible experimentation. This enables faster development of catalysts for CO2 utilization and other decarbonization-critical processes, improving the feasibility and economics of low-carbon fuels and chemicals. To ensure uptake and long-term success, the next needs are: (1) demonstration at industrially relevant conditions (extended duration, realistic impurities, operating windows) and validation of durability/aging behavior; (2) integration with industrial partner workflows and clear performance benchmarks tied to techno-economic metrics; (3) standardisation and interoperability (common reporting formats for catalyst performance and stability, audit-ready data/provenance, and alignment with emerging best practices for autonomous labs); (4) robust IPR and data governance frameworks that protect customer confidentiality while enabling scalable learning and commercialization; and (5) continued access to markets and finance to scale platform capacity and industrial deployments, including internationalisation through strategic partnerships and pilot sites. Together, these steps position Dunia to convert technical advantage into sustained industrial adoption and measurable climate and economic impact.

Automated Electrocatalyst Deposition Module