The team’s breakthrough centers around an innovative deep learning framework coined the Crossbreeding Neural Network (CBNN). Unlike conventional AI models trained within narrowly defined material boundaries, CBNN simultaneously assimilates data from two chemically distinct catalyst families: carbon-supported single-atom catalysts and perovskite oxide catalysts. This dual learning strategy enables CBNN to capture complementary insights—single-atom catalysts elucidate atomic-scale surface activity while perovskite oxides provide critical information on bulk crystal structure effects. The result is an unprecedented ability to predict properties of a hybrid catalyst class previously unexplored: single-atom catalysts anchored upon perovskite oxide supports.

Mechanistically, this hybrid system exploits the atomic precision of single metal atoms dispersed on catalytically active substrates, merged seamlessly with the structural and electronic versatility inherent in perovskite frameworks. Surface atomic arrangement characteristics are encoded in the model as image data, allowing the AI to interpret local atomic environments visually. Simultaneously, the bulk crystal structure of the oxide, represented by its graph information, informs the network about extended lattice periodicity and connectivity. By integrating these multidimensional data modalities, CBNN constructs a holistic representation of catalyst behavior anchored in both micro- and macro-scale structural features.

To enhance the model’s predictive rigor and interpretability, the research group implemented an automated descriptor-selection pipeline. This pipeline leverages classical statistical techniques coupled with advanced natural language processing (NLP) methods to identify physicochemical descriptors that robustly correlate with catalytic performance across both catalyst families. Key factors singled out include oxidation state, ionic radius, valence d-electron count, electronegativity, and coordination number. Such descriptors succinctly capture the electronic and geometric environment influencing OER activity and allow the AI to generalize learned relationships to novel material domains.

Experimental validation of the CBNN predictions involved synthesizing catalysts within the targeted hybrid family of single-atom-perovskite composites. Twenty-seven distinct catalysts spanning various elemental combinations were produced and rigorously tested under alkaline OER conditions. Impressively, the AI’s activity rankings for twelve of these synthesized materials precisely matched experimental measurements, confirming that the model did not simply interpolate within its training data but extrapolated genuine chemical insights to unknown material classes.

The research did not stop at single-element models; it further expanded into the complex realm of multimetallic catalysts. By computationally screening a vast candidate space of 8,008 permutations containing tungsten (W), molybdenum (Mo), ruthenium (Ru), and rhodium (Rh) single atoms embedded on a calcium–praseodymium cobalt iron oxide perovskite scaffold (Ca0.8Pr0.2Co0.8Fe0.2O3−δ, or CPCF), the AI identified optimal multimetallic compositions. These predictions were experimentally verified, with the top-performing multimetallic catalyst demonstrating superior oxygen evolution rates compared to all mono- and bi-metallic counterparts tested, as well as outperforming traditional perovskite oxides and carbon-supported single-atom catalysts.

Beyond raw catalytic activity predictions, the CBNN framework integrates explainable AI techniques to unpack the atomic-scale design principles responsible for enhanced performance. Visualization of feature importance within the model highlighted synergistic effects arising from specific neighboring metal atom configurations, revealing how electronic interactions at the atomic interface promote catalytic turnover. This level of interpretability is crucial for guiding rational catalyst design and advancing mechanistic understanding beyond black-box computational models.

Director HYEON Taeghwan eloquently summarizes the significance of this work: the AI system transcended conventional boundaries by leveraging heterogeneous datasets to explore uncharted catalyst territories rather than simply optimizing within known categories. This achievement opens the door to a paradigm wherein artificial intelligence holistically synthesizes fragmented experimental knowledge and catalyzes innovation at the interface of multiple material classes.

Looking ahead, the implications of this cross-material AI-driven discovery framework extend far beyond the oxygen evolution reaction or even catalysis. Similar integrative approaches could revolutionize materials development arenas such as battery electrode formulation, complex energy storage device design, and pharmaceutical drug discovery, where disparate heterogeneous datasets remain challenging to unify. By enabling AI to “speak” the common scientific language across domains, the potential for unveiling unanticipated material architectures and performance regimes dramatically expands.

In a landscape where the grand challenge is to design materials that reconcile often competing factors such as activity, stability, and cost, the fusion of AI with cross-class learning exemplified by CBNN offers a visionary roadmap. This study, published in the esteemed journal Nature Materials, marks a watershed moment in the journey toward generalized materials artificial intelligence capable of uncovering entirely new classes of functional materials through hybridized data-driven insight.

As the clean hydrogen economy accelerates, breakthroughs like this underscore the indispensable role of machine learning in pushing the frontiers of catalyst innovation beyond current scientific constraints. The ability to rationally engineer catalysts by bridging atomic to bulk scale phenomena unlocks unprecedented efficiency pathways essential for a sustainable energy future.

Subject of Research: Not applicable

Article Title: Cross-material catalyst discovery via deep learning

News Publication Date: 28-May-2026

Web References: 10.1038/s41563-026-02622-6

Image Credits: Institute for Basic Science

Keywords: Catalysis, Oxygen evolution reaction, Green hydrogen production, Single-atom catalysts, Perovskite oxides, Deep learning, Artificial intelligence, Multimetallic catalysts, Water electrolysis, Catalyst design, Machine learning, Nanomaterials

Catalysts underpin a vast majority of chemical manufacturing processes, facilitating reactions with enhanced speed and selectivity, thereby reducing energy consumption and raw material usage. However, the traditional trial-and-error approach to catalyst development is painfully slow and resource-heavy, limiting innovation at the speed industry demands in the race against climate change. ASCEND addresses these limitations head-on by fusing state-of-the-art digital catalysis methodologies with cutting-edge thin-film catalyst technologies. Digital Catalysis employs Artificial Intelligence (AI), high-fidelity simulations, and autonomous self-driving laboratories (SDLs) to explore and identify high-performance catalyst materials with unprecedented speed.

The thin-film catalyst technology complements this by minimizing material usage while maximizing surface area through innovative nanostructures and 3D architectures. Such designs allow for enhanced interaction between reactants and catalytic sites, which leads to superior catalytic performance and durability. By integrating digital discovery platforms with novel physical embodiments of catalysts, ASCEND aims to deliver sustainable syn-fuels and foundational chemicals that seamlessly substitute fossil-based inputs in critical industrial processes.

At the core of ASCEND is the transformative role of AI-powered autonomous research systems. These SDLs leverage machine learning algorithms to continuously build and refine digital twins—virtual replicas—of experimental systems. The AI system iteratively designs and executes experiments via robotic platforms, analyzing outcomes and adaptively steering the subsequent experimental parameters to optimize catalyst performance metrics. This closed-loop, iterative learning paradigm dramatically compresses experimentation timescales from months or years to mere days or weeks. Notably, while AI orchestrates rapid decision-making, scientists maintain crucial oversight, defining research objectives, interpreting complex results, and ensuring alignment with industrial needs.

This synergy between human ingenuity and autonomous systems epitomizes the future of materials science research. ASCEND builds upon the rich legacy of collaboration between FHI and HZB, leveraging decades of expertise in catalysis and materials characterization. Dr. Karsten Reuter of FHI highlights the strategic leap this approach represents, noting that AI’s capacity to navigate vast, previously uncharted chemical spaces “fundamentally changes how fast science can deliver solutions urgently needed by the chemical sector.” Michelle Browne from HZB echoes this sentiment, emphasizing the acceleration potential that transcends traditional research boundaries.

Dunia Innovations plays a pivotal role in bridging the divide between digital design and real-world, scalable catalyst synthesis. By integrating stress testing protocols under manufacturing-relevant conditions, Dunia ensures that AI-driven discoveries translate into practical, industry-ready solutions. According to Dunia’s CTO, Marcus Tze-Kiat Ng, this combined methodology “accelerates learning while maintaining confidence at scale,” a crucial factor for industrial adoption where reliability and robustness are paramount.

From a technological leadership standpoint, ASCEND aims to drastically shorten the pathway from material discovery to commercial deployment. The project targets catalytic breakthroughs vital for the economic and environmentally sustainable production of green hydrogen and other renewable chemicals. These developments are indispensable prerequisites for heavy industries seeking to decouple from fossil coal and oil feedstocks. BASF Senior Vice President Wolfram Stichert underscores the project’s value in identifying promising new catalysts early, an essential step towards transitioning cutting-edge research into industrial practice.

The urgency for ASCEND’s objectives is underscored by the chemical industry’s significant environmental footprint. It accounts for approximately six percent of global greenhouse gas emissions, equivalent to the annual emissions of the entire European Union, according to S&P Global Ratings and the EDGAR database. A substantial portion of these emissions emanates from fossil-fuel-powered electricity generation and the chemical synthesis of plastics, fertilizers, and pharmaceuticals—fields heavily reliant on fossil feedstocks. Catalysts present one of the most effective levers for reducing these emissions, as about 80% of chemical products involve catalytic stages in their production. Innovation in catalyst design, therefore, constitutes a linchpin for the sector’s transition to greenhouse gas-neutral manufacturing by 2050.

ASCEND is therefore poised as a transformative initiative that not only accelerates fundamental research but also tightly integrates digital innovation with material engineering and industrial validation. Its ambition is to establish new paradigms for catalyst development and deployment, positioning Europe at the forefront of sustainable chemical technology. Success in this endeavor could redefine how industrial catalysis responds to global climate imperatives, enabling scalable, economically viable alternatives to fossil-derived chemicals and fuels.

As the project kicks off in April 2026, the eyes of the scientific and industrial communities will be on ASCEND to witness how its AI-driven experimental workflows and nanotechnology-enabled catalyst designs will reshape the landscape of sustainable chemistry. This initiative represents a critical step forward, harnessing emergent technologies and collaborative expertise to meet global energy and environmental challenges in the most pivotal sectors of industry.

Subject of Research: Accelerator-driven discovery and development of sustainable catalysts for chemical manufacturing through AI and nanotechnology.

Article Title: ASCEND Consortium Launches €30 Million AI-Powered Initiative to Revolutionize Catalyst Development for Decarbonizing the Chemical Industry

News Publication Date: Not specified (Project start date: April 1, 2026)

Web References:
https://mediasvc.eurekalert.org/Api/v1/Multimedia/51024eaf-706d-4778-80d2-f1721b807273/Rendition/low-res/Content/Public

Image Credits: ASCEND Consortium: Helmholtz-Zentrum Berlin, Fritz-Haber-Institut der Max-Planck-Gesellschaft, BASF, Dunia Innovations, Siemens Energy, Technische Universität Berlin / BasCat

Keywords

AI-driven catalyst discovery, self-driving laboratories, digital catalysis, thin-film catalysts, nanotechnology, sustainable chemical manufacturing, green hydrogen, industrial decarbonization, catalytic materials, autonomous experimentation, green chemical synthesis, syn-fuels

The research, recently published in ACS Catalysis, expands the frontier of catalyst discovery by integrating an advanced AI framework directly with state-of-the-art laboratory automation and robotics. This streamlined loop allows the proposed system to perform rapid synthesis, testing, and analysis cycles with unparalleled efficiency. Crucially, the study’s subject—the catalytic conversion of propane into propylene—represents a cornerstone reaction within the chemical industry, underpinning the production of polymers and fibers that permeate everyday life. Traditionally, enhancing catalysts for such essential processes has been a painstakingly slow endeavor, constrained by the need for exhaustive experimental validation and mechanistic insight.

Conventional AI approaches in materials science tend to operate as inscrutable black boxes. They generate predictive results optimized for specific outputs but offer limited or no mechanistic rationale, leaving chemists uncertain about the underlying principles governing observed enhancements. This has raised concerns about the verifiability and reliability of AI-generated discoveries in chemical sciences. Addressing these challenges, the newly developed “gray-box” AI model has been meticulously engineered to balance exploratory agility with transparency. This affords researchers not only improved catalysts but also insight into the synergistic relationships among catalyst components.

The essence of this advancement lies in adaptive experimental planning. By intelligently selecting experiments that yield maximal information about both performance and mechanistic factors, the AI avoids exhaustive brute-force approaches while navigating an enormous combinatorial space. Specifically, it explored over 10 trillion potential multi-promoter catalyst formulations but required fewer than 50 experiments to identify superior candidates. Such efficiency is unprecedented and demonstrates the potential of combining heuristic AI planning with high-throughput experimentation.

Beyond mere identification of superior catalysts, the methodology effectively elucidates the roles of individual promoters—the minor additives that tweak catalytic behavior—and their interactions. These subtle synergies, previously elusive through traditional methods, emerged as central to optimizing catalyst activity and selectivity. The AI’s analytical framework translates complex performance data into chemically intuitive language, bridging the gap between machine intelligence and human expertise. This interplay reveals nuanced catalytic mechanisms, enriching the scientific knowledge base rather than simply improving outcome metrics.

This paradigm shift carries profound implications for the design and discovery of materials across numerous chemical transformations. The study exemplifies how AI is transitioning from a mere black-box optimizer to an agentic scientific partner capable of hypothesis generation, mechanistic interpretation, and adaptive refinement. In doing so, it counters skepticism about the scientific validity of AI-assisted discovery and addresses concerns regarding reproducibility and reliability.

From a broader perspective, the fusion of AI-guided experimentation with robotics embodies the future of autonomous laboratories. These self-driving labs can iterate through experimental cycles at speeds incomprehensible to human operators, systematically exploring vast compositional and processing spaces. However, without interpretability frameworks like the gray-box approach, the output risks being disregarded due to the “black-box” problem. This study’s integrative model demonstrates a viable pathway to overcome that limitation, offering reproducible, interpretable, and actionable insights.

The choice of the propane-to-propylene reaction as the testbed is particularly strategic due to its industrial significance. Propylene is fundamental in manufacturing a wide spectrum of polymers, detergents, and fibers, making catalyst improvements here immediately impactful both economically and environmentally. Enhanced catalysts could reduce energy consumption, increase selectivity, and lower byproduct formation, contributing to greener and more cost-effective production processes.

Moreover, the reduction in experimental workload—down to under 50 from trillions of possibilities—highlights the profound efficiency gains afforded by such AI strategies. This alone could revolutionize research timelines, enabling faster scaling from laboratory discovery to industrial deployment. Importantly, the system’s transparency ensures that accelerated timelines do not compromise the rigor or depth of scientific understanding.

The publication of these findings marks a promising step toward embedding AI more deeply into the scientific method itself. As AI evolves from an assistive tool to a partner capable of offering explanatory narratives behind discoveries, it paves the way for more collaborative workflows. Researchers can leverage AI’s capacity for exhaustive data processing alongside their expertise in mechanistic chemistry, achieving breakthroughs that neither alone could accomplish as efficiently.

In conclusion, this groundbreaking study dispels the myth that rapid AI-guided discovery must come at the expense of interpretability and scientific insight. By pioneering a gray-box framework that seamlessly couples acceleration with understanding, the research unlocks a new dimension of catalyst optimization, poised to transform materials science. The approach not only accelerates the pace of innovation but also enriches the foundational knowledge guiding future explorations, positioning AI as a truly collaborative agent in the scientific quest.

Subject of Research: Catalyst development and AI-driven experimental planning in industrial chemical reactions

Article Title: Adaptive Experiment Planning for Inverse Design and Understanding: Synergistic Interactions as Key to Optimized Multi-Promoter Formulations

News Publication Date: 19-Mar-2026

Web References: 10.1021/acscatal.6c00286

Image Credits: © ACS Catal. 2026

Keywords: Artificial Intelligence, Catalyst Discovery, Experimental Planning, Automation, Propane Conversion, Propylene Production, Multi-Promoter Formulations, Synergistic Interactions, Gray-Box AI, Materials Science, Chemical Industry, High-Throughput Experimentation