We demonstrate a framework for real-time active learning in state-changing experimental systems, integrating path-dependent acquisition functions with a fully autonomous, cloud-controlled experimental platform to accelerate materials discovery.
\nSolid state materials are central to many modern technologies in which a given material may be exposed to a variety of environments. The material properties often vary with the sequence of environments in an irreversible manner, resulting in a quintessential path-dependency in experimental observables. While sequential learning techniques have been effectively deployed for accelerating learning of state properties of materials, they often use a consistent environment path in all experiments. To elevate such techniques for making optimal decisions in experimental investigations of path-dependent properties, we introduce an iterated expected information gain acquisition function that optimizes over entire experimental trajectories. This approach is implemented within a cloud-based Materials Acceleration Platform architecture utilizing an event-driven stateful broker coupled with remote HELAO (Hierarchical Experimental Laboratory Automation and Orchestration) instances and an AI science manager. The platform's efficacy was demonstrated through a case study optimizing multi-step spectro-electrochemical experiments to identify optically stable potential windows in (Co–Ni–Sb)Oz metal oxides. The system successfully integrated AI-driven experiment design, remote laboratory automation, and cloud-based data infrastructure, validating the platform's capability for managing complex, adaptive, path-dependent workflows in materials discovery.
", "doi": "10.1039/d5dd00325c", "issn": "2635-098X", "publisher": "Royal Society of Chemistry", "publication": "Digital Discovery", "publication_date": "2025-12-01", "series_number": "12", "volume": "4", "issue": "12", "pages": "3674-3682" }, { "id": "authors:djvrc-zcv46", "collection": "authors", "collection_id": "djvrc-zcv46", "cite_using_url": "https://authors.library.caltech.edu/records/djvrc-zcv46", "type": "article", "title": "Accelerated screening of carbon dioxide capture by liquid sorbents", "author": [ { "family_name": "Jones", "given_name": "Ryan J. R.", "orcid": "0000-0002-4629-3115", "clpid": "Jones-Ryan-J-R" }, { "family_name": "Lai", "given_name": "Yungchieh", "orcid": "0000-0001-9392-1447", "clpid": "Lai-Yungchieh" }, { "family_name": "Kan", "given_name": "Kevin", "orcid": "0000-0002-5840-8928", "clpid": "Kan-Kevin" }, { "family_name": "Guevarra", "given_name": "Dan", "orcid": "0000-0002-9592-3195", "clpid": "Guevarra-Dan" }, { "family_name": "Haber", "given_name": "Joel A.", "orcid": "0000-0001-7847-5506", "clpid": "Haber-Joel-A" }, { "family_name": "Ramirez", "given_name": "Natalia M.", "clpid": "Ramirez-Natalia-M" }, { "family_name": "Zito", "given_name": "Alessandra", "orcid": "0000-0002-1111-5415", "clpid": "Zito-Alessandra" }, { "family_name": "Li", "given_name": "Clarabella", "orcid": "0000-0002-5551-4506", "clpid": "Li-Clarabella" }, { "family_name": "Yang", "given_name": "Jenny Y.", "orcid": "0000-0002-9680-8260", "clpid": "Yang-Jenny-Y" }, { "family_name": "Appel", "given_name": "Aaron M.", "orcid": "0000-0002-5604-1253", "clpid": "Appel-Aaron-M" }, { "family_name": "Gregoire", "given_name": "John M.", "orcid": "0000-0002-2863-5265", "clpid": "Gregoire-J-M" } ], "abstract": "The ideation of carbon capture, concentration, and utilization technologies is establishing a need for carbon dioxide sorbents with specific binding, release, and chemical specifications. While computational workflows can help navigate the broad search space of molecular sorbents, automated experimental screening platforms are relatively underdeveloped. We present a carbon capture screening instrument to characterize the carbon dioxide binding and sorption capacity of liquid sorbent media. We discuss the extension of this capability to characterize the loaded liquid sorbent as well as the headspace to facilitate study of the carbon dioxide adduct, for example its electrochemical activation. The fabrication and computer automation instructions are provided so that the experimental technique can be implemented in a broad range of materials acceleration platforms involving gas sorption.
\nOperational durability is poorly characterized by traditional (photo)electrocatalyst discovery workflows, creating a barrier to scale-up and deployment. Corrosion is a prominent degradation mechanism whose thermodynamics depend on the concentration of corrosion products in electrolyte. We present an automated system for characterizing the equilibration of (photo)electrodes with dissolved metals in electrolyte for a given electrode, pH, and electrochemical potential. Automation of electrode selection, electrolyte preparation, and electrolyte aliquoting enables rapid identification of self-passivating electrodes and estimation of the equilibrium dissolved metals concentrations. The technique is demonstrated for metal oxide photoanodes in alkaline electrolyte, where BiVO4 is found to continually corrode, in agreement the literature. An amorphous Ni−Sb−O photoanode is found to passivate with a Ni-rich coating on the order of 1 monolayer with less than 1\u2005μM total dissolved metals in electrolyte, demonstrating its suitability for durable photoelectrochemical operation. The automation and throughput of the instrument are designed for incorporation in accelerated electrocatalyst discovery workflows so that durability can be considered on equal footing with activity.
\nThe materials research community is increasingly using automation and artificial intelligence (AI) to accelerate research and development. A materials acceleration platform (MAP) typically encompasses several experimental techniques or instruments to establish a synthesis-characterization-evaluation workflow. With the advancement of workflow orchestration software and AI experiment design, the scope and complexity of MAPs are increasing, however each MAP typically operates as a standalone entity with dedicated experiment, compute, and database resources. The data from each MAP is thus siloed until subsequent efforts to integrate data into complex schema such as knowledge graphs. To lower the latency of data integration and establish an extensible community of MAPs, we must expand our automation efforts to include data handling that is decoupled from the resources of each MAP. Event-driven pipelines are well established in the computational community for building decoupled data processing systems. Such pipelines can be difficult to implement de novo due to their distributed nature and complex error handling. Fortunately, the broader computational science community has established a suite of cloud services that are well suited for this task. By leveraging cloud computing resources to establish event-driven data management, the MAP community can better realize the ideals of extensibility and interoperability in materials chemistry research.
\n