HZB's Tailored Digital Lab Workflows Towards AI-ready Datasets

Efficient data management and structured digital workflows are essential for transforming experimental
science toward FAIR datasets. We implement a comprehensive digital lab infrastructure that links
experimental data across the full sample lifecycle—from synthesis to advanced characterization—
ensuring machine-readable, metadata-rich datasets.
At HZB, Data Stewards and Laboratory Scientists collaborate closely to integrate software, metadata,
and experimental data. Data Stewards develop the open-source platform NOMAD Oasis to structure data
according to FAIR principles and harmonize metadata using collaboratively created vocabularies (e.g.,
voc4Cat, TFSCO). Laboratory Scientists generate and document heterogeneous experimental data,
which are captured through digital laboratory workflows and subsequently analyzed using jointly
developed, customized Jupyter notebooks. Through this close partnership, both groups produce
interoperable, reusable datasets that support automated analyses and AI-driven applications.
In this contribution, we present two workflows developed for Thin Film Catalyst laboratories focusing
on Thermocatalysis and Electrocatalysis. These workflows systematically capture and structure
research data across multiple lab processes, enabling high-throughput analysis, AI-driven insights, and
efficient reuse. By highlighting our design decisions for optimizing experimental parameters using
Bayesian optimization and for creating linked datasets to efficiently build combinatorial libraries, we aim
to share practical strategies for implementing FAIR-aligned, metadata-rich digital lab workflows in
heterogeneous experimental environments.