Autonomous scanning electrochemical cell microscopy enables rapid exploration of large compositionally complex material spaces

Abstract: Alloying is a central strategy in electrocatalysis, enabling fine-tuning of electronic structure. In particular, compositionally complex solid solutions (CCSS) often called high-entropy alloys are of high interest as they allow active site design. However, the "combinatorial explosion" in the number of possible compositions poses a critical bottleneck for the discovery of active CCSS electrocatalysts. We present an autonomous scanning electrochemical cell microscopy (SECCM) system for ultrahigh-throughput and large-scale CCSS activity screening. The platform rapidly establishes composition-electrocatalytic activity relationships for large compositional spaces across multiple thin-film CCSS materials libraries via active learning and automated library exchange. Embedding analytical expressions of voltammetry in the algorithm enables the learning of whole voltammograms rather than a single selected metric. As a demonstration, we investigated hydrogen evolution reaction (HER) activities of Au-Ir-Rh, where Ir and Rh exhibit strong metal-hydrogen binding and Au exhibits relatively weak binding as derived from the HER volcano plot. The composition-activity trend was accurately predicted after measuring only 15% of all 966 measurement areas. Au30Ir20Rh50 and Au10Ir35Rh55 exhibit highest activities with standard rate constants of about 0.012 cm/s, demonstrating positive synergistic contributions from elemental mixing. The autonomous robotic SECCM platform is broadly applicable to a wide range of electrocatalytic reactions, providing a general pathway for accelerating CCSS electrocatalyst discovery and optimization.

• The paper introduces an autonomous robotic SECCM platform that rapidly screens complex electrocatalyst libraries by capturing complete LSVs for enhanced kinetic analysis.
• It employs an active learning framework with multi-output Gaussian processes to decouple mass transport effects and rigorously map kinetic parameters in compositionally complex solid solutions.
• The study demonstrates high-fidelity HER activity mapping in Au-Ir-Rh catalysts with only 15% of measurements, showcasing significant improvements in resource efficiency and analytical accuracy.

Autonomous Robotic SECCM for Accelerated Screening of Compositionally Complex Electrocatalysts

Introduction and Problem Context

The optimization of electrocatalysts for reactions such as the hydrogen evolution reaction (HER) is increasingly reliant on exploring compositionally complex solid solutions (CCSS), including high-entropy alloys. These multimetallic systems offer a high-dimensional parameter space for catalyst design, promising finely tailored electronic structures and site-specific activity tuning. The combinatorial explosion of possible compositions, however, creates a critical bottleneck for conventional experimental techniques. Even at moderate grid resolutions, the number of unique compositions in ternary, quaternary, and higher-order systems quickly becomes intractable, impeding rapid discovery—even when leveraging thin-film combinatorial libraries.

Recent advances in high-throughput synthesis and electrochemical characterization techniques, such as scanning droplet cell (SDC) and long-range scanning electrochemical cell microscopy (SECCM), have improved the characterization rates and quality for thousands of samples per campaign. Yet, conventional approaches still sequentially probe libraries and typically reduce the rich information in voltammetric curves to single scalar metrics, sacrificing mechanistic insight and kinetic detail.

Autonomous SECCM Platform Architecture

This work introduces an autonomous robotic SECCM platform integrating the following modules: long-range SECCM with a nanocapillary probe, a storage cassette for multiplexed handling of up to ten thin-film CCSS libraries, and a six-axis robotic arm for hands-free sample exchange. This instrument operates entirely within a grounded Faraday cage and includes an environmental chamber surrounding the SECCM tip to maintain a controlled atmosphere (humidified Ar), avoiding competing oxygen reduction during HER characterization. Automated positioning leverages force sensors, piezoelectric actuators, and computer-vision-based feedback using microscopy imaging to guarantee precise tip-sample approach and robust wetting.

The continuous, unattended operation is enabled by a workflow that encompasses library loading, tilt correction, surface approach, measurement, analytical data fitting, and algorithmic selection of the next measurement target. The system executes active learning cycles across multiple libraries, dynamically optimizing the information gain per experiment and handling sample transfer without user intervention.

Data Acquisition, Analytical Framework, and Active Learning

The robotic SECCM acquires full linear sweep voltammograms (LSVs) for HER in each measured composition. Each LSV is analytically fitted within the Butler-Volmer formalism under steady-state mass transport constraints, yielding the standard rate constant ($k^0$) and the transfer coefficient ($\alpha$) as kinetic descriptors decoupled from mass transfer effects. This contrasts with most earlier methods that utilized only a fixed current density or a single parameter as the performance metric, thus discarding significant kinetic information.

The input features for compositional mapping are derived from high-throughput EDX, and validated through XPS to ensure surface-sensitive provenance (mean Euclidean distance between surface/volume compositions: 4.2 at.%). Phase stability and absence of extrinsic morphological variation are confirmed via XRD (single-phase fcc solid solution throughout) and AFM (roughness 0.7–1.6 nm across libraries).

The active learning framework employs multi-output Gaussian processes with a coregionalization kernel, capturing correlations between $k^0$ and $\alpha$. Cost-aware acquisition is used, assigning measurement priorities based on epistemic uncertainty and estimated sample transfer/measurement overhead. Notably, the platform is equipped with a noise-aware extension of the GP likelihood, incorporating the variance from triplicate LSV fits to avoid overfitting when learning in log-space—a common failure mode in high-throughput kinetic screening.

Experimental Demonstration: Au-Ir-Rh Ternary Catalyst System

The platform’s capability is demonstrated using three combinatorial thin-film libraries in the Au-Ir-Rh ternary system, chosen to span a range of metal-hydrogen binding affinities and probe for synergistic effects in HER. The campaign comprises 966 unique measurement areas, each subjected to SECCM-based voltammetry, LSV fitting, and model updating.

Key outcomes:

• The LSV fitting approach enables extraction of distribution maps for the limiting current, $k^0$, and $\alpha$ across the composition space with no discontinuities at library boundaries, reflecting the continuous atomic arrangement distributions in CCSS.
• The highest HER activities (by $k^0$) are realized at compositions near Au$_{30}$Ir$_{20}$Rh$_{50}$ and Au$\alpha$0Ir$\alpha$1Rh$\alpha$2, with $\alpha$3 cm s$\alpha$4. This represents a measurable synergistic enhancement relative to the binary edges and constituent elements.
• The entire composition-activity landscape is inferred with high fidelity after only $\alpha$515% of all possible measurements (140 out of 966), as indicated by the plateauing MAE in active learning and corroborated by real-time predictive variance.
• Use of the full LSV as the predictive target rather than a fixed current metric enables generalization to kinetic regimes and accurate mechanistic probing across the full dataset.

Implications, Limitations, and Future Directions

This autonomous SECCM platform establishes a new paradigm for rapid electrocatalyst discovery and optimization, especially in high-dimensional materials spaces where exhaustive measurement is prohibitively expensive. By integrating advanced robotic automation, analytical fitting, compositional validation, and robust uncertainty-aware active learning, the platform offers significant improvements in data quality, throughput, and resource efficiency.

The approach is agnostic to the particular electrocatalytic reaction under study, provided an appropriate analytical voltammetric model is available for fitting. Further expansion to more than 50 libraries per campaign is technically feasible and would enable systematic mapping of higher-order compositional spaces, such as quaternary or quinary CCSS.

Potential future developments include:

• Extension beyond HER to oxygen evolution/reduction and CO$\alpha$6 reduction reactions, with broader application in energy conversion and storage materials.
• Deeper integration of atomic-scale characterization (e.g., in situ spectroscopy) synchronized with SECCM kinetic mapping.
• Hierarchical model-based optimization combining autonomous SECCM with computational screening approaches (DFT, ML-based predictors).
• Deployment of multi-modal data acquisition (e.g., integrating optical, electronic, and catalytic descriptors) in active learning cycles.

Conclusion

The described autonomous robotic SECCM system achieves scalable, resource-efficient exploration of large, complex material spaces by combining high-quality nanoscale electrochemistry, robust automation, and advanced active learning. The study demonstrates that with only a fraction of the possible measurements, the critical features of the composition-activity landscape in multimetallic thin-film electrocatalysts can be faithfully reconstructed. This establishes a practical and generalizable workflow for accelerated discovery and optimization in catalysis and related fields, enabling swift navigation of the combinatorial complexity inherent in CCSS materials (2606.00779).