Multimetallic OER Catalyst Discovery Dataset from Autonomous Robotic Synthesis and Electrochemical Characterization

Datasets from experimentation on the FastCat robotic platform with alkaline nickel catalyst with dopants Cr Al Fe Co Mn Ni Cu Zn. In total 502 experiment and in total 7449 datasets and 22.901 attributes related to the experiments (eg. temperature, humidity, pressure, date, etc.)

This dataset supports the publication "Autonomous Discovery of Multielement LDH Alloy Catalysts for Alkaline OER" in Advanced Intelligent Discovery 2025. It presents data from the platform FastCat based on the N9 from North Robotics which was used for the synthesis of OER catalyst for alkaline water splitting using Ni, Fe, Cr, Co, Mn, Cu, Al, Zn metal nitrates, in combination with Nickel foam substrates, forming Ni'X' Layered Double Hydroxides (LDH).

The dataset in the HDF5 container comprehensively includes all data and scripts related to the synthesized catalysts. The easiest is to go to README.md and afterwards run the main.py which will execute all python files. This will in essence plot data and combine metadata into a metadata.xlsx file for manual investigation of key parameters.

How to refer to this dataset?

Use this citation when citing the dataset:
Fisker-Bødker, Nis (2025). Multimetallic OER Catalyst Discovery Dataset from Autonomous Robotic Synthesis and Electrochemical Characterization. Technical University of Denmark. Dataset. https://doi.org/10.11583/DTU.28494185

Kindly also consider to cite the related paper:
Fisker-Bødker, Nis, Autonomous Discovery of Multielement LDH Alloy Catalysts for Alkaline OER, Advanced Intelligent Discovery, 2025.

Use Cases of the Dataset

This dataset provides a rich, structured collection of electrochemical measurements and experimental metadata across a systematically varied composition space of multimetal oxide catalysts. Its design supports several advanced use cases:

1. Machine Learning for Materials Optimization

The dataset is ideally suited for supervised ML tasks such as:

• Regression: Predicting catalytic performance metrics (e.g., overpotential at given current densities) from composition and synthesis parameters.
• Classification: Identifying high-performance vs. low-performance catalysts.
• Bayesian optimization: Guiding autonomous experimentation through learned surrogate models that predict outcomes and uncertainties.
• Feature attribution: Understanding which compositional or procedural variables most affect performance (e.g., using SHAP, LIME).

The temporal structure, replicated measurements, and detailed metadata (e.g., temperature, pressure, precursor treatment) also allow:

• Uncertainty quantification
• Learning causal relationships
• Generalization to out-of-sample compositions

2. Experimental Validation of DFT Predictions

The dataset is valuable for validating or calibrating DFT-predicted trends:

• Benchmarking: Compare experimentally measured overpotentials with theoretical OER energetics (e.g., ΔG for *OH, *O, *OOH intermediates).
• Descriptor modeling: Use DFT-derived properties (e.g., d-band center, charge transfer) as inputs and correlate with experimental outcomes.
• Data fusion: Combine DFT and experimental data into joint models (e.g., via multi-fidelity Gaussian processes or co-kriging).
• Inverse design: Use experimental performance labels to train models that prioritize DFT screening of unexplored compositions.

3. Structure–Property Relationship Mapping

The granularity of your dataset enables building interpretable structure–property models:

• Relate precursor ratios and processing conditions to final electrocatalytic behavior
• Investigate sensitivity to impurities, synthesis duration, or environmental factors
• Map non-linear or coupled effects (e.g., Cr–Fe–Ni synergistic behavior)

Dataset Structure

📁 root (HDF5)
├── 📄 keyParameters
├── 📁 ...
├── 📁 899_Cr0.05_Al0.05_Fe0.35_Co0.55_Mn0.0_Ni0.0_V0.0_FeCl0.55
│   ├── 📄 Attributes
│   ├── 📄 0EISacv
│   ├── 📄 0CVact0.2mVsx25
│   ├── 📄 1CVact0.01mVsx1
│   ├── 📄 2CVact0.2mVsx100
│   ├── 📄 3CVact0.05mVsx3
│   ├── 📄 4CVact0.075mVsx3
│   ├── 📄 5CVact0.1mVsx3
│   ├── 📄 6CVact0.15mVsx3
│   ├── 📄 7CVact0.2mVsx3
│   ├── 📄 8EISacv
│   ├── 📄 9CP100.0mA
│   ├── 📄 10CP50.0mA
│   ├── 📄 11CP20.0mA
│   ├── 📄 12CP10.0mA
│   ├── 📄 13CP5.0mA
│   ├── 📄 14CP2.0mA
│   ├── 📄 15CP1.0mA
│   ├── 📄 16CVact0.01mVsx1
├── 📁 ...
├── 📁 1315_Cr0.0_Al0.15_Fe0.4_Co0.05_Mn0.0_Ni0.2_Cu0.2_Zn0.0
│   ├── 📄 Attributes
│   ├── 📄 0EISacv
│   ├── 📄 0CV0.4mVsx100
│   ├── 📄 1CV0.01mVsx2
│   ├── 📄 8EISacv
│   ├── 📄 10CP50.0mA
│   ├── 📄 11CP20.0mA
│   ├── 📄 12CP10.0mA
│   ├── 📄 13CP5.0mA
│   ├── 📄 14CP2.0mA
│   ├── 📄 15CP1.0mA

Group Naming Convention

Each group is named UID_chemical composition. The composition describes the dispensed % of each metal nitrate stock solution.

Dataset Naming Convention

Each dataset follows a convention where the leading number indicates the acquisition order.

• CVXX... = Cyclic Voltammetry
• CPXX... = Chronopotentiometry, where XX indicates the applied current (in mA)
• EISacv = Electrochemical Impedance Spectroscopy

Column Descriptions

CV datasets (Cyclic Voltammetry):

• Time (s)
• Potential (WE vs. RHE) [V]
• Vu (V)
• Current [A]
• Vsig
• Ach (V)
• IERange
• Overbit1
• Stop Test
• Scan cycle
• Temperature (C)

CP datasets (Chronopotentiometry):

• Time [s]
• Potential (WE vs. RHE) [V]
• Vu (V)
• Current [A]
• Charge Q
• Vsig
• Ach (V)
• IERange
• Overbit1
• Stop Test

EIS datasets (Impedance Spectroscopy):

• Point
• Time [s]
• Freq
• Zreal [ohm]
• Zimag [ohm]
• Zsig
• Zmod
• Zphz
• Idc
• Vdc
• IERange

Attributes in Each Measurement Folder

Each measurement group contains an Attributes object that stores metadata including:

• UID, Date, Group_name, composition (Cr, Al, Fe, Co, Mn, Ni, Cu, Zn)
• Electrochemical parameters (e.g., Potential @ XX A/cm², ohmic resistance)
• Temperature, humidity, pressure (initial, start, end)
• Pre-treatment and synthesis steps (e.g., oxide removal time, dip time, electrolyte used)
• Status (string indicating data quality or flag)

keyParameters Dataset

A centralized summary table across all UIDs with:

• UID
• Current [A/cm²]
• Potential (uncorrected) [V]
• Potential (ohmic-corrected) [V]
• Ohmic resistance [Ω]