Electrocatalysis R&D | Data-Driven Catalyst Development

PEMFC · PEMWE · Pt / PtCo · IrOx · ORR / OER
Electrochemistry · XRD · DoE · Bayesian Optimization · Research Automation

I work on electrocatalyst development for PEM fuel cells and water electrolysis, with experience spanning catalyst synthesis, characterization, electrochemical evaluation, and process scale-up.

My current focus is on connecting experimental design, catalyst characterization, and data-driven optimization to better understand synthesis–structure–activity relationships and make catalyst R&D more efficient.

Many of the tools below originated from practical problems encountered during catalyst development rather than from software development alone.

Closed-loop experimental optimization for catalyst development

A Gaussian-process-based workflow for recommending the next experiment from existing catalyst datasets.

Methods

• Gaussian Process surrogate modeling
• Matérn 5/2 + ARD kernel
• EI / PI / UCB acquisition functions
• Constraints on stability, cost, particle size, etc.
• Batch experiment recommendation
• Multi-objective optimization
• Uncertainty reporting and experiment traceability

Example applications

• PtCo L1₀ ordering optimization
• IrOx OER catalyst optimization
• synthesis–structure–activity exploration

→ bayesian-optimization-electrocatalyst

Offline electrochemical analysis for PEMFC catalyst evaluation

A browser-based tool for calculating:

• ECSA from H-upd
• ORR mass activity
• ORR specific activity
• N₂ background correction
• kinetic current using the Koutecký–Levich equation

Also includes utilities for processing Autolab .paax and Origin .opju data.

The goal is to make routine catalyst evaluation more reproducible and reduce repetitive manual processing.

→ PEMFC-Electrocatalyst-Activity-Analyzer

XRD analysis for PtCo intermetallic ordering

Offline tool for estimating PtCo L1₀ chemical ordering from characteristic XRD reflections.

Methods include:

• SNIP background subtraction
• Gaussian peak fitting
• Levenberg–Marquardt optimization
• overlapping peak deconvolution
• characteristic superlattice/fundamental peak analysis

Developed from practical PtCo catalyst characterization work and currently focused on PtCo/Pt-based catalyst systems.

→ XRD-PtCoOrdering

Tools for estimating catalyst crystallite size using:

• Scherrer analysis
• Williamson–Hall analysis
• integral-breadth approaches
• Pt / PtCo-specific diffraction workflows

→ XRD-Scherrer

My longer-term goal is to connect these tools into a practical closed-loop catalyst-development workflow:

Synthesis Variables
        ↓
Design of Experiments
        ↓
Catalyst Synthesis
        ↓
Characterization
XRD · Electrochemistry · Particle Size
        ↓
Structure–Activity Dataset
        ↓
Statistics / Gaussian Process
        ↓
Bayesian Optimization
        ↓
Next Experiment

Rather than treating modeling as a separate activity, I am interested in using computational methods directly inside the experimental workflow.

• Electrocatalyst synthesis and scale-up
• PEM fuel-cell catalysts
• PEM water-electrolysis catalysts
• Pt and Pt-alloy ORR catalysts
• Ir-based OER catalysts
• synthesis–structure–activity relationships
• catalyst benchmarking and measurement uncertainty
• Design of Experiments
• Bayesian optimization and active learning
• data-driven experimental planning

I also build small tools and integrations to improve scientific and technical workflows.

• catalyst-search-skill — Literature retrieval and structured evidence synthesis for catalysis research.
• catalyst-design-skill — Traceable catalyst-design reasoning with literature sources, confidence levels, and validation paths.
• opencode-responses-bridge-skill — Local compatibility bridge between Chat Completions and Responses-style APIs.

• Zotero integrations
• literature/RAG workflows
• Hermes Agent extensions
• research automation utilities

These projects are secondary to my catalyst work but support the same objective: reducing repetitive work and shortening the path from an experimental question to a testable result.

My background is primarily in industrial electrocatalyst R&D.

I use programming, statistics, and AI-assisted development mainly as tools for solving experimental and engineering problems rather than as ends in themselves.

Current direction:

experiment → characterization → data → model → next experiment

Y. Peng · Electrocatalysis R&D · Research Tools · Data-Driven Experimentation