Machine-Learning-Enabled Quantification of Metal-Based Nanoparticle Sizes Using Linear Sweep Voltammetry

Metal-based nanoparticles serve as critical catalysts
in reactions,
such as hydrogen evolution and oxygen reduction, with their catalytic
activity strongly influenced by particle size due to the higher density
of active sites in smaller particles. Current techniques for nanoparticle
size characterization often rely on costly and complex instrumentation,
limiting their efficiency. Here, we present “LSV2NP”,
an innovative machine learning approach leveraging the gradient boosting
regression (GBR) model to predict nanoparticle sizes with the help
of the selected overpotential data from a linear sweep voltammetry
(LSV) test. Utilizing an extensive database of nanoparticle compositions
and electrochemical properties, LSV2NP achieves accurate size predictions
through data-driven insights. Validation with commercial Pt/C and
PtCo/C catalysts confirms the model’s high predictive accuracy.
Furthermore, we demonstrate the characterization capabilities of LSV2NP
following a 15 h chronoamperometry test on Pt/C. A subsequent LSV
measurement, analyzed via LSV2NP, provided particle size estimates
consistent with transmission electron microscopy observations, revealing
aggregated particle sizes. LSV2NP offers a powerful tool for catalyst
development and the broader study of nanomaterials, streamlining nanoparticle
size characterization.