Machine Learning for Estimating Electron Transfer Rates From Square Wave Voltammetry.

Area of Science:

• Surface electrochemistry
• Computational chemistry
• Machine learning applications

Background:

• Electron transfer rates of surface-bound molecules are critical for electronic and sensor applications.
• Determining these rates typically requires intensive experimental or computational methods.
• Developing faster, more accessible methods is essential for broader application.

Purpose of the Study:

• To evaluate machine learning (ML) models for extracting electron transfer rates from large electrochemical datasets.
• To compare the performance of different ML algorithms against first-principles computational measures.
• To establish ML as a viable tool for rapid rate determination in surface electrochemistry.

Main Methods:

• Utilized large voltammetry datasets from experimental electrochemistry.
• Applied machine learning techniques including decision tree ensembles, neural networks, and Gaussian process regression.
• Validated ML model predictions against electron transfer rates computed using first-principles methods.

Main Results:

• A random forest model with 80 trees achieved a root mean squared error (RMSE) of 0.37 s⁻¹ (0.38% mean error).
• A neural network with Bayesian regularization yielded an RMSE of 0.49 s⁻¹ (0.52% mean error).
• Both models demonstrated high accuracy in reproducing computationally derived electron transfer rates.

Conclusions:

• Machine learning methods can accurately and rapidly determine electron transfer rates from experimental electrochemical data.
• This work validates ML as a powerful tool for analyzing large datasets in surface-bound electrochemistry.
• The developed ML approaches facilitate widespread applications in electronics and sensor development.