Maximal Predictability Approach for Identifying the Right Descriptors for Electrocatalytic Reactions.

Area of Science:

• Computational materials science
• Electrochemistry
• Quantum chemistry

Background:

• Density functional theory (DFT) is widely used for discovering new materials with high activity.
• DFT calculations have inherent uncertainties that hinder the accurate prediction of material activity.
• Existing methods struggle to distinguish between materials with genuinely high activity.

Purpose of the Study:

• To develop a framework for propagating uncertainty through thermodynamic activity models in DFT.
• To introduce a new metric, prediction efficiency, for quantifying the distinguishability of material activity.
• To identify optimal descriptors for predicting material activity.

Main Methods:

• Implementing uncertainty propagation through thermodynamic activity models.
• Calculating probability distributions for computed activity and expectation values.
• Defining and applying the prediction efficiency metric.

Main Results:

• Demonstrated a method to generate probability distributions of computed material activity.
• Introduced prediction efficiency as a quantitative measure for material distinguishability.
• Applied the framework to hydrogen evolution, chlorine evolution, oxygen reduction, and oxygen evolution reactions.

Conclusions:

• The developed framework enhances the ability to predict and distinguish highly active material candidates.
• Prediction efficiency offers a quantitative approach to optimize descriptors for material discovery.
• This method can significantly improve the accuracy of identifying promising materials for electrochemical applications.