Discovery of Ni^(I) Complexes for CO₂ Insertion Enabled by a Machine Learning-Computational-Selection Sequence.

Area of Science:

• Catalysis
• Computational Chemistry
• Materials Science

Background:

• Catalyst speciation is crucial for efficiency, reactivity, and selectivity.
• Understanding factors dictating catalyst speciation is limited, often relying on trial-and-error.
• Predictive tools for ligand selection to control metal speciation are needed.

Purpose of the Study:

• To evaluate machine learning combined with computational activation barrier predictions for guiding ligand selection.
• To achieve CO2 insertion at room temperature for vulnerable Ni(I)-Ph complexes.
• To identify ligands favoring the Ni(I) oxidation state for enhanced reactivity.

Main Methods:

• Computational rationalization of Ni(I) vs. Ni(II) reactivity towards CO2 insertion.
• Construction of an in silico descriptor database for machine learning.
• Machine learning prediction of ligands favoring Ni(I) oxidation state, filtered by activation barriers.
• Synthesis and experimental testing of predicted ligands for CO2 insertion.

Main Results:

• Identified ligands that favor the reactive Ni(I)-Ph intermediate.
• Predicted and confirmed room temperature reactivity for CO2 insertion.
• Demonstrated alignment between computational predictions and experimental results.

Conclusions:

• Machine learning and computational chemistry offer a blueprint for predicting ligands that control metal complex oxidation state and reactivity.
• This approach can guide ligand design for desired catalytic transformations.
• Enables prediction of ligands, including novel ones, for targeted catalyst performance.