From computational screening to the synthesis of a promising OER catalyst

Sai Govind Hari Kumar, Carlota Bozal‐Ginesta, Ning Wang, Jehad Abed, Chung Hsuan Shan, Zhenpeng Yao, Alán Aspuru‐Guzik

TL;DR

This paper presents a high-throughput computational framework to accelerate the discovery of new oxygen evolution reaction (OER) catalysts. By screening over 6,000 ternary spinel oxides, they identified promising candidates, synthesized three, and discovered a novel material, Co₂.₅Ga₀.₅O₄, that rivals current state-of-the-art catalysts.

What Problem Does This Paper Address?

The urgent need for efficient and cost-effective catalysts for water splitting, a key technology for green hydrogen production, is hampered by the slow and expensive process of materials discovery. Current state-of-the-art OER catalysts, like IrO₂ and RuO₂, are rare and expensive. While spinel oxides offer a promising avenue due to their earth-abundant nature and tunable properties, the vast chemical space of possible spinel compositions makes experimental exploration infeasible. This paper aims to bridge this gap by developing a computational approach to rapidly screen and identify high-performing spinel OER catalysts.

Methods

The researchers developed a multi-stage computational workflow:

  1. Spinel Library Creation: They generated a database of 6,155 ternary-phase spinel oxides (AxB₃₋ₓO₄) by considering 52 elements and three different metal ratios (x=0.5, 1, 1.5), including both normal and inverse spinel structures.
  2. Synthesizability Prediction: The thermodynamic stability of these compounds was assessed using a convex hull analysis, filtering out unstable materials.
  3. Performance Evaluation: The theoretical overpotential for OER on the surfaces of the stable spinels was calculated using the computational hydrogen electrode method, which involves determining the binding energies of key intermediates (O, OH, and *OOH).
  4. Machine Learning Model: A random forest regression model was trained using DFT-calculated electronic structure descriptors (e.g., Bader charges, band centers) to predict O, OH, and *OOH binding energies, enabling a deeper understanding of structure-activity relationships.
  5. Synthesis and Characterization: Promising candidates from the computational screen were synthesized (via electrodeposition) and characterized using techniques like SEM, EDS, XRD, and XPS to confirm their structure and composition.
  6. Electrocatalytic Testing: The synthesized catalysts were evaluated for OER activity using linear sweep voltammetry (LSV) in a 1 M KOH solution.

Datasets

  • In-house Dataset: A computational database of 6,155 ternary-phase spinel oxides was generated for the high-throughput screening.
  • External Databases: The Open Quantum Materials Database (OQMD) was used to construct the convex hull for thermodynamic stability assessment.
  • Literature Data: For benchmarking, the performance of the synthesized catalysts was compared against 38 other state-of-the-art OER catalysts reported in the literature.

Key Insights and Ideas

  • Computational Screening as a Catalyst Discovery Engine: The study demonstrates the power of high-throughput computational screening to efficiently navigate vast chemical spaces and identify promising novel materials for challenging applications like OER.
  • Spinel Oxides as a Promising Class: The research reinforces the potential of earth-abundant spinel oxides as cost-effective alternatives to precious metal catalysts for OER.
  • Electronic Structure Descriptors are Key: The machine learning model highlights the importance of electronic structure descriptors, particularly the Bader charge and O₂ₚ bandcenter of adsorbate oxygen, in predicting OER activity. This provides valuable insights for future catalyst design.
  • Surface Effects Matter: The study acknowledges that while bulk descriptors are useful, surface effects (like coordinatively unsaturated sites) can significantly influence OER activity, underscoring the need for surface-sensitive computational approaches.
  • Co-based Spinels Dominate: Cobalt appears to be a particularly effective element for OER-active spinel catalysts, forming the largest number of low-overpotential candidates.

Key Results

  • 33 Promising Candidates: Out of 6,155 screened ternary spinel oxides, 33 were predicted to have theoretical overpotentials below 0.5 V.
  • Novel Catalyst Discovery: Co₂.₅Ga₀.₅O₄ was synthesized and demonstrated competitive OER performance with a low overpotential of 220 mV at 10 mA cm⁻² and a Tafel slope of 56.0 mV dec⁻¹.
  • Machine Learning Model Performance: The random forest model achieved an R² score of 0.83 for the training set and 0.78 for the test set in predicting O, OH, and *OOH binding energies.
  • Benchmarking Success: The synthesized Co₂.₅Ga₀.₅O₄ outperformed many existing Co, spinel oxide, and Ru catalysts in terms of overpotential and Tafel slope.

Limitations and Future Work

  • Thermodynamic Stability vs. Kinetic Stability: The study primarily focused on thermodynamic stability. Future work could explore kinetic stability and defect engineering (e.g., oxygen vacancies) to further refine catalyst selection.
  • Machine Learning Model Outliers: The ML model struggled to predict the overpotentials of materials containing less common elements (Nb, Pb, Cd, Ir), likely due to their underrepresentation in the dataset. Further expansion of the training data could improve this.
  • Surface Coverage Assumptions: The assumption of 50% monolayer coverage for adsorbates in the computational model simplifies calculations but may not fully capture real-world conditions.
  • Broader Application: The authors suggest that this computational framework can be adapted to discover novel, high-performing OER catalysts from other material classes beyond spinels.
  • Experimental Validation: While three candidates were synthesized, further experimental validation of the remaining 30 predicted candidates could yield more discoveries.

Summary updated

Abstract

The search for new materials can be laborious and expensive. Given the challenges that mankind faces today concerning the climate change crisis, the need to accelerate materials discovery for applications like water-splitting could be very relevant for a renewable economy. In this work, we introduce a computational framework to predict the activity of oxygen evolution reaction (OER) catalysts, in order to accelerate the discovery of materials that can facilitate water splitting. We use this framework to screen 6155 ternary-phase spinel oxides and have isolated 33 candidates which are predicted to have potentially high OER activity. We have also trained a machine learning model to predict the binding energies of the *O, *OH and *OOH intermediates calculated within this workflow to gain a deeper understanding of the relationship between electronic structure descriptors and OER activity. Out of the 33 candidates predicted to have high OER activity, we have synthesized three compounds and characterized them using linear sweep voltammetry to gauge their performance in OER. From these three catalyst materials, we have identified a new material, Co<sub>2.5</sub>Ga<sub>0.5</sub>O<sub>4</sub>, that is competitive with benchmark OER catalysts in the literature with a low overpotential of 220 mV at 10 mA cm<sup>-2</sup> and a Tafel slope at 56.0 mV dec<sup>-1</sup>. Given the vast size of chemical space as well as the success of this technique to date, we believe that further application of this computational framework based on the high-throughput virtual screening of materials can lead to the discovery of additional novel, high-performing OER catalysts.
BibTeX
@inproceedings{kumar2024from,
  author = {Sai Govind Hari Kumar and Carlota Bozal‐Ginesta and Ning Wang and Jehad Abed and Chung Hsuan Shan and Zhenpeng Yao and Alán Aspuru‐Guzik},
  title = {{From computational screening to the synthesis of a promising OER catalyst}},
  booktitle = {Chemical},
  year = {2024},
  doi = {10.1039/d4sc00192c},
}
Figure