Navigating the unknown with AI: multiobjective Bayesian optimization of non-noble acidic OER catalysts

Jenewein, Ken J.; Torresi, Luca; Haghmoradi, Navid; Kormányos, Attila; Friederich, Pascal; Cherevko, Serhiy

doi:10.1039/d3ta06651g

Cited by 13 publications

(4 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These droplet cells find significant application in electrocatalysis, where they facilitate the scanning of various solid electrocatalysts with diverse chemical compositions. [49,50] These electrocatalysts are typically prepared using a custom-programmed pipetting robot [49,50] or, alternatively, through a physical vapor deposition process. [51] Scanning droplet setups are also employed in corrosion research.…”

Section: Experiments Automation and Standardizationmentioning

confidence: 99%

Key Requirements for Advancing Machine Learning Approaches in Single Entity Electrochemistry

SHKIRSKIY,

Kanoufi

2024

Preprint

View full text Add to dashboard Cite

Despite the noteworthy progress in Single Entity Electrochemistry (SEE) in the last decade, the field still must undergo further advancements to attain the requisite maturity for facilitating and propelling machine learning (ML)-based discoveries. This mini-review presents an analysis of the required developments in the domain, using the success of AlphaFold in biology as a benchmark for future progress. The first essential requirement is the creation and support of high-quality, centralized, and open-access databases on the electrochemical properties of single entities. This should be facilitated through the automation and standardization of experiments, promoting high-throughput output and facilitating comparison between datasets. Finally, the creation of a new type of interdisciplinary specialist, trained to pinpoint critical issues in SEE and implement solutions from applied informatics, is vital for ML approaches to flourish in the SEE field.

show abstract

Section: Experiments Automation and Standardizationmentioning

confidence: 99%

Key Requirements for Advancing Machine Learning Approaches in Single Entity Electrochemistry

SHKIRSKIY,

Kanoufi

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…As catalyst behavior also depends on the temperature, a temperature-controlled flow field could improve the cell design. The conventional SFC can be used for automated high-throughput approaches, , so an adaption to the S-GDE’s flow field to analyze multiple catalysts would be beneficial for accelerating catalyst layer testing aimed at high-throughput analysis.…”

Section: Conclusion and Outlookmentioning

confidence: 99%

Scanning Gas Diffusion Electrode Setup for Real-Time Analysis of Catalyst Layers

Reichmann,

Lloret,

Ehelebe

et al. 2024

ACS Meas. Sci. Au

View full text Add to dashboard Cite

The scanning gas diffusion electrode (S-GDE) half-cell is introduced as a new tool to improve the evaluation of electrodes used in electrochemical energy conversion technologies. It allows both fast screening and fundamental studies of real catalyst layers by applying coupled mass spectrometry techniques such as inductively coupled plasma mass spectrometry and online gas mass spectrometry. Hence, the proposed setup overcomes the limitations of aqueous model systems and full cell-level studies, bridging the gap between the two approaches. In this proof-of-concept work, standard fuel cell electrodes are investigated at elevated oxygen reduction reaction current densities, while dissolved Pt x+ ions in the electrolyte and gaseous CO 2 in the outlet gas stream are detected to track platinum dissolution and carbon corrosion, respectively. Relevant current densities of up to 0.75 A cm −2 are demonstrated. The electrochemically active surface area, oxygen reduction reaction activity, and Pt dissolution rates are quantified and benchmarked to the values obtained in the conventional stationary GDE half-cell. Moreover, it is found that Pt dissolution is suppressed when O 2 is purged into the catalyst layer. Overall, this work demonstrates the feasibility of fast fuel cell electrode screening obtaining, complementary to electrochemical, mass spectrometry data necessary in fundamental studies on structure/ performance relationships under actual reaction conditions. While Pt/C, in relevance to its fuel cell application, is used in this study, the proposed setup can be applied in water electrolysis, CO 2 conversion, metal-air batteries, and other neighbor technologies.

show abstract

“…The Cantor alloy electrocatalyst surfaces completely lose OER activity after prolonged cycling (∼2000 CV cycles), exhibiting pseudocapacitor characteristics (Figure 1a), which was also observed for MnTiO x for the acidic OER. 39 Previous studies on extraordinarily stable HEA electrocatalysts 8,27,34 typically measure the changes in overpotentials or perform chronoamperometry or chronopotentiometry measurements to examine stability. In our study, we observed that the decrease in the level of the OER activity in the Cantor alloy is relatively slow compared to that of unary metals, for example, Co. 17 The increase in overpotential of the Cantor alloy microelectrode from ∼360 to ∼380 mV took about 3.5 h of the OER cycling (∼500 cycles, Figure 1b).…”

Section: Dissolution Measurements Of the Cantor Alloy During The Oermentioning

confidence: 99%

Revealing Dynamic Surface and Subsurface Reconstruction of High-Entropy Alloy Electrocatalysts during the Oxygen Evolution Reaction at the Atomic Scale

Luan,

Escalera-López,

Hagemann

et al. 2024

ACS Catal.

View full text Add to dashboard Cite

High-entropy alloys (HEAs) represent a promising material systems in the search for next-generation high-performance oxygen evolution reaction (OER) electrocatalysts. Developing HEA electrocatalysts requires a thorough understanding of surface and subsurface reconstruction during the OER and their effects on activity and stability. However, it is difficult for most characterization techniques to resolve the atomic-scale elemental distribution of multiple elements and their surface composition. Herein, we combine atom probe tomography and transmission electron microscopy with online and offline inductively coupled plasma mass spectrometry to unveil the surface and subsurface reconstructions of a model CrMnFeCoNi Cantor alloy electrocatalyst during the OER. We reveal that the Cantor alloy suffers from dissolution once in contact with the electrolyte, whereby the surface is deficit in Cr, Mn, and Co before the OER. At the onset of the OER cycling, the Cantor alloy surface is activated by forming an ∼5 nm amorphous NiFe-rich (oxy)hydroxide. Below the surface (oxy)hydroxide, an ∼3 nm Cr-rich oxide layer and an oxygen-rich layer (∼3–5 nm) are formed in the subsurfaces, which impede the outward diffusion of Mn, Fe, Co, and Ni to the surface, retarding their continuous dissolution from the bulk, while Cr dissolution occurs steadily. As the OER proceeds, the Cr-rich oxide layer collapses due to steady Cr dissolution, which results in the exfoliation of the amorphous (oxy)hydroxide layer. Simultaneously, the fastest diffusing Mn segregates to the surface, forming a Mn-rich oxide, which leads to a deterioration in the OER activity. Our atomic-scale data advance the fundamental understanding of how concerted thermodynamically and kinetically driven elementary processes occur for different elements in HEAs during the OER. More importantly, our study highlights the importance of establishing the structure–activity–stability correlations of HEA electrocatalysts during the OER in order to validate the hypothesis of synergistic effects in enhancing their activity and stability.

show abstract

Navigating the unknown with AI: multiobjective Bayesian optimization of non-noble acidic OER catalysts

Abstract: Experimental catalyst optimization is plagued by slow and laborious efforts. Finding innovative materials is key to advancing research areas for sustainable energy conversion, such as electrocatalysis. Artificial intelligence (AI)-guided optimization...

Cited by 13 publications

References 69 publications

Key Requirements for Advancing Machine Learning Approaches in Single Entity Electrochemistry

Key Requirements for Advancing Machine Learning Approaches in Single Entity Electrochemistry

Scanning Gas Diffusion Electrode Setup for Real-Time Analysis of Catalyst Layers

Revealing Dynamic Surface and Subsurface Reconstruction of High-Entropy Alloy Electrocatalysts during the Oxygen Evolution Reaction at the Atomic Scale

Contact Info

Product

Resources

About