Dresden 2026 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

MM: Fachverband Metall- und Materialphysik

MM 8: Materials for the Storage and Conversion of Energy II

MM 8.9: Vortrag

Montag, 9. März 2026, 18:00–18:15, SCH/A216

LDH Under Stress: Assessing Degradation Pathways of Ni-Fe-V Catalysts Under Technical Operating Conditions — •Juan Manuel Lombardi, Charles Pare, Karsten Reuter, and Christoph Scheurer — Fritz-Haber-Institut der MPG, Berlin

The transition to a sustainable energy landscape relies on electrocatalysts that are not only active and affordable but also structurally robust under operating conditions. Ni-based layered double hydroxides (LDHs) doped with Fe and V are among the most promising OER candidates for anion-exchange membrane water electrolyzers (AEMWE), owing to their tunable lattice chemistry and rich redox behavior. However, the mechanistic influence of these dopants and how they reshape catalytic performance and structural stability remains insufficiently understood due to the immense configurational complexity.

In this contribution, we present an integrated strategy tailored to address this complexity directly. Our evolutionary exploration framework EZGA provides a systematic route through the high-dimensional composition-structure landscape, employing chemically informed operators and explicit diversity control to generate physically meaningful structural candidates. Machine-learning interatomic potentials (MLIPs) enable high-throughput sampling of thermally accessible configurations with near first-principles fidelity, revealing how dopants modulate stability and reactivity. Together, these elements deliver a predictive workflow that reveals how dopants reshape the accessible configurational landscape and provides a mechanistic picture of their influence on stability and reactivity.

Keywords: machine-learning interatomic potentials ( MLIP ); evolutionary materials optimization; NiFeV layered double hydroxides ( LDH ); operando-inspired modeling