Quaternary mixture designs applied to the development of multi-element oxygen electrocatalysts based on the Ln0.58Sr0.4Fe0.8Co0.2O3−δ system (Ln = La1−x−y−zPrxSmyBaz): Predictive modeling approaches

The experimental data generated through the optimization of oxygen electrocatalysts based on the perovskite Ln0.58Sr0.4Fe0.8Co0.2O3−δ system (Ln = La1−x−y−zPrxSmyBaz) have been modeled following different approaches. The main application of these catalysts is as fuel cell (SOFC) cathodes and activation layers on oxygen-transport membranes. Among the different La, Pr and Sm combinations, those containing at a time Sm–La–Ba or alternatively Pr–La–Ba show the lowest polarization resistance values. Within the same substitution degree, Pr–Ba-based compositions have lower electrode resistance than samarium-based ones. The experimental datasets available for the series of materials can be divided into: composition data, structural data (X-ray diffraction patterns), and electrochemical characterization data (electrochemical impedance spectra). Electrochemical characterization was performed for each electrode composition as a function of the operating temperature and oxygen partial pressure. Different ways of reducing the dimensionality of the spectral descriptors (XRD patterns and impedance spectroscopy) were applied based on knowledge-guided and unsupervised approaches. Different material descriptors were studied as input variables in the modeling of the electrochemical properties.