As a critical industrial gas, the global oxygen market is projected to reach $48 billion/year within this decade. Meanwhile, oxygen production is highly energy intensive due to the limited efficiency of the commercial cryogenic air separation technology. With a pressing demand for emission and energy reduction, there is an urgent need to intensify the air separation technology and to increase the separation efficiency. Chemical looping air separation (CLAS), a cyclic redox scheme that relies on a metal oxide for oxygen storage and release, offers a potentially promising route for oxygen production at a higher thermal efficiency. In this method, the PO2 and/or temperature swing between the reduction and oxidation step is what drives the separation. CLAS has an excellent potential to produce oxygen with a lower energy consumption. However, it is imperative to develop advanced oxygen sorbents with suitable redox thermodynamics and fast redox kinetics for improved efficiency and smaller reactor sizes. This study aims to identify new and promising candidates by rationally optimizing mixed oxide sorbents. Using SrFeO3-δ as a base structure, we systematically investigated a wide range of redox potential and temperatures using high throughput density functional theory calculations (DFT), covering 2401 perovskite compositions with a general formula of SrxA1-xFeyB1- yO3-δ. Machine learning (ML) techniques were subsequently applied to fit the DFT results with reasonable accuracy. Using the model, redox properties of 227,273 perovskites containing 5 cation elements were screened, significantly expanding the materials design space. The ML model was then confirmed by both DFT and experimental results. The DFT and ML models not only correctly identified 11 superior oxygen sorbent materials reported in CLAS literature but also predicted many new and unexpected compositions for air separation. Of the predicted materials, 15 were experimentally investigated. 13 out of 15 materials tested exhibited better performance than the baseline SrFeO3 sorbent, indicating the effectiveness of the DFT and ML based high throughput approaches for accelerated materials discovery. In addition, detailed characterizations were performed on a few promising, high entropy perovskite oxide sorbents to determine their redox properties and CLAS performance.

Engineering the oxygen potential of perovskite oxides for chemical looping air separation

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.