Fe3+ reduction during melt‐synthesis of LiFePO4. (17th July 2019)

• Record Type: Journal Article
• Authors: Sauriol, Pierre; Li, Delin; Hadidi, Lida; Villazon, Hernando; Jin, Liling; Yari, Bahman; Gauthier, Michel; Dollé, Mickaël; Chartrand, Patrice; Kasprzak, Wojciech; Liang, Guoxian; Patience, Gregory S.
• Is Part Of: Canadian journal of chemical engineering. Volume 97:Number 8(2019)
• Page Start: 2196
• Page End: 2210
• Publication Date: 2019-07-17
• Subjects: Fe3+ reduction -- hybrid induction heating -- LiFePO4 -- melt‐synthesis -- scale‐up; Chemical engineering -- Periodicals; Technology -- Periodicals; 660.05
• DOI: 10.1002/cjce.23522
• Languages: English
• ISSNs: 0008-4034

Abstract

Abstract : LiFePO4 (LFP) is a safe and low cost cathode material for Li‐ion batteries. Its solid‐state synthesis requires micron‐sized reactants yielding high production costs. Here, we melt‐synthesized up to 5 kg batches of LFP from low‐cost coarse Fe2 O3 (509 µm) in an induction furnace. Graphite from the crucible was an effective reducing agent. Adding metallic Fe or CO increased the Fe 2+ content and reaction kinetics. Metallic Fe improves the lifetime of the graphite crucible but requires a premixing step for it to be effective, otherwise the Fe powder agglomerates due to the presence of a eutectic in the LiPO3 ‐Fe‐Fe2 O3 system. In a pushout furnace configuration, for an hour‐long holding period, injecting CO into the melt increased the Fe 2+ content from 0.301 to 0.315 g/g, which we attributed to melt protection. Likewise, graphite powder floating on top of the melt further improved the Fe 2+ content to 0.331 g/g. The Fe 2+ content reached 0.325 g/g when using fine Fe 3+ (142 µm) and CO as reducing agent at half the holding period at 1150 °C. We attribute the higher reaction rate to the improved contact between the suspended Fe 3+ and the CO reducing gas. When the graphite crucible is the unique reducing agent, the reaction rate was proportional to the crucible base surface area. A zero‐order kinetic model characterized the solids disappearance with time. A thermal model developed to compare lab‐scale data against small pilot‐scale demonstrated that the charge lagged the furnace temperature by as much as 22 min at 1000 °C.