Energetics of Lithium Insertion into Magnetite, Defective Magnetite, and Maghemite

At low concentrations of lithium insertion into inverse spinel magnetite Fe3O4, a phase change to rock-salt-like LixFe3O4 has been observed. We used density-functional-theory-based (DFT-based) calculations to study the structural origins of this phase change, the concentration at which it occurs, the role of iron vacancies, and the stability of the various motifs that form during the electrochemical reduction process in the Li–Fe–O ternary space up to x = 1.33. We compared our results to new experimental measurements of the open circuit voltage for 8–9 nm magnetite particles over a comparable range of lithium insertion. Of the vacant sites in magnetite (16c, 8b, and 48f) lithium insertion was found to be most stable on 16c. Coulomb interactions between the added lithium and iron at the 8a site in magnetite led to substantial displacement of the iron. As further lithium was added, the most energetically favored motif involved lithium clustering in 16c sites around the shifted 8a iron up to a total of three lithiums. In competition with the lithium clustering motif, lithium insertion could be accompanied by the full displacement of all 8a iron to 16c sites, to form the rock-salt-like LixFe3O4, saturating at x = 1. The defective rock-salt structure was found to be more stable than the lithium clustering motif for x ≥ 0.5. The rock-salt-like LiFe3O4 was found to be stable in the Li–Fe–O ternary space for a continuous range of Li–Fe organization on the 16c sites, stabilized by Coulomb interactions. For x < 1, neither the lithium clustering motif, nor the defective rock-salt-like structure for LixFe3O4 were stable against phase segregation to LiFe3O4 and Fe3O4. This phase segregation (0 < x < 1) occurred at a predicted voltage of ∼2.0 V. However, when iron vacancies on the 16d site were introduced, lithium insertion to those vacant 16d sites in Fe2.875O4, and γ-Fe2.67O4 (maghemite), resulted in stable intercalated materials at a predicted voltage of ∼3.0 V. Beyond the concentration of such iron vacancy sites, phase segregation was predicted to the rock-salt-like Li1.125Fe2.875O4 and Li1.33Fe2.67O4, again at ∼2.0 V. These results were consistent with measured open circuit voltages. Finally, the relative stability for several lithium compositions along the FeO to LiFeO2 tie line in the defective rock-salt structure suggested stable compound formation for a range of lithium–iron compositions, but without long-range order in the cation sublattice, consistent with what has been commonly observed in the literature.