Qualitative Trends versus Quantitative Accuracy: A Case Study of LiMn1.5Ni0.5O4 Cathode Material †

Abstract

LiMn_1.5Ni_0.5O₄ is a high-voltage cathode material for Li-ion batteries. The effects of Ni/Mn ordering and nickel content on its electrochemical behavior remain ambiguous. DFT can provide essential insights into these phenomena. We benchmark four exchange-correlation functionals (PBE, PBE+U, r²SCAN, and HSE06) for Li_xMn_1.5Ni_0.5O₄ (x = 1, 0.5, 0) against experimental data. We evaluate key properties: operating voltage, voltage differences between Ni³⁺/Ni²⁺ and Ni⁴⁺/Ni³⁺ redox couples, magnetic and crystallographic structures, structural response upon delithiation, and oxygen K-edge EELS spectra. PBE underestimates the average voltage (3.94 V vs. 3.73 V) and overestimates redox voltage differences (114 mV vs. 15 mV), but captures essential trends in structural and electronic properties at moderate computational cost. While PBE+U improves voltage accuracy, it predicts an unstable half-lithiated phase and incorrect magnetic configurations. r²SCAN yields mixed results. HSE06 provides the highest accuracy for some results but at significant computational expense. PBE emerges as the best compromise for modeling Li_xMn_1.5Ni_0.5O₄ phases.