Li Ion Transport and Electrochemical Decomposition of Garnet Li7-XLa3Zr2-XTaxO12 Solid Electrolytes

Abstract

Ionic conductivity and chemical stability are two properties that are critical for the successful deployment of solid electrolytes in next-generation Li ion batteries. Among reported candidate materials up to date, garnet Li7-xLa3Zr2-xTaxO12 (LLZrTaO) has been viewed as one of the most promising oxides for solid electrolyte use, capable of reaching conductivity levels that come very close to conventional liquid/polymer electrolytes (~10-3 S/cm). It is also known to possess good stability at the reductive potential regime, technically enabling the use of Li metal anode for high energy density. However, understanding of the Li ion transport and the electrochemical stability at the oxidative potential regime for doped garnet oxides is still limited. In this work, we present our first-principles density functional theory (DFT) calculations and experimental results on LLZrTaO, focusing on two main points: i) the atomistic-level analysis of Li ion transport and ii) investigation of reactivity during battery charge process at the electrolyte-cathode/interphase region. DFT optimizations of doped garnet model cells suggest a trend following Vegard’s law, i.e., a decreasing lattice parameter with an increasing Ta content which is in good agreement with experimental data. By molecular dynamics, bulk Li ionic conductivities for compositional end members LLZrO and LLTaO were estimated to be 1.06 x 10-4 (0.33 eV) and 1.63 x 10-3 S / cm (0.26 eV), respectively. The former agrees well with those in experiments while a difference is noted for the latter, in which the experimental value is lower by three orders of magnitude than our prediction. Possible reasons that could explain for the difference are as follows: i) an ion transport process at high temperatures that differs from that in the low temperature region ii) an unaccounted effect from intentional/unintentional Al doping and Li+/H+ exchange reactions in experiments, and/or ii) finite model size effects inherent in the calculation which prevent the phase-space sampling for superstructures.We analyzed the observed capacity fading during the charge process of an air-isolated Li ion battery with an LLZrTaO solid electrolyte (at x = 0.375 doping level), Li metal anode, and LiFePO4+carbon (LFP+C) active cathode material. Cyclic voltammetry (CV) measurements of assembled Li/LLZrTaO/(LLZrTaO+C) and Li/LLZrTaO/Al cells showed an increase and an absence of oxidation current, respectively. No reduction current is noted in the reverse CV sweep direction which means that the reaction is irreversible. These CV profiles indicate that the garnet solid electrolyte may have been decomposed via a reaction with carbon, forming a defective garnet and carbon-containing oxide products. XRD data for the post-charge solid electrolyte showed no detectable impurity phases which is consistent with the formation of decomposition product(s) that is (are) made up of light elements. Meanwhile, DFT calculations showed an interesting trend in the operating stability of garnet compounds, i.e., the smaller is the relative electronegativity of the framework octahedral cation, the more thermodynamically stable the base garnet compound becomes against carbon reaction; the order of increasing stability is Nb ≈ Ti < Ta < Zr < Hf.