Perovskite-type solid electrolytes have attracted attention as model materials for solid-state battery materials due to their high ionic conductivity and simple structure. A perovskite, La_0.57Li_0.29TiO₃ (LLTO), shows ionic conductivity of 1×10^-3 Scm^-1 at room temperature. It has a double perovskite structure consisting of a La-rich layer in which La occupies 96% of the A site and a La-poor layer in which La occupies 23% of the A site. Due to its crystal structure, anisotropic lithium conduction, in which lithium ions migrate through the La-poor layer, is expected, but anisotropic ionic conduction is rarely observed even in LLTO single crystals [1, 2]. This is believed to be due to the presence of 90° domains. In addition, the complex hierarchical structure, including 90° domains and grain boundaries, makes it difficult to separate bulk and grain boundary diffusion.

This study focuses on the use of pulsed field gradient nuclear magnetic resonance (PFG-NMR) spectroscopy to elucidate the bulk diffusion mechanism of LLTO. Sample is a commercially available LLTO (Toho Titanium) polycrystalline ceramic. The microstructure of the polycrystals was characterized by SEM, EBSD, and STEM. LLTO polycrystals are consisted of random grains (10-200 μm) and 90° domains (10-100 nm). PFG-NMR were performed using an ECA-400 NMR system with a diffusion probe in the temperature range of 273K to 393K. An ECA-500 spectrometer and a high temperature diffusion probe [3] were used for measurements at higher temperatures. A pulsed gradient-guided echo sequence was used. Typical diffusion time Δ is 10-100 ms [4].

In the case of homogeneous diffusion, the echo attenuation of PFG-NMR follows the Stejskal-Tanner equation and decays linearly with the square of the intensity of the gradient pulse (g). In LLTO, however, the decay is not linear but curvilinear. This curve fits very well in the theoretical equation for randomly oriented two-dimensional crystals, providing evidence for two-dimensional diffusion of LLTO. We found that at 393 K, the diffusion coefficient of LLTO is 6.1 × 10^-7 cm² s^-1 (ab-axis) and 3.3 × 10^-8 cm² s^-1 (c-axis). This diffusion anisotropy was maintained up to 693 K, the highest temperature measured. On the atomic scale, it is suggested that the diffusion path of LLTO may change at high temperatures [5], but on the millisecond scale, the macroscopic anisotropy does not appear to change.

The average diffusion coefficient (D_NMR) is 2/3 of the diffusion coefficient for the ab direction. When compared to the conductivity diffusion coefficient (D_σ) obtained from the bulk ionic conductivity, it agrees well over a wide temperature range in an Arrhenius plot. The ionic conductivity is also an average of the anisotropic conductivity of the LLTO. Non-Arrhenius behavior is observed above 450 K for both D_NMR and D_σ. The origin of the non-Arrhenius behavior is not due to a change in the number of carriers, but to a change in the mobility. In addition, the grain boundary diffusion coefficient can be elucidated by tracer isotope diffusion using secondary ion mass spectrometry (SIMS) [6]. In conclusion, PFG-NMR is a promising method for elucidating the diffusion mechanism of polycrystalline solid electrolytes.