Effect of the 3D Structure and Grain Boundaries on Lithium Transport in Garnet Solid Electrolytes

Abstract

Lithium metal anodes are vital enablers for high-energy all-solid-state batteries (ASSBs). To promote ASSBs in practical applications, performance limitations such as the high lithium interface resistance and the grain boundary resistance in the solid electrolyte (SE) need to be understood and reduced by optimization of the cell design. In this work, we use our 3D microstructure-resolved simulation approach combined with a modified grain boundary transport model for the SE to shed some light on the aforementioned limitations in garnet ASSBs. Using high-resolution volume images of the SE electrode sample, we are able to reconstruct the SE microstructure. Using a grain segmentation algorithm, we further distinguish individual grains and account for the influence of the SE grain size and grain boundaries. We focus our simulation work on the trilayer cell architecture, consisting of two porous SE electrodes separated by a dense layer. Even though the highly porous SE electrodes reduce the lithium interface resistance by providing a higher active surface area, the increased electrode tortuosity also reduces the effective ionic conductivity in the SE. We confirm via impedance simulation studies and validation against experimental results that with increasing SE electrode porosity, the lithium transport becomes limited by grain boundaries. We also correlate the area-specific resistance to different lithium infiltration stages in the trilayer cell by spatially resolving the current density distribution. This analysis allows us to suggest a plausible deposition mechanism, and moreover, we identify current density hot spots in the proximity of the dense layer. These hot spots might lead to dendrite formation and long-term cell failure. The joint theoretical and experimental study gives guidelines for cell design and optimization which allow further improvement of the trilayer architecture.

Elucidating the Role of Microstructure in Thiophosphate Electrolytes – a Combined Experimental and Theoretical Study of β‐Li3PS4

“…Liu et al claimed that since the nanoporous 𝛽-LPS has a high surface-to-bulk ratio (rodshaped particles between 10 and 30 μm with an average BET specific surface area of 15.6 m 2 g -1 ), surface conduction is dominant. Although a certain amount of porosity appears to have a beneficial effect on ion conduction, only a few papers [21,22,23] have addressed the role of microstructure in thiophosphates, confirming the experimental results of Liu et al [20] In fact, while the role of microstructure and internal interfaces-such as grain boundaries (GBs, a term commonly used to describe interparticle interfaces in polycrystalline samples, as it is the case in this work)-has been extensively studied in the case of oxide SEs, [24,25,26,27,28] this aspect has been largely neglected in sulfides. However, sulfide SEs can suffer from issues at GBs too, e.g., lithium tends to propagate along the GBs resulting in dendrite growth.…”

“…This range was chosen based on previous studies on porous oxide‐based SEs. [ 24 ] Variation of the respective structural parameters in Equation 2 allows a qualitative assessment of the effect of internal particle constriction on the transport resistance. An overview of the resulting effective bulk conductivity of the particles for the given combination of tortuosity and porosity is given in Table S2 (Supporting Information).…”

“…In fact, while the role of microstructure and internal interfaces—such as grain boundaries (GBs, a term commonly used to describe interparticle interfaces in polycrystalline samples, as it is the case in this work)—has been extensively studied in the case of oxide SEs, [ 24 , 25 , 26 , 27 , 28 ] this aspect has been largely neglected in sulfides. However, sulfide SEs can suffer from issues at GBs too, e.g., lithium tends to propagate along the GBs resulting in dendrite growth.…”

Oxide‐Based Solid‐State Batteries: A Perspective on Composite Cathode Architecture

“…[ 19a,20,28 ] Computational techniques have also been adopted to estimate the above‐mentioned effective properties and compare with experiments, [ 29,30 ] which can be broadly divided into the following categories: Image‐based reconstruction of the LLZO microstructure, which uses X‐ray computed tomography, or focused ion beam scanning electron microscopy (FIB‐SEM). [ 26c,27,31 ] Random addition of spheres representing cathode and SE domains. [ 32 ] Discrete element‐based methodologies.…”

“…Mitigation strategies aimed at reducing the contributions from secondary phases and GBs are outlined in detail in Section 2. By avoiding secondary phases and assuming GB contributions similar to those measured in dense pellets [ 27 ] ($$\sigma _{{\rm{LLZO}}}^0 = 8 \times {10^{ - 4}}{\rm{S}}\;{\rm{c}}{{\rm{m}}^{ - 1}}$$, β GB = 0), one can boost the effective ionic conductivity to σ Li,eff = 3.3 × 10 −5 S cm −1 . This optimization results in a jump of the predicted specific energy to more than 200 Wh kg −1 and 300 Wh kg −1 for the electrodes with 50 µm and 100 µm thickness, respectively.…”