Recent advances in in situ and operando characterization techniques for Li₇La₃Zr₂O₁₂-based solid-state lithium batteries

	Fig. 3 (a) Schematic illustration of the in situ OM setup. (b) Schematic diagrams and field effects for in-plane and through-plane configurations. Reproduced from ref. 22 with permission from Elsevier. Schematic diagram of a micro-cell structure in (c) in-plane and (d and e) through-plane configurations for in situ OM viewing. Reproduced from ref. 22, 28 and 29 with permission from Elsevier.

Analogous to normal cells, the one designed for in situ OM visualization is composed of a working electrode, an electrolyte, and a counter electrode. The distinction is that it needs a window on one side of the lens, which is constructed of an optically transparent material such as quartz glass or polypropylene. Fig. 3c illustrates a schematic diagram of a micro-cell structure in an in-plane configuration for in situ OM viewing.²⁸ Two Li electrodes are placed parallel at ∼4 mm distance to the SSE surface. Electron beam evaporation is used to deposit a 100 nm Cu layer as a current collector on an optical glass plate. The glass plate with the Cu layer is covered on the Li electrode to ensure good electronic contact. The micro-cell is sealed with epoxy resin. The two electrodes are connected to an electrochemical workstation for charging and discharging. OM in real-time observes the area between the two electrodes. Fig. 3d displays a schematic diagram of a micro-cell structure in a through-plane configuration.²⁹ The SSE is sandwiched between the Li electrode and the electrode to be observed. Ni screens are placed on the Li electrode side to buffer the applied stacking pressure. Also, two electrodes are connected to an electrochemical workstation for charging and discharging tests. Fig. 3e presents a schematic diagram of another micro-cell structure in a through-plane configuration.²² Minor distinctions are that areas apart from the active LLZO surface are masked with Kapton tape to prevent air contamination and Ni foil is adhered to both sides. Notably, the materials for the OM test are transparent or translucent. Therefore, OM is a highly suitable technique for in situ analysis.

In situ OM has been used to view the Li dendrite growth in LLZO-based SSLBs. For LLZO pellets, Guo et al.²⁸ first prepared transparent Ta-doped LLZO (LLZTO) pellets, and then viewed the Li growth behavior inside the cell directly by in situ OM. As shown in Fig. 4a, Li dendrites always penetrate from one electrode to the other at low currents (100 μA) or high currents (500 μA). The penetration of Li dendrites changes from a single direction to multiple directions with increasing current to maintain a large exchange current at the interface. Accordingly, the shape of Li dendrites varied greatly at different currents. The Li dendrites formed at high currents were similar to those in liquid cells. In addition, during the plating/stripping process, dead Li appeared inside LLZTO. In subsequent cycles, Li was preferentially deposited in dead Li regions.

	Fig. 4 (a) Li dendrite evolution at 100 μA and 500 μA. The scale bar is 1 mm. Reproduced from ref. 28 with permission from Elsevier. (b) Schematic diagram for different types of morphologies of Li penetration. Reproduced from ref. 22 with permission from Elsevier. (c) The optical evolution photographs of bare Li and SbF3@Li during Li deposition at about 4 mA cm−2 for 240 min. Reproduced from ref. 29 with permission from Elsevier.

Kazyak et al.²² revealed that Li penetration includes four morphologies, straight, branching, spalling and diffused (Fig. 4b), indicating that a single mechanism cannot solely explain Li penetration. However, before the cell short-circuit, these morphologies of Li can be reversibly cycled. Moreover, a comparable Li penetration behavior was also found in glassy Li₃PS₄ (LPS) SSE. Furthermore, the Li filament propagation rate was proportional to the current density. Under deep discharge at high current density, void formation, dewetting and reduction of the contact area were seen at the Li electrode. Importantly, as temperature increases, the interface resistance decreases rapidly and the critical current density (the highest current density that the cell can withstand before short-circuiting) increases. Consequently, higher temperatures suppress Li dendrite growth at high current densities.³⁰ The working temperature of the cell is 195 °C which is just above the Li melting point, and Li changes from the solid to molten state. Meanwhile, the mechanical properties of Li change in a step-wise manner, resulting in a step increase in the critical current density. The internal Li anode displays pressure relaxation, suppressing Li dendrite formation.³¹

The Li dendrite growth mechanism was also investigated with in situ OM. Both single crystals and polycrystalline LLZO invariably contain surface defects. Initially, due to surface defects, Li deposits at local locations at the interface between LLZO and Li metal electrodes. Then individual points of the surface experience high local pressure, resulting in mechanical cracking. Li fills the cracks and spreads gradually to form Li filaments, ultimately penetrating the SSE and contributing to a cell short circuit. Furthermore, current is also concentrated on this low resistance path to transport, generating high localised heat. Owing to the Joule heating effect, partial Li filaments are melted into Li spheres that disperse in and outside the SSE.³² Therefore, the failure mechanism of brittle LLZO pellets is mainly Griffith-like rather than the shear-modulus criterion proposed by Monroe and Newman.³³ It is essential to minimize LLZO surface defects to avoid Li dendrite growth. The glassy LPS SSE displays similar Li dendrite penetration mechanisms.³⁴ To alleviate the Li dendrite formation owing to surface defects, Kim et al.³⁵ revealed the function of artificial interlayer modifications via in situ OM. The kinetics of alloying and precipitation reactions using interlayer species remarkably alter Li distribution and deposition morphology. In this way, the interlayer performs a dual role of a dynamic buffer layer for Li redistribution and a matrix layer for facile Li precipitation.

The lithiation/delithiation process of SSLBs constructed with LLZO-based CEs was viewed with in situ OM. In tetraethylene glycol dimethyl ether (G4)-based gel electrolyte solidification, LLZO particles were deposited on the underside forming a hierarchical layered structure by gravity.³⁶ In Li-symmetric cell structures, as Li⁺ was deposited on the underside, Li anodes maintained a smooth surface without dendrite formation over the whole cycling process. This benefit arose from LLZO particle layer protection, resulting in homogeneous Li deposition and inhibition of side reactions at the interface. In contrast, when Li plating is on the top side, the active deposition sites rapidly develop, causing heterogeneous Li deposition. Moreover, a self-adaptive structure electrolyte (SASE) structure can also maintain intimate interfacial contact during cycling.³⁷ It is well-known that Li dendrite growth in cells assembled with poly(ethylene oxide) (PEO)-based polymer electrolytes promptly contributes to cell failure and short circuits. It is also hard to suppress Li dendrite generation by adding LLZO particles. Fortunately, Wang et al.²⁹ sprayed SbF₃ on Li foil surfaces to in situ form a Li/Li₃Sb interlayer, improving the SSE/Li interface. As shown in Fig. 4c, it was revealed by in situ OM that Li was initially deposited in particle interstices of Li/Li₃Sb interlayers, which resulted in higher local current density in interlayers. Subsequently, Li was homogeneously and densely deposited on the interlayer surface, which can be attributed to the certain electronic conductivity, high chemical diffusion coefficient, and high Li absorption energy of Li₃Sb, as well as to the high ionic conductivity of LiF.

Another characteristic of OM that distinguishes it from electron microscopy techniques is its ability to recognize colors in the visible range. This can open up directions for researching specific electrode layers for lithiation, relevant interface reactions and heterogeneous electrochemical reactions via color variations. Cross-sectional imaging with in situ OM was performed to investigate the interface behavior between an active sulphur cathode and CE (LLZTO as a filler, PEO as a polymer and lithium bis(trifluoromethylsulfonyl) imide (LiTFSI) as a Li salt) at 65 °C.³⁸ Originally at an open circuit voltage (OCV) of 2.60 V, the CE remains bright-white when viewed from the optical glass window and on discharging to 2.34 V, the brightness of CE color diminishes markedly. As the potential shifts to 2.25 V, the CE color transforms from white to light brown, and color evolution spreads from the cathode side to the whole CE. Such a behavior was attributed to the polysulfides (PSs) dissolved in CE. In a fully discharged state at 1.50 V, CE color transforms completely to deep brown. Subsequently, on charging back to 2.60 V, the brown CE color is still maintained, indicating that PSs stay in CE instead of returning to the S cathode. The irreversible reaction induces rapid capacity degradation and results in low Coulomb efficiency of SSLBs. The CE suffers severe deformation with continuous cycling, resulting in S cathode fracture.³⁹ In addition, temperature variations strongly influence PS shuttling, CE irreversible volume changes, and Li metal volume expansion.³⁸ Interestingly, CE containing LiTFSI-lithium bis(fluorosulfonyl) imide (LiFSI) binary salts can achieve self-regulation in terms of interface compatibility and side reactions of dissolved PSs.⁴⁰

Nevertheless, in situ optical microscopy (OM) still faces several challenges. The most critical challenge is the limited spatial resolution, which cannot reveal the reaction mechanisms on a nano or an atomic scale. For such high magnification imaging, electron microscopy techniques are employed (discussed in Section 3.2). OM is unable to visualize opaque materials, allowing it to be used mostly in surface/interface research. Encouragingly, optical interferometric scattering microscopy (iSCAT) has been developed as a rapid and low-cost imaging technique and feature positions (such as phase boundaries) with “precision” less than 5 nm can be identified in iSCAT images. Unfortunately, iSCAT has not been utilized for analysing LLZO-based SSLBs. However, scientists have visualized and determined the ion diffusion kinetics in single LiCoO₂ (LCO) particles by iSCAT.⁴¹ During charging and discharging processes, LCO particles showed heterogeneous iSCAT intensity distributions (determined from local dielectric properties of the samples), indicating uneven Li⁺ diffusion and phase transition.⁴²

	Fig. 5 (a) Schematic illustration of Raman spectroscopy studies. (b) Schematic illustration of an in situ Raman testing cell. Reproduced from ref. 45 with permission from Springer. (c) Schematic illustration of an operando Raman microscopy cell for measuring the cross-section of the sample. Reproduced from ref. 46 with permission from the American Chemical Society. (d) Schematic illustration of in situ Raman spectroscopy of a thin-film battery. Reproduced from ref. 47 with permission from Elsevier. (e) Schematic illustration of a 3D-printed in situ symmetric cell device attached with a micro-heater. Reproduced from ref. 48 with permission from the American Chemical Society.

Fig. 5b illustrates a schematic diagram of an in situ Raman spectroscopy cell used to test the sample positioned on the top.⁴⁵ The cell consists of a working electrode, an electrolyte, and a counter electrode. In situ Raman cell design requires an extra special unit, a window, to allow detected light to enter without obstruction. The window, generally a quartz glass, is placed on top of the cell structure and is sealed using glue at the point of contact with an Al plastic package. To observe the top region, a current collector with a small hole or a mesh current collector with pores must be used to expose the observation area. The Al tab and Ni tab are connected to an electrochemical workstation for charging and discharging tests. Fig. 5c shows a schematic diagram of an operando Raman spectroscopy cell for viewing the cross-section of the sample⁴⁶ with an O-ring used to perform sealing. Fig. 5d presents a schematic diagram of an in situ cell for Raman experiments on cathodes in thin film cells (SSE thickness of 3–4 μm).⁴⁷ Indium tin oxide (ITO) electrodes can be utilized as current collectors due to their transparency to laser light. As electrodes undergo large volume changes during the charging and discharging process, in situ/operando Raman spectroscopy cells should be highly air-tight to ensure a high pressure. In theory, in situ/operando cells should feature a high confining pressure similar to normal cells. Fig. 5e illustrates a 3D-printed in situ symmetric cell device attached to a micro-heater.⁴⁸ For Raman microscopy, a flat observation area is necessary to obtain clear images of the samples. Thicker samples can be finely polished with SiC paper, while thinner samples or composite electrodes are needed for Ar ion milling.⁴⁹ When conducting ion milling, the power of ion beam should be controlled to prevent damage to the samples. Therefore, electrodes, electrolytes, and the interfaces between them are accessible for investigation using Raman spectroscopy.

Initially, the annealing process of LLZO films was studied with in situ Raman spectroscopy. The Raman profiles indicate that the temperature range of 23°−700 °C was adequate for cubic LLZO generation since this system benefits from Li₃N Li reservoirs that compensate for Li loss during low-temperature vacuum deposition.¹³ However, high background noise levels result owing to the long distance between the sample in the heater and the Raman microscope since only 50× lenses are used in this setup. It is appropriate to investigate the synthesis temperature of cubic LLZO utilizing in situ Raman spectroscopy. As shown in Fig. 6a, LLZTO pellets after cycling exhibit compressive stress regions and stress-free regions (originally tensile stress regions) when viewed by microscopic two-dimensional (2D) and 3D stress mapping methods utilizing Raman spectroscopy.⁵⁰ The high tensile stress region with high overpotential induces Li deposition preferentially. Excess Li is deposited along the grain boundaries in LLZTO pellets and accumulates progressively, contributing to the emergence of compressive stress regions. While the development of stress-free regions is attributed to cracks inside LLZTO pellets and voids at interfaces, Li deposition along cracks and grain boundaries connects through the cathode and anode, resulting in cell short-circuits. Introducing an artificial interlayer at the SSE/Li anode interface can effectively homogenize Li⁺ deposition and suppress Li dendrite formation.⁹In situ Raman spectroscopy revealed that h-BN interlayer-modified LLZO-based SSLBs produce neither structural nor chemical variations during cycling.⁴⁸ In contrast, unmodified LLZO-based SSLBs display structural phase transition from cubic to tetragonal LLZO, leading to enhanced interfacial resistance and cell degradation. It is well known that the Li₂CO₃ layer is generated rapidly on the LLZO pellet surface in air.⁵¹ However, the quantitative analysis of the Li₂CO₃ generation rate has not been adequately conducted. Thus semi-quantitative rate analysis using in situ Raman spectroscopy can be an excellent solution.

	Fig. 6 (a) 2D and 3D stress mapping for LLZTO before and after electrochemical cycling. Reproduced from ref. 50 with permission from Elsevier. (b) In situ Raman spectra of I2@KB cathode at different discharge and charge states for the first (left) and second (right) cycle. Reproduced from ref. 45 with permission from Springer.

The charging and discharging processes of CE-assembled SSLBs have been further researched with in situ Raman spectroscopy. In Li-LiFePO₄ (LFP) cell systems, LLZTO encapsulated in polyacrylonitrile (PAN) eliminates decomposition of succinonitrile (SN) additives effectively.⁵² Conversely, in unencapsulated systems, SN decomposes severely and spreads throughout CE at higher voltages (>3.5 V), which generates a thick Li⁺-blocking layer (CN–C layer) at the interface. In Li–S cell systems, the re-emergence of S₈ or S₈²⁻ Raman peaks at 151, 217 and 471 cm⁻¹ at 2.65 V confirms the occurrence of a reversible reaction mechanism.³⁹ Notably, peak intensity post-cycling is weaker than that seen pre-cycling, indicating that S cathodes form PSs partially and dissolve in CE (as verified by in situ OM). In Li–I₂ cell systems, the I₃⁻ Raman peak at 116 cm⁻¹ apparently disappears in the first cycle, while the I₅⁻ Raman peak at 162 cm⁻¹ becomes visible (Fig. 6b, left). This phenomenon suggests that the first cycle is irreversible while the in situ Raman spectroscopy study of the second cycle shows reversible conversion (Fig. 6b, right). Accordingly, a two-step reaction mechanism in all-solid-state Li–I₂ cells with a cutoff voltage of 4.0 V has been proposed: (1) 3I₅⁻ + 2e⁻ ⇌ 5I₃⁻ and (2) 5I₃⁻ + 10e⁻ ⇌ 15I⁻.⁴⁵ Overall, these results show that studying the charge and discharge mechanism in a specific cell is practical using in situ Raman spectroscopy.

Besides LLZO-based SSEs, the sulfide SSE family has also been studied by in situ Raman spectroscopy. In Li |LPS| LCO cells, an unreacted LCO region occurs in the composite cathode.⁵³ The poor contact between SSE and cathode active material generates low utilization and a heterogeneous distribution. Hence, finding the right strategy to ensure perfect contact between SSE and cathode active material is necessary. Moreover, the c-axis lattice parameter of LCO increases during cycling via peak intensity evolution from Raman spectroscopy.⁴⁷ Importantly, Li⁺ diffusion in the cathode requires time, which results in a time-delay between electrochemical charge/discharge and Raman measurements. This is a difficult issue to avoid and address currently. In Se |LPS| C systems, Raman spectroscopy reveals the formation of a PS_4−xSe_x³ interface phase.⁵⁴ During the total charging and discharging process, the electrochemical reactions of Se_n chains and PS_4−xSe_x³ were discovered to be reversible despite the varying extent of lithiation. In addition, operando Raman spectroscopy reveals that the PS₄³⁻ unit emerges inside LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) particles in NCM811 |Li₆PS₅Cl (LPSCl)| Li cells during long term cycling, suggesting SSE decomposition and diffusion and cathode particle deterioration.⁵⁵

As discussed above, most research studies have focused on varying ion concentrations. However, conventional Raman spectroscopy features low spatial and temporal resolution, resulting in real-time failure to monitor ion concentrations.⁵⁶ Stimulated Raman scattering (SRS), a nonlinear Raman technique, improves signal resolution and accuracy. Due to the utilization of two types of spatiotemporally synchronized picosecond laser pulse sequences, such technique achieves an effect 10⁸ times stronger than conventional Raman spectroscopy.⁵⁷In situ SRS successfully studied the variation of Li⁺ concentration and dendrite growth on the Li metal surface in liquid cells, allowed the 3D visualization of Li deposition and revealed the Li deposition procedure.⁵⁸ Regrettably, in situ SRS is seldom used for LLZO-based SSLBs. Researchers are expected to utilize in situ SRS to further reveal the Li deposition procedure in LLZO-based SSLBs for optimal interface engineering. Besides low spatial and temporal resolution, low signal accuracy is an obvious weakness of conventional Raman spectroscopy. Micro-Raman, surface-enhanced Raman, and confocal Raman spectroscopy can enhance signal accuracy and it is expected that all these techniques will be adopted for LLZO-based SSLBs investigation in the future.

FTIR spectroscopy is a vibrational spectroscopy technique. Over a wide spectral range (14300 to 20 cm⁻¹), dipole movements are monitored by exciting molecular vibration, rotation, and lattice modes based on the absorption of infrared beams.⁵⁹ The resulting absorption peaks correspond to various modes of the sample, which provide information on the composition and structure of molecules.⁶⁰ FTIR spectroscopy is characterized by high detection sensitivity, high resolution, high measurement accuracy, and fast measurement speeds and it is ideally advantageous for determining the structure and functional groups of polymers and organic compounds. Therefore, in situ FTIR is vital for investigating electrolyte decomposition and solid electrolyte interphase (SEI) film formation in conventional Li-ion and nascent lithium batteries.⁶¹ Surprisingly, in situ FTIR is seldom used in the LLZO-based SSLB research area.

On one hand, the interface between the LLZO pellet and cathode/anode is mainly modified using inorganic materials while LLZO pellets are also inorganic in nature. In situ FTIR fails to achieve a desirable result due to its preference to analyze organic instead of inorganic materials. On the other hand, CE studies mainly concentrate on modulating the species, content and morphology of components, while rarely focusing on the CE degradation and SEI film formation at the interface with the cathode/anode. Consequently, researchers seldom use in situ FTIR characterization. Furthermore, FTIR spectroscopy is inadequate for investigating Li dendrite nucleation and growth. As organic modification of interfaces and degradation of CE will become increasingly researched in the future, in situ FTIR will become critical. The cell structure for in situ FTIR characterization is probably similar to that in in situ Raman spectroscopy. According to the experience of liquid cell research, FTIR is susceptible to O₂ and CO₂ reactions, resulting in low reliability of outcomes. Several infrared spectroscopy modes, including diffuse reflectance infrared transform spectroscopy (DRIFTS), double modulation FTIR (DM-FTIR), subtractive normalized interfacial FTIR (SNI-FTIR), photoacoustic infrared spectroscopy, and transmission infrared spectroscopy will be implemented in future SSLB research studies.⁵⁹

	Fig. 7 (a) Schematic diagram for SEM studies. (b) Schematic diagram of a plan view in situ SEM cell structure. Reproduced from ref. 63 with permission from Springer. (c) Schematic diagram of the in situ SEM setup which is typically equipped with an FIB system. Reproduced from ref. 64 with permission from Wiley. (d) Schematic diagram of an SEM experimental setup equipped with a microelectrode module and two micromanipulators. Reproduced from ref. 62 with permission from Elsevier. (e) Schematic diagram of a cross-section view in situ SEM cell structure. Reproduced from ref. 65 with permission from Wiley.

Distinct in situ cell setups inside the sample room allow different viewing perspectives resulting in variable information. Fig. 7b illustrates a schematic diagram of a plan view in situ SEM cell structure.⁶³ 34 μm Li metal electrodes are pressed on both sides of the electrolyte. A smaller Li electrode is utilized on the top to induce dendrite growth to have edge effects. A Cu spring is employed as a contact electrode and to apply pressure on Li films to press them against LLZO electrolyte to achieve good contact. The lower Li electrode, as a contact electrode and with the same area as LLZO, is rested on a flat Al sample holder that pushes the lower Li electrode on the LLZO electrolyte. The electrode is connected to a cycler using an electrical feedthrough installed in one SEM port. The top Li electrodes can be substituted with Au or Pt electrodes.^14,64 As shown in Fig. 7c, to observe correlated information inside electrolytes during electrochemical cycles, the in situ SEM setup is typically equipped with focused ion beam (FIB) technique for milling the surface.⁶⁴ In addition, Fig. 7d shows an SEM experimental setup equipped with a microelectrode module and two micromanipulators.⁶² The micromanipulator attached to a tungsten needle is directly placed on the LLZO pellet surface as a working electrode, while that attached to Li metal acts as a counter electrode. The micromanipulator enables external operation in x, y and z directions under high vacuum conditions using a control unit. The electrically-shielded connections for microelectrode and micromanipulator control units are guided through a channel outside of the SEM chamber. Adopting a comparable SEM experimental setup, Fig. 7e shows a schematic diagram of a cross-sectional view of the in situ SEM cell configuration.⁶⁵ The LLZO pellet is fractured manually into two semicircular pieces; one piece is placed vertically in a homemade sample holder and fixed with an electrically conductive grub screw. One side of the fractured LLZO pellet is pasted with a Li electrode acting as a counter electrode and a reference electrode, which is electronically contacted via a grub screw. Another side is directly in contact with the Al frame. A Li|LLZO half-cell is fixed in a customized sample holder for real time visualization. Notably, another side of the fractured LLZO pellet can be attached with an In electrode or Si electrode for research purposes.^66,67

Since the experimental effect of electron beam on Li (as compared to Na/K) is relatively low, in situ SEM technique is frequently adopted to characterize the growth morphology and formation mechanism of Li dendrites on a micro-scale. At first, in situ SEM was used to view the LLZO/Li interface in Li symmetric cells during cycling.⁶³ The results suggest that the LLZO/Li interface characteristics are crucial for Li dendrite growth. At the interface, uneven dissolution of Li causes inhomogeneous current distribution and Li deposition. The generated Li dendrites have three morphologies: mossy, needle, and bump (Fig. 8a). Both bump and mossy features were seen mostly in thin Li metal areas due to the low exerted stress. Moreover, combining SEM and EDS data revealed that the formed Li dendrites contain C and O, which is probably related to Li₂CO₃, Li_xC_y and Li₂O. An in situ SEM device with micromanipulators was adopted to conduct and analyse electrodeposition experiments.⁶² Such a device reveals the inhomogeneous electrodeposition behavior and Li dendrite and whisker growth (Fig. 8b), which were attributed to the microstructural variations (including one- and two-dimensional defects⁶⁸) of the LLZO substrate. In theory, a stable and clean LLZO/Li interface exhibits ultrafast charge transfer kinetics (<10⁻¹ Ω cm²).⁶⁵ Ideally, the transport limitations derived from vacancy injection and diffusion are the exclusive kinetic parameters fundamentally restricting the deposition rate capability. However, inherent defects such as porosity, grain boundaries and impurities are inevitable in the practical polycrystalline SSEs. These defects results in rapid Li deposition kinetics and high nucleation trends, resulting in non-uniform Li deposition.¹⁴ In general, short, and thick whiskers are developed at low current densities, while long and thin structures are seen at high current densities. These phenomena reflect that the surface morphology, current density, and surface crystal nucleation kinetics completely control Li's nucleation and deposition behavior. Therefore, minimizing inherent defects to achieve a clean and defect-free interface can prevent the development of local nucleation hotspots. In addition, during the preparation of LLZO pellets, it is extremely probable that microcracks are introduced on the surface. As current is applied, inhomogeneous electric field intensity will be generated around the microcracks. As shown in Fig. 8c, cross-sectional in situ SEM reveals that Li dendrites grow along the cracks and ultimately penetrate LLZO.⁶⁶

	Fig. 8 (a) SEM images of the cell at various cycle times. Reproduced from ref. 63 with permission from Springer. (b) SEM images of dendritic-like and whisker-like Li-metal deposition. Reproduced from ref. 62 with permission from Elsevier. (c) SEM images during the cycles. The scale bars are 200 μm. Reproduced from ref. 66 with permission from Elsevier. (d) Time-lapse images showing the formation of a bowl-shaped crack. Reproduced from ref. 64 with permission from Wiley.

Furthermore, an in situ SEM device equipped with an FIB system revealed that Li propagates predominantly along transgranular cracks.⁶⁴ The crack propagation initiates from the interior of LLZO beneath the electrode surface and then propagates by curving toward the surface, forming a bowl-shaped crack (Fig. 8d). The resulting crack resembles hydraulic cracks caused by high fluid pressure, indicating that the Li deposition-induced inner stress is the major driving force for crack initiation and propagation. Chemical etching or laser treatment of SSE surfaces can realize crack-free surfaces, resulting in homogeneous electric field.

In fact, the pure LLZO pellet surface is not completely clean and defect-free. Hence, an artificial interlayer is introduced to achieve homogeneous Li deposition. The capability of the interlayer to suppress Li dendrites was directly visualized by in situ SEM.⁶² The Au layer can be alloyed with Li, and because of this, Li metal penetration is delayed. However, after generation of the alloy phase, Li metal still nucleates and penetrates at the interface. Due to the inhomogeneous plating characteristics and the high nucleation overpotential of Cu layers, Li metal penetrates immediately. Therefore, the alloyed interlayer will not fundamentally change the kinetics of the Li anode. Li nucleation will not occur if the alloy's vacancy diffusion coefficient can enable Li fluxes to exceed the external applied current density. It is advantageous to introduce a Li metal reservoir layer or an alloy layer with fast Li diffusion properties at the LLZO/Li interface to assemble LLZO-based SSLBs. Moreover, an electronically insulating LiF layer effectively improves the physical interface contact and suppresses the Li dendrite formation, as revealed by in situ SEM.⁶⁹ Serial block-face scanning electron microscopy (SBFSEM) can image the whole bulk material morphology especially the interior. Imaging of Li and g-C₃N₄ composite anodes by SBFSEM revealed that Li₃N is enriched on the surface.⁷⁰ It is comparable to introducing a Li₃N layer at the LLZO/Li interface, resulting in homogeneous Li deposition. Analogously, introducing a C₃N₄ interlayer at the Li_1.5Al_0.5Ge_1.5(PO₄)₃ (LAGP)/Li interface not only homogenizes the Li⁺ flux and alleviates the stress induced by uneven Li deposition, but also restrains the interface side reactions between LAGP and Li.⁷¹In situ SEM demonstrated that the absence of C₃N₄ interlayer causes successive generation of side reaction layers and mechanical cracks at the interface. In general, utilization of non-Li anodes essentially suppresses Li dendrite formation. The interface compatibility and stability between Si anodes with various thicknesses and LLZO was investigated via in situ SEM.⁶⁷ Excellent interfacial contact and electrochemical performance were realized for Si anodes with thicknesses less than 180 nm. Unfortunately, as the Si anode thickness increases, the cell performance deteriorates. Recently, the in situ SEM technique has also been critical in cathode failure research. It is verified that the Li₃BO₃ (LBO) and LCO interfaces in the composite cathode are cracked as the cycle evolution occurs, resulting in higher interface resistance and cell degradation.⁷²

Importantly, adding various equipment in the in situ SEM sample chamber can realize diverse in situ property measurements. For example, coupling a SEM system with mechanical test systems co-located in a large vacuum chamber can help conduct in situ experiments on mechanical properties.⁷³ Based on this system, the variability of mechanical properties during Li dendrite growth can be adequately investigated. When an organic solvent (e.g., CE systems or interface modifications) or low melting point Na/K metals (e.g., construction of composite anodes) are introduced in the investigation system, environmental SEM (ESEM) is feasible since the sample chamber allows working with wet samples. However, the resolution of ESEM is inferior to that of SEM (∼10 nm at a low vacuum of <10³ Pa). Notably, numerous artefacts, including surface damage and alkali metal melting, are observed to arise at room temperature during cross-section viewing using FIB techniques. For example, SEI films formed in liquid cells are susceptible to high-energy Ga⁺, resulting in incorrect detection of the elemental distribution. To address this phenomenon, cryo-FIB/SEM technique has been successfully applied for spectroscopy and imaging on a nano-scale in liquid cells.⁷⁴ Such a technique can be further employed in LLZO-based SSLB systems. In that case, several surprising results have been seen, such as natural formation of nano-scale interlayers at the interface or on the Li surface that exhibits heterogeneity influencing Li deposition kinetics.

	Fig. 9 (a) Schematic diagram for TEM investigations. (b) Schematic diagram for STM-TEM holder and MEMS-TEM holder. Reproduced from ref. 10 with permission from Elsevier. (c) Schematic illustration of the anode-free Cu|LLZO|Li nanobattery setup for in situ TEM probing. Reproduced from ref. 77 with permission from Springer. (d) Schematic illustration of chip-based holders and batteries, an SEM image of the chip, and the step-by-step schematic illustration of the battery fabrication process. Reproduced from ref. 79 with permission from the American Chemical Society. (e) Configuration of TEM holder and micro-battery. Reproduced from ref. 19 with permission from Springer.

For in situ TEM measurements, it is critical to design miniaturized in situ electrochemical cells because of the demands of high vacuum environments and extremely thin samples.⁵⁹ Various in situ TEM holders have been developed to research electrical, thermal, magnetic, and force properties in various gas and liquid environments. As shown in Fig. 9b, these holders are mainly divided into scanning tunneling microscopy-TEM (STM-TEM) holders and micro-electro-mechanical system-TEM (MEMS-TEM) holders. STM-TEM holders utilize a scanning probe unit at the end of the sample holder and incorporate current–voltage measurements for in situ operation and electrical measurements. Individual nano-wires or nano-particles generally serve as the electrodes under observation, materials with low saturation vapour pressure serve as electrolytes, and cathode materials or Li serve as counter electrodes.⁷⁶ Since the reaction cell in the holder is always open, the electrolyte and electrodes are not sealed, so the electrons directly interact with the observed part. Fig. 9c illustrates a schematic diagram of an in situ cell setup with an STM-TEM holder.⁷⁷ LLZO particles are first distributed on the glass substrate. Then, a Cu rod with Li metal attached is pressed onto the substrate, enabling LLZO particles to be attached and submerged in Li metal. The whole part is mounted on one end of the holder. Li metal adhered to a Cu rod acted as a Li source and LLZO particles semi-submerged in Li metal served as SSEs. The other end of the holder is mounted with a metal probe, namely, Cu, sender W or Cu attached with carbon nanotubes (CNTs) (providing different mechanical constraints), as a working electrode. The working electrode is brought into contact with a LLZO particle to form a solid-state nanocell, realizing a cross-sectional view of the interface. A constant bias is applied to the Li metal to initiate the electrochemical process. Accordingly, STM-TEM holders have the advantages of simple sample preparation and atomic-level spatial resolution. Unsurprisingly, there are several issues as well. First, there is a weak contact between the electrolyte and electrode; second, it is essential to apply a large over-potential for driving Li⁺ transport due to the low ionic conductivity of the electrolyte, and lastly, there is a difference between the working environment and actual batteries and this can lead to incorrect conclusions.

MEMS-TEM holders conduct in situ electrical measurements and other functions by carrying a MEMS chip in the sample holder. They rely on contact between conductive probes and electrical contacts of pre-patterned circuits on silicon wafers. The MEMS chip equipped with a nanometer-thick robust Si₃N₄ window results in a sealed environment.⁷⁸ Nanoparticles or active material lamellae (prepared by FIB) connected to a Pt electrode serve as observed electrodes, while Li or Au is the counter electrode. Fig. 9d illustrates a schematic diagram of the holder, chip, and in situ electrochemical cell.⁷⁹ First, a lamella is cut from a micron-sized LCO particle by using FIB milling. The LCO lamella is connected to a gold wire by Pt deposition on the nano-chip. The gold wire is parallel to the [001] direction of the LCO lamella. Subsequently, an LLZO electrolyte slice and Au anode are cut out. A whole cell is formed using a Pt deposition connection. An ion beam is utilized to cut Pt deposition on the electrode in case a short circuit occurs during the charging process. Notably, one or two FIB cuts are conducted on the LCO cathode to thin the sample, which makes an electron-transparent area for in situ TEM observation. The two ends of the microcell are connected to two electrodes in the nano-chip using silver paste (Fig. 9e).¹⁹ Lastly, the nano-chip is placed in the in situ bias sample holder. The two electrodes are connected to an electrochemical workstation via lead wire for charge and discharge tests. Hence, MEMS-TEM holders not only compensate for the disadvantages of STM-TEM holders but also ensure atomic-level resolution. However, in situ electrochemical cells for MEMS-TEM holders are tough to fabricate.

Research on LLZO-based SSLBs has been conducted with a series of in situ electrochemical cells developed using the two holders described previously. When Al-doped LLZO (Al-LLZO) comes in contact with Li metal, an ultrathin (∼6 nm) interphase is viewed depending on the STM-TEM holder.⁸⁰ The interface was observed to be tetragonal LLZO, as confirmed by STEM mode and EELS analyses. This phenomenon was attributed to the local phase conversion that occurred due to the slight reduction of the Al-LLZO surface in contact with Li metal. Notably, the interface was stable and avoided the complete reduction of Al-LLZO by Li metal and enabled Li⁺ transportation. In contrast, cracks, deformation or phase conversion are undetected when Al, Nb co-doped LLZO nanoparticles come into contact with Li metal.⁸¹ In fact, the chemical/electrochemical stability between LLZO and Li metal is dependent on the LLZO-doped elements.⁹ Subsequently, the in situ TEM holder can be used to investigate the evolution of the LLZO/Li metal interface.⁷⁷ Under stronger mechanical restraints and lower charging rates, LLZO particles clearly display a “splitting” or “peeling” form (Fig. 10a). These single-crystal LLZO particles are free of apparent defects, indicating that the local stresses can reach GPa levels during Li “eruption”. Once the mechanical restraints are weakened, LLZO will not be damaged even if the local current density is increased to A cm⁻² level. Therefore, the crack initiation at the interface is associated with the route and efficiency of mass/stress release. Rapid mass/stress release can be realized by either accelerated Li⁺ transport in carbon nanotubes or increasing temperature to facilitate Li⁰ diffusion. Moreover, Li⁺ heterogeneous accumulation at the LLZO/Li interface during the charging process is revealed by EH and EELS.⁸² Even after the discharge process, Li⁺ is seen to be present. Besides the LLZO/Li interface, the interior of LLZO was also studied using in situ TEM. The grain boundaries with narrow band gaps (∼1–3 eV) in LLZO are potential leakage channels.⁸³ When 10 V bias voltage is applied, the brightness of the triple-junction voids progressively weakens to a level comparable with surrounding grains as a function of time. The decrease in brightness is attributed to the filling of Li into voids. As a result, Li⁺ is precociously reduced by electrons to form Li filaments at the grain boundaries, and these Li filaments are ultimately interconnected, causing short circuits. In summary, the electronic conductivity of grain boundaries should not be neglected.

	Fig. 10 (a) Crack initiation and Li penetration in LLZO by Li eruption at the constrained Cu|LLZO interface. Reproduced from ref. 77 with permission from Springer. (b) In situ STEM micrographs of the LCO cathode structure variations. The Li, oxygen, and cobalt ions are green, purple, and cyan balls, respectively. Reproduced from ref. 79 with permission from the American Chemical Society. (c) Changes in the Li maps at 3–18 nA h, B 39–53 nA h, and the open-circuit state for 30 min between the charge and discharge reactions, respectively. The scale bars are 100 nm. Reproduced from ref. 19 with permission from Springer.

MEMS-TEM holders are regularly used to research the cathode material structure evolution in SSLBs during charging and discharging processes. At high voltage delithiation, primary single-crystal LCO cathodes are transformed into nano-sized polycrystalline phases connected by coherent twin boundaries and anti-phase domain boundaries (Fig. 10b).⁷⁹ Such transformations could be preferable for Li⁺ migration. For LiNi_0.5Mn_1.5O₄ (LNMO) cathodes, transition metal ions, in particularly Ni ions, migrate to the 4a sites (16c in the Fdm space group), leading to structural stability reduction.¹⁸ The 〈100〉, 〈110〉, and 〈111〉 zone axes are converted from ordered to disordered structures. The 〈112〉 zone axis produces inhomogeneous structural evolution forming three regions: a transition metal-rich region, an anti-phase boundary region, and a transition metal migration front region. Fortunately, the analysis revealed that the improvement of stability can be realized by doping low-valence cations on an atomic scale. It is expected that the structural transition of commercially available LFP and LiNi_xCo_yMn_1−x−yO₂ (NCM) cathode materials during charging and discharging will also be investigated in future research. The reaction kinetics of Li⁺ in LCO cathodes with LAGP SSE was studied by operando EELS and sparse coding (image de-noising, improved imaging time and spatial resolution).¹⁹ Importantly, Li⁺ migrates along the vertical direction to the LCO/LAGP interface and along the parallel direction (Fig. 10c) and also migrates to the interior of LCO even in the open-circuit state. Moreover, in situ differential phase contrast-STEM (DPC-STEM) technique revealed that discontinuously coating BaTiO₃ (BTO) nanoparticles on LCO particles can suppress space charge layer (SCL) formation and facilitate Li⁺ migration.⁸⁴

Besides imaging at room temperature, several investigations demand in situ imaging at high or low temperature. For instance, the LLZO phase evolution investigation during calcination demands high temperature imaging.⁸⁵ Since Li/Na metal and SEI films are sensitive to extremely high electron doses in high-resolution imaging, imaging at low temperatures is necessary.⁸⁶ Accordingly, it is essential to utilize several heating and/or cooling techniques to realize the desired experimental temperature. Current heating technologies involve furnace-type holders,⁸⁷ MEMS-based heating holders,⁸⁸ filament wire holders,⁸⁹ and laser heating holders.⁹⁰ Unfortunately, these techniques need to address image drift to satisfy the requirement of continuous imaging during heating. MEMS-based heating holders can deliver a drift of less than 1 nm min⁻¹ under optimal working conditions. Based on this, the phase conversion of Ga-doped LLZO (Ga-LLZO) during high-temperature calcination was revealed.⁸⁵ As shown in Fig. 11, initially, LZO is formed according to a 2-to-1 epitaxial growth process along the crystallographic orientations of (−111)_LZO//(−111)_ZrO2 and [211]_LZO//[101]_ZrO2 at 750 °C. Then, Li and Ga are diffused into LZO layer by layer along the [011] orientation to form Ga-LLZO at 900 °C. Cooling technologies include cooling holders (the cold source is liquid nitrogen or liquid helium) and MEMS-based thermoelectric cooling chips (similar to the heating holder).⁹¹ The lowest temperature of cooling holders is as low as liquid nitrogen or liquid helium temperature but fails to stabilize at any temperature in this cooling range. The flow or boiling of liquid nitrogen or liquid helium also introduces vibration that causes drift. MEMS-based thermoelectric cooling chips exhibit the advantages of vibration-free and precise temperature control. Thanks to the cooling techniques, cryo-TEM realizes high-resolution visualization on electron beam-sensitive materials. However, electrochemical studies of in situ cryo-TEM are not available due to space constraints of TEM holders, complexity of chip function, and others. Notably, an initial investigation of LLZO-based SSLBs has been carried out with ex situ cryo-TEM technique.⁹² Results indicate that microcracks which originate from Li dendrite growth are introduced by mechanically polishing the LLZO surface to eliminate the contaminant layer. An interlayer consisting of amorphous carbon and nanocrystalline LiF is formed in situ by the introduction of CF_x and molten Li. Such a layer exhibits superb lithiophobic nature and thus the capability to suppress Li dendrite formation, providing an effective strategy for enabling the use of LLZO-based SSLBs in energy storage applications.

	Fig. 11 In situ atomic-scale investigation of Ga-LLZO Li+-conducting solid electrolyte during calcination-related growth. Reproduced from ref. 85 with permission from Elsevier.

Overall, in situ TEM has been successfully used to investigate the interface, electrolyte and cathode in LLZO-based SSLBs. Notably, several issues remain: (1) in situ electrochemical cells are detached from practical working situations due to the limitation of the sample chamber; (2) extremely thin samples enhance the difficulty of cell preparation; (3) high-vacuum testing conditions increase the difficulty of laboratory manipulation; (4) sensitivity of Li metal to high electron doses produces unreliable results, thus requiring cooling techniques or low electron doses (without sacrificing resolution); and (5) lower imaging speed and higher image drift. It is believed that after addressing these issues, LLZO-based SSLBs can be more thoroughly and comprehensively investigated by utilizing in situ TEM techniques.

	Fig. 12 (a) Schematic diagram for XRD studies. (b) Schematic diagram of an in situ XRD cell. Reproduced from ref. 94 with permission from Wiley. (c) Schematic diagram of an operando XRD cell setup and cell inside a diffractometer. Reproduced from ref. 95 with permission from Elsevier. (d) Schematic diagram of an in situ XRD measure setup for phase conversion analysis during high-temperature heating. Reproduced from ref. 96 with permission from Elsevier.

However, beryllium metal is easily oxidized and produces beryllium oxide peaks that disturb the test results. Kapton films suffer from having lower conductivity although they have no sharp XRD peaks. Al metal windows are not frequently adopted due to strong X-ray absorption and the formation of alloy with Li below 0.45 V. Be metal windows are frequently used in LLZO-based SSLBs systems. The thickness of the window has to be optimal to ensure that the smooth transmission of X-rays is not hindered (high thickness) while ensuring that it is not easily fractured during assembly (low thickness) of in situ cells. If using an SR light source, smaller X-ray-transparent windows can be used and in some cases, a window is not even required. Hence, two conditions should be considered for in situ XRD cell design: first the cell should be simple to assemble/disassemble and reusable and second, an ideal window material should enable X-rays to reach electrodes easily and be stable in cell systems. Fig. 12c illustrates a schematic diagram of an operando XRD cell setup and cell inside a diffractometer for solid-state batteries.⁹⁵ The cell is made of Ti and polyether ketone. A thin Li foil is placed at the bottom and covered with SSE. Then, a self-standing film of electrode material is placed at the top covered with a window (beryllium metal). The window is compressed with a rubber O-ring to seal the cell. The in situ cell structure design discussed above is suitable for the charging and discharging process. However, most current research reports are focused on changes in the LLZO-based SSE bulk or cathode structure during high-temperature heating. Fig. 12d presents a schematic diagram of an in situ XRD measurement setup for phase conversion analysis during high-temperature heating.⁹⁶ A bulk or powder sample is placed in a crucible, which is inserted in a heating chamber. The sample is heated to a specific temperature, and XRD patterns are recorded before moving to the next temperature. The final in situ XRD pattern during heating is acquired. Therefore, during charging and discharging or high-temperature heating, in situ XRD techniques enable real-time visualization of material phase conversion and structural transformation. In addition, the working principle of SRXRD is analogous to that of conventional XRD employing laboratory X-ray sources. It can help determine crystallographic information, including lattice parameters, site occupancy, microstructure and strain/stress of many materials.⁹³ High resolution, robust signal strength and low test time of SR are more favorable for measurement of dilute phase scattering, residual stress analysis, time-resolved XRD, and for in situ/operando XRD.⁹³

The structural evolution of Sn anodes in initial lithiation/delithiation was investigated using operando XRD.⁹⁷ During lithiation, Li–Sn alloy peaks appear, matching the expected Li₂Sn₅ phase. This phase is converted into the LiSn phase before the end of lithiation. In contrast, during delithiation, the LiSn phase disappears and the Li₂Sn₅ phase reappears. However, the latter does not disappear at the end of delithiation, indicating irreversibility of the cycle. The results indicate that the formation of an irreversible Li₂Sn₅ phase during charging and discharging process results in poor electrochemical performance. Currently, in situ XRD is mainly adopted to examine the phase conversion process of LLZO or cathode materials in LLZO-based SSLBs research.⁹⁸ During annealing, the evolution of undoped LLZO, Ga-LLZO and Al-LLZO films is revealed by in situ grazing incidence (GI) XRD.^99,100 The in situ cell employed for GI-XRD is placed under a flowing O₂ atmosphere. As shown in Fig. 13a, an undoped amorphous LLZO film is formed initially as Li₄Zr₃O₈ and Zr₃O at 300 °C, while the tetragonal LLZO phase occurs at 500 °C. At 500°–700 °C, the tetragonal LLZO phase and the cubic LLZO phase are seen to co-exist. As the temperature continuously increases, the cubic LLZO phase gradually becomes dominant. Finally, LLZO is completely converted from the tetragonal phase to the cubic phase at 700 °C. The temperature for complete conversion is lowered to 650 °C through Ga doping.⁹⁹ At 650°–700 °C, Al-LLZO films also fully showed cubic phase formation.¹⁰⁰ Therefore, Li-site element doping effectively decreases the temperature of complete conversion of LLZO from the tetragonal to cubic phase, indicating that doping can stabilize the cubic phase at lower temperatures. Furthermore, phase evolution of LLZO particles during synthesis was also investigated by in situ high-temperature XRD. As shown in Fig. 13b, the calcination process of LLZTO and Ca/W dual-doped LLZO (LLCZWO) are similar.¹⁰¹ The whole procedure consists of three reaction stages, namely, the formation of La(OH)₃ (25° to ∼300 °C), La₂O₂CO₃ with unreacted ZrO₂ (∼300° to ∼760 °C) and cubic LLZO (∼760° to 820 °C). Notably, the LLZTO sample shows an additional reaction from Ta₂O₅ to LiTaO₃ compared to LLCZWO in the second stage. Operando synchronous XRD demonstrates that Al-LLZO particles are developed in a cubic phase at 900°–1000 °C with remarkable particle coarsening.¹⁰² The simulation results reveal that particles with small size and bimodal distributions are preferable for subsequent sintering densification. Besides the use of different doping elements to change the phase conversion behavior during LLZO particle synthesis, the heating technique can also be a factor that introduces changes. The peak signal of La₂Zr₂O₇ (LZO) (2θ = 3.54°) is seen to rapidly increase in intensity and sharpness until its disappearance at 600° to 1065 °C when a reactive flash sintering (RFS) technique was used.⁹⁶ In contrast, LZO disappeared progressively in conventional heating experiments. This phenomenon suggests that RFS facilitates intermediate LZO growth during the heating process.

	Fig. 13 (a) Phase evolution of cosputtered film LLZO with Li2O. Reproduced from ref. 99 with permission from the American Chemical Society. (b) Contour plots of 2D in situ high-temperature XRD patterns of LLCZWO and LLZTO. Reproduced from ref. 101 with permission from the American Chemical Society. (c) Contour plots of in situ high-temperature XRD patterns of a composite mixture of NCM523 charged to 3.8 V with Ga-LLZO in the selected 2θ range under air and N2 flowing conditions. Reproduced from ref. 105 with permission from Wiley.

It is well known that LLZO on exposure to air not only forms Li₂CO₃ and LiOH contaminants on the surface, but also transforms partially into Li_7−xH_xLa₃Zr₂O₁₂ in the bulk. In order to recover cubic LLZO, high-temperature heat treatment is an effective way as confirmed by in situ high-temperature XRD.¹⁰³ Under an air atmosphere, high-temperature heat treatment consists of five stages: (i) Li_7−xH_xLa₃Zr₂O₁₂ + xLiOH = LLZO + xH₂O; (ii) Li_7−xH_xLa₃Zr₂O₁₂ = LLZO + LZO + (x/2) H₂O; (iii) LZO + Li₂CO₃ = LLZO + CO₂; (iv) LLZO = LZO + Li₂O (volatilization); and (v) LLZO + H₂O + CO₂ = Li_7−xH_xLa₃Zr₂O₁₂ + LiOH + Li₂CO₃.

Although large amounts of Li_7−xH_xLa₃Zr₂O₁₂ are recovered to cubic LLZO, the volatilization of Li₂O causes stoichiometric irreversibility during the heat treatment. The degree of stoichiometric irreversibility is dependent on the kind of doping additive. Furthermore, re-protonation reactions occur in the cooling process. For preventing such reactions, heat treatment in an argon atmosphere is demonstrated to be useful by in situ grazing incidence synchrotron XRD (GISXRD).¹⁰⁴ The cell performance of LLZO with heat treatment is superior to that seen when air is used. It is noteworthy to avoid re-exposure of heat-treated LLZO to air which induces protonation reactions. In situ high-temperature XRD also revealed the chemical compatibility and structural stability between charged LiNi_0.5Co_0.2Mn_0.3O₂ (NCM523) and Ga-LLZO when co-sintered at high temperatures in various atmospheres.¹⁰⁵ As shown in Fig. 13c, Li diffuses from Ga-LLZO to NCM523 with increasing temperature under an air atmosphere. Ga-LLZO is decomposed to LZO and charged spinel NCM523 is recovered from being layered. In contrast, Li diffuses from NCM523 to Ga-LLZO under a nitrogen atmosphere. Ga-LLZO is converted from the cubic to tetragonal phase and NCM523 is reduced to Ni metal. Therefore, co-sintering technique is inappropriate for improving the poor contacts between layered cathodes and LLZO.

X-Ray scattering occurs in the lattice of regularly spaced crystalline materials and this can be used to obtain information about the grain size, shape, or pore size.¹⁰⁶ However, there is no reported research study about in situ/operando X-ray scattering in the field of LLZO-based SSLBs. In principle, the in situ/operando X-ray scattering cell structure is similar to the in situ/operando XRD cell. Recently, the nucleation and nucleus size evolution of the Li metal at different current densities in liquid cells has been studied through synchrotron-based transmission mode grazing-incidence small-angle X-ray scattering.¹⁰⁷ The high overpotential induced small Li particle size and dense Li dendrites, which enlarged the reaction-specific surface area and consumed more electrolytes.

XRI techniques mainly consist of transmission X-ray microscopy (TXM) and X-ray tomography (XRT). Fig. 14a illustrates the schematic diagram of measurements for both imaging techniques.⁵ XRT, also known as computed tomography (CT), adopts a monochromator to select an incident beam for the sample. A scintillator and a charge-coupled device (CCD) camera converts the transmitted photons into an optical image with micrometer spatial resolution. XRT enables visualization of the internal sample information without slicing and is a real-space full-field imaging method and obeys the optical geometry principle. TXM generally adopts hard X-rays with shorter wavelengths and couples a monochromator and a capillary condenser to establish a micro-focused beam. A Fresnel strip plate (which amplifies the imaging), phase ring and CCD camera are used to realize phase contrast by converting the transmitted photons into an optical image with nanometer spatial resolution. In comparison to other imaging techniques, TXM offers advantages of simple manipulation and usage at atmospheric pressure. From above, the resolution of XRI techniques is decided by the optics in device systems. The spatial resolution ranges from ∼15 nm to several mm.¹⁰⁹ Furthermore, coupled tomography and far-field high-energy diffraction microscopy (FF-HEDM) have been used to evaluate the grain-level chemo-mechanics.¹¹¹ As shown in Fig. 14b, compared to XRT, FF-HEDM captures diffraction images on a 2D area detector. The images are processed to provide the mass center, sizes, crystallographic orientations, and lattice strain tensor information for each grain. FF-HEDM can quantify the stress evolution at an individual grain level in response to electrochemical cycling. In addition, coupled calorimetry and operando SR X-ray fast radiography evaluate the thermal behavior of a cell.¹¹²Fig. 14c illustrates a plan of a calorimeter and an instrumented cell irradiated with X-ray beam with thermocouple recording (TCR) and pressure recording (PR). The in situ cell of this setup is a commercial 18650 type battery while the calorimeter is made of a stainless-steel tube. The closed system is thin enough to allow high-energy X-rays to be transmitted through the cell. External measurements of temperature and pressure sensors are recorded over the whole extent of the protocol. This setup successfully establishes a link between external measurements and cell internal structures. In general, in situ XRI employs Swagelok-type cells¹¹³ and Fig. 14d shows the corresponding in situ Swagelok-type cell structure. 3D images are derived by rotating the in situ cell to acquire photographic images from various angles and reconstructing them with computer software.¹¹⁴ More importantly, with the benefits of high flux, high brightness, high power and low incident energy of SR X-rays, XRI techniques enable in situ/operando measurements on commercially available CR2023 coin-type cells.¹¹⁵ As shown in Fig. 14e, high-energy X-rays are incident along the direction of the sample cross-section to directly observe the internal structure and transmission imaging is acquired. Mapping analysis of transmission images allows real-time detection of changes in the internal cell structure during the charging and discharging reaction. Therefore, in situ XRI techniques can realize 3D visualization of the changes in the local morphology, structure, and chemical composition within solid-state batteries without causing any damage.

	Fig. 14 (a) Schematic diagram for XRI. (b) End-station detectors for carrying out tomography and FF-HEDM measurements with the corresponding information expected. (c) Plan of a calorimeter and an instrumented cell irradiated with X-ray beam with TCR and PR. Reproduced from ref. 112 with permission from the American Chemical Society. (d) Schematic diagram of an in situ Swagelok-type cell structure. (e) Schematic diagram of a commercially available CR2023 coin-type cell. Reproduced from ref. 115 with permission from AIP Publishing.

Due to the low attenuation coefficient, the pore (air) and Li metal regions are transparent to X-rays. In contrast, the LLZO ceramic electrolyte region intensely attenuates the X-ray displaying opacity, thereby appearing in the image. Owing to the distinct X-ray absorption ability of pores and LLZO, the LLZO porous structure was successfully investigated using XRT technique.^116–118 LLZO porous structures were initially prepared by the freeze casting method with tert-butyl alcohol (TBA) as the solvent (the mass fraction of LLZO is 20%).¹¹⁶ 3D images captured by SRXRT reveal that the pore size is ∼50 μm and pores are distributed homogeneously in the whole structure. These pore channels have identical orientations and are even bridge-free over the length range of several hundred μm (far exceeding the LLZO thickness of ∼100 μm). In order to fit the practical manufacturing process, LLZO porous structures are prepared by the freeze tape casting method.¹¹⁷ The effect of various LLZO volume fractions on porous structures was investigated using XRT technique. Fig. 15a shows that unblocked, opened, and homogeneous pore channels are formed for the 7.5 and 12.5 vol% LLZO slurries, while randomly distributed macro and closed pores are formed for the 17.5 vol% sample. Therefore, the higher the LLZO volume fraction, the smaller the pore size, the lower the porosity and the greater the number of closed pores. Such phenomenon is attributed to the higher LLZO volume fraction, as interactions between particles and solvent lower the particle dispersion stability. The particles fail to be uniformly excluded during freezing procedure. Compared to the freeze casting method, pore orientation performed by the freeze tape casting method is not vertical to the surface but at an angle. The pore channels are slanted as the tape-casting process causes the temperature gradients to be not perfectly aligned from the top to bottom. The temperature gradient also leads to a structure with small pores at the bottom and large pores at the top. Subsequently, the bilayer structure was prepared with a porous structural layer and dense layer.¹¹⁸ LiNi_0.6Co_0.2Mn_0.2O₂ (NCM622) or other cathode slurries were infiltrated into the porous structural layer. Unidirectional and open pore channels favour cathode particle infiltration, shortening of Li⁺ diffusion path length, and enlargement of the contact area. The Li anode was coupled on the dense layer side and the assembled SSLB exhibited good charge/discharge capability.

	Fig. 15 (a) Reconstructed thin films and longitudinal subvolume together with the front view images extracted from the tomographic results for the 7.5%, 12.5%, and 17.5% structures, respectively. Reproduced from ref. 117 with permission from the American Chemical Society. (b) XRT reconstruction of the void phase in the interior of LLZO electrolytes sintered at 1050°, 1100°, and 1150 °C before and after failure. Reproduced from ref. 122 with permission from the American Chemical Society. (c) The 3D rendering of the pillar with internal Li protrusions and voids is viewed from three different directions. The LLZO is in transparent light yellow, Li protrusions are in opaque grey or transparent yellow, and voids inside cracks are in opaque red. Reproduced from ref. 123 with permission from Wiley. (d) X-Ray radiography during the thermal runaway of liquid electrolyte Li-ion batteries and all-solid-state batteries. The white bar gives the 2 mm scale. Reproduced from ref. 112 with permission from the American Chemical Society.

Similarly, polymers are also transparent due to non-attenuation of X-rays. Accordingly, large attenuation contrast is seen in the region between the polymer and LLZO, which allows the visualization of LLZO particles directly in CE. The factors affecting the improved CE transport properties were examined using XRT technique based on this principle.¹¹⁹ In theory, the contact surface area with polymer should be enlarged as LLZO nanoparticle sizes decrease when the content increases. However, LLZO particle aggregation is clearly viewed when the content increases from 5 to 50 vol%. The particle clustering significantly lowers the accessible area, resulting in increased particle aggregation size.¹²⁰ The dielectric relaxation strength will decrease, indicating that the movement of amorphous polymer chains near the LLZO surface is constrained resulting in lower ionic conductivity. Consequently, ionic conductivity is highly dependent on the accessible area instead of the content of inorganic particles and an enhanced transport in CE was observed mainly due to polymer/inorganic interactions. In addition, an ordered distribution of inorganics is the ideal structure for designing CE. An asymmetric structural design was confirmed by XRT.¹²¹ An asymmetric CE with a LLZO-rich layer on the anode side and a polymer-rich layer on the cathode side was constructed via the selective adsorption of a glass fiber (GF) separator. This structure not only improves the ionic conductivity and Li⁺ migration ability but also addresses the SSE/electrode interface issue to a certain extent.

XRI has also been widely adopted to investigate the meso-scale transformation in LLZO-based SSLBs and the microstructural transformation in LLZO pellets. Stresses, cracks, and electrode volume changes all contribute to mechanical degradation in LLZO SSEs. Most studies have focused on characterizing the microstructural changes in LLZO pellets in response to electrochemical cycling. However, high-energy X-rays dramatically lower the attenuation contrast sensitivity of Li metals and voids. Therefore, it is hard to identify Li deposits, Li dendrites, and void regions. Fortunately, XRT can trace the evolution of LLZO phases and void regions. If isolated Li deposits or Li dendrites occur, void regions are significantly varied. Microstructural details can be analyzed by directly comparing XRT images of pristine LLZO pellets and failed samples. As shown in Fig. 15b, LLZO pellets sintered at 1050°, 1100°, and 1150 °C exhibit distinct void distributions before and after failure.¹²² In comparison to pristine samples, the increase in the number of voids in the failed LLZO demonstrate Li deposition. It suggests that Li metal is deposited in an isolated form or as an agglomerate and expanded in the void regions. Furthermore, samples sintered at higher temperatures (1150 °C) displayed more connected and tortuous void networks. Samples with higher connectivity exhibited more rapid Li dendrite formation and propagation and hence, high-density LLZO pellets are essential. In order to distinguish voids and Li metal, XRT with a high-resolution imaging mode is employed.¹²³ Such an imaging device utilizes a comparatively low X-ray energy of 5.4 keV to enhance Li and X-ray interactions. Inside the cracks in the XRT image, the lighter part corresponds to Li and the darker part corresponded to voids. According to greyscale discrepancies, Li, voids and LLZO can be clearly segmented and rendered as yellow, red and transparent light yellow, respectively (Fig. 15c). On one hand, Li protrusions mainly propagate intergranularly through LLZO, forming a wavy plane whose geometry is templated by grain boundaries. On the other hand, Li protrusions also form flat branches in a transgranular mode, located at complex geometrical features such as triple-junction grain boundaries. This indicates that Li deposition and Li dendrite formation favorably occurred at cracks, voids, and grain boundaries.¹²⁴ Importantly, coupled tomography and FF-HEDM can evaluate the grain-level chemo-mechanics in LLZO pellets.¹²⁵ High location overlaps between stress hotspots (grains with maximum stress values) and cold spots (grains with minimum stress values) and second phases were revealed by coupled FF-HEDM and tomography data; Li metal was preferentially deposited near such locations. Results indicate that failure in LLZO pellets was initiated locally since trace secondary phases cause mechanical stresses and result in ion transportation response inhomogeneities. In addition, a stable anode morphology is critical to achieve high electrochemical performance in LLZO-based SSLBs. Nevertheless, pore formation and local hotspot generation in Li metals are visualized by XRT with a high-resolution imaging mode, confirming the non-uniform interface dynamics.¹²⁶ Advanced image processing techniques revealed that Li metal hotspots correlate to anisotropic microstructures in LLZO. In particular, the failure cores are the local regions with suboptimal effective properties (such as ion transportation and mechanical stresses). Li dendrites that grew parallel to the LLZO/anode interface were detected via CT imaging.¹²⁷ The morphology of Li dendrites was “bowl-shaped” as observed by in situ SEM and “spalling” observed by in situ OM, indicating that internal stresses induced by Li deposition as the major driving force for propagation. Overall, LLZO shows severe chemo-mechanical degradation during electrochemical cycling. Cracks and secondary phases at the interface resulted in the emergence of local high-stress regions. Li was deposited in such regions preferentially, which led to LLZO failure. Application of stack pressure (1 to 20 MPa) or enhancement of temperature (25° to 100 °C) enabled homogeneous Li⁺ deposition at the interface.¹²⁸ Assessing the chemo-mechanical response of LLZO under relevant operating conditions is thus essential. Nonetheless, designing experimental setups that are compatible with X-ray end-stations with desired pressures and temperatures in imaging systems is challenging. It is expected that future work on setup design will enable controlled in situ and operando experiments.

In practical work on LLZO-based SSLBs, battery safety has received scarce attention and has not been assessed significantly until now. Fortunately, the thermal runaway of liquid electrolyte LIBs and SSLBs was visualized by coupled calorimetry and operando SR X-ray fast radiography (Fig. 15d).¹¹² Compared to LIBs, heat, and gas release in SSLBs during thermal runaway decreased by 11 ± 10% and 40 ± 20%, respectively, while reaction kinetics increased by 42 ± 20%. The decreased heat and gas release in SSLBs was attributed to fewer combustibles, such as liquid electrolytes and separators in the system. The increased reaction kinetics were attributed to improved thermal conductivity of LLZO ceramic particles and evaporation of a part of the liquid electrolyte.

In summary, in situ X-ray techniques offer straightforward visualization of microstructural evolution in LLZO-based SSLBs, including electrolytes, electrodes, and electrolyte/electrode interfaces. It is noted that the application of XRI in SSLBs is still at an early stage. XRI with laboratory X-ray sources is low-cost and operator-friendly, while low-intensity restricts resolution and makes it difficult to separate lighter elements. Although SR X-ray sources with high energy and precision capabilities enhance resolution, exorbitant experimental costs limit its use in normal laboratories. Hence, further practical design work will contribute to a comprehensive understanding of both microscopic and chemical structural evolution in SSLBs through XRI techniques.

	Fig. 16 (a) Schematic diagram for XPS measurements. (b) Schematic diagram of an in situ XPS cell and the configuration of in situ electrochemical XPS measurements. Photographs of (c) the sample holder and (d and e) the vacuum suitcase for transferring samples without air exposure. (f) Laboratory-type XPS apparatus. (g) The simplified schematics of the analysis chamber, electrical connections, and the stage with the sample holder. The terminals A and B are connected to a potentiostat via vacuum feed through external coaxial cables. Reproduced from ref. 130 with permission from the American Chemical Society.

Fig. 16b illustrates an in situ XPS cell schematic diagram and the configuration of in situ electrochemical XPS measurements.¹³⁰ An active electrode material layer was deposited on LLZO, such as an amorphous Si layer, by using Ar radio frequency magnetron sputtering. The sputter-deposited electrode active material layer was coated with a Cu collector by direct current (DC) sputtering. Importantly, during Cu plating, the central part of the electrode active material layer was masked with a stainless-steel stencil mask. The obtained uncoated Cu part of the active electrode material served as an analysis area for XPS measurements. The Li metal layer was thermally deposited on the other side of the LLZO. The Li metal layer of the in situ XPS cell was bonded to the Cu foil loaded on a sample holder (Fig. 16c). The bottom side was electrically isolated from the sample holder via a polyimide film and connected to terminal A. The top side was connected to the sample holder and terminal B. The cell-loaded sample holder was sealed in a vacuum suitcase and attached to the load lock (Fig. 16d, e). This was then transferred into the XPS apparatus without exposure to atmosphere (Fig. 16f). The sample holder was mounted on a stage in the analysis chamber for in situ electrochemical measurements (Fig. 16g). Terminal B and the sample holder were electrically grounded to the hemispherical analyzer of the XPS apparatus and then bias was applied to terminals A and B by using an electrochemical workstation.

LLZO and the doped (Al, Ta, Nb, and Ga) versions typically exhibit a wide electrochemical stability window and superior stability during charging and discharging processes. However, in situ and operando XPS revealed that an oxygen-deficient interfacial layer (ODI) was developed at the Al-LLZO/Li interface.¹³¹ As shown in Fig. 17a, ODI formation was attributed to the reduction of Zr⁴⁺ to Zr²⁺ and Zr⁰ by the introduction of higher energy when Li came into contact on the LLZO surface. Operando XPS measurements reveal that ODI formation is self-limiting, with very little change in the degree of Zr⁴⁺ reduction with increased Li deposition time after ∼1.5 min. Similarly, depth profile XPS also revealed the reduction of Zr⁴⁺ and generation of Li-O species at the Ta-LLZO/Li interface.¹³² The formation rate of ODI is highly affected by current density at the interface. At low current densities, ODI has a passivation effect to protect LLZO from sustained reduction and constrain interfacial resistance. In contrast, ODI formation was probably obstructed, and Li dendrite formation at interface was accelerated at high current densities. Interestingly, electrochemical interface response was not changed by the existence of ODI; in other words, interfacial charge transfer was not impeded. The lithiation/delithiation of amorphous Si electrodes on Ta-LLZO in thin film cell structures was investigated by in situ electrochemical XPS.¹³⁰ As shown in Fig. 17b, during initial lithiation/delithiation, the formation of irreversible species such as Li-silicates, Li₂O and Li₂CO₃ at the interface was observed. Such species contribute to irreversible capacity loss while interface reactions were also found at the cathode/LLZO interface. The first example is the LCO cathode where depth profile XPS revealed that deleterious Li₂CO₃, LZO and LaCoO₃ species were formed at the interface due to Co and La diffusion during annealing (Fig. 17c).¹³³ Owing to the formation of interfacial phases, Li migration ability was decreased, and interface impedance was increased. Results suggest that lower temperatures, shorter time and CO₂-free environments are processing conditions to achieve an ideal cathode/LLZO interface. Second, Mn in LiMn₂O₄ (LMO) also diffuses into LLZO during higher temperature cycling, which causes significant deterioration of cell properties.¹³⁴ Mn regions at the interface were detected by depth profile XPS and they were found to become wider with increasing cycles. Therefore, understanding the dynamic transformation of the interface and its role in electrochemical degradation is beneficial in stabilizing such interfaces.

	Fig. 17 (a) Operando XPS measurements of individual core-level spectra during electrochemical deposition of Li showing their evolution with increasing Li deposition time. Zr 3d core-level spectra compare the extent of Zr4+ reduction as a function of the deposition technique. Reproduced from ref. 131 with permission from the American Chemical Society. (b) Reproduced from ref. 130 with permission from the American Chemical Society. (c) Evolution of surface oxide, carbonate and hydroxide species with heat treatment temperature for LLZO solid electrolyte. Reproduced from ref. 133 with permission from the American Chemical Society. (d) XPS depth profile of Li3N- or LiF-rich layers at LLZO/Li anode interfaces. Reproduced from ref. 139 and 140 with permission from the American Chemical Society and Elsevier. (e) XPS depth profile of a freshly dissociated sample after 24 h air exposure: O 1s and Li 1s core spectra for various sputtering time periods from 1 to 15 min. Reproduced from ref. 142 with permission from the American Chemical Society.

Until now, many interface modification approaches have been successfully proved via depth profile XPS. First, introducing minor liquid electrolytes at cathode/LLZO and LLZO/Li anode interfaces can alleviate but not completely solve the issue cell electrochemical performance deterioration. XPS depth profile experiments show a spontaneous chemical reaction between LLZO and liquid electsrolyte.¹³⁵ The reaction layer formed on the LLZO surface is composed of a Li-rich phase near the surface and Li–La–Zr oxide on the sub-surface. A continuous increase in reaction layer thickness led to lower cell capacity efficiency and retention. For addressing chemical reactions induced by minor liquid electrolytes, nanometer-scale self-assembled monolayers (SAMs) with acidic anchoring groups were adopted.¹³⁶In situ Li⁺/H⁺ exchange or intense interactions between LLZO and liquid electrolyte were weakened by the compact SAM passivated layer, which achieved an interface resistance close to 0. Second, introducing ionic liquid-based gel interlayers on anode and cathode sides dramatically lowered the interface resistance and stabilized the interface.¹³⁷ The formation of SEI films in the interface region similar to those in conventional liquid cells was revealed by depth profile XPS. The amorphous nano-interlayer was characterized by an outer layer which was rich in organic species while the inner layer (near the LLZO surface) was dominated by inorganic species. Such structural character enables good battery performance. Third, the introduction of alloy layers at the LLZO/Li anode interface is also a promising approach. Depth profile XPS indicated that a ∼10 nm Sb metal layer was transformed to a Li–Sb alloy layer during lithiation/delithiation.¹³⁸ The alloy layer permitted efficient Li⁺ and e⁻ percolation and mitigated cavity and Li dendrite formation. The batteries assembled with intercalation-type V₂O₅ cathodes delivered favorable areal capacity and peak current density. Lastly, molten state composite anodes were constructed to achieve close contact and a stable LLZO/Li anode interface. Composite anodes were prepared by incorporating nitrides such as 2D boron nitride nanosheets¹³⁹ or graphitic carbon nitride (g-C₃N₄)⁷⁰ into molten Li resulting in high viscosity and low surface tension. The LLZO/Li anode interface allowed close integration due to the transformation from point contact to surface contact. Simultaneously, a Li₃N-rich layer was formed at the interface which featured high ionic conductivity and low electronic conductivity (Fig. 17d, left). Analogously, the incorporation of fluorides such as MXene with surface terminal functional group of –F¹⁴⁰ or AlF₃¹⁴¹ into molten Li also achieved close contact and a LiF-rich layer at LLZO/Li anode interface (Fig. 17d, right). Such approaches reduced interface resistance and improved the critical current density value, suppressing Li dendrite formation and thus achieving stabilization of electrochemical cycling.

As previously mentioned, a thin and insulating Li₂CO₃ layer was formed on LLZO pellets upon air exposure resulting in high interface resistance. The formation mechanism of the Li₂CO₃ layer was investigated by depth profile XPS. Initially, it was suggested that the air-contaminated LLZO structure was composed of three layers, and these were surface, sub-surface and inner layers which were the Li₂CO₃ layer (from CO₂ and LiOH reaction), intermediate oxide or hydroxide layer, and bulk LLZO layer, respectively.¹⁵ Later research indicated that the sub-surface layer was essentially comprised of protonated Li_7−xH_xLa₃Zr₂O₁₂ and LiOH due to Li⁺/H⁺ exchange (Fig. 17e).¹⁴² The Li₂CO₃ layer significantly obstructs Li⁺ migration at the interface which facilitates massive void generation and ultimately causes cell failure. Heat treatment is a simple approach to eliminate contaminated layers completely. The temperature range for decomposition of Li₂CO₃ to CO₂ is 620°–1000 °C. In contrast, in situ XPS showed that the decomposition of carbonate species on the LLZO surface initiates at ∼150 °C.¹⁴³ Notably, it is essential to prevent re-exposure of the clean surface to air after heat treatment. Otherwise, it will result in the formation of contaminated layers. Based on the knowledge of contaminated layer composition, this can be converted into an advantageous interlayer through different electrochemical reactions. For instance, the contaminant layer was converted to an aromatic layer via a simple azo reaction at 60 °C.¹⁴⁴ XPS depth profile characterization indicated that the aromatic layer was composed of a C–F-rich surface layer and a LiF-rich bottom layer. The aromatic layer not only offers favorable lithophilicity, but also facilitates Li⁺ migration and homogeneous Li⁺ deposition. Alternatively, acid treatment was adopted to eliminate Li₂CO₃ from the surface and grain boundaries while introducing protons.¹⁴⁵ Proton doping suppresses formation of conductive by-products and releases residual stress for retaining the intact interface contact, which remarkably enhances stability and energy density of LLZO-based SSLBs. Moreover, the laboratory-used Li foil surface was found to contain impurities such as Li₂O, Li₂CO₃, LiOH or LiF as seen from the XPS depth profile.¹⁴⁶ It is critical to provide a clean Li surface for achieving close contact at the LLZO/Li anode interface. The clean Li surface can be realized by heating Li foil to a molten state and then scraping off the contaminated layer which floats on the surface.

LLZO-based SSEs prepared by various synthesis methods generally exhibit distinct structures, properties, and applications. When LLZO particles were ball-milled together with LiBH₄, a unique amorphous dual coating was formed on the surface, which consisted of a LiBO₂ inner layer and a LiBH₄ outer layer and this was revealed by XPS depth profiles.¹⁴⁷ The amorphous coating acts as a filler, binder, and bridge, effectively enhancing contact among particles. The cold-press molded, and un-sintered pellets exhibited high densities, high ionic conductivity, high Li⁺ transference numbers, and low electronic conductivity. The mechanism of phase conversion in LLZO nanofibers prepared by electro-spinning was investigated using XPS depth profiles.¹⁴⁸ The structure and chemical composition of LLZO nanofibers sintered at various temperature stages were analyzed in detail. The diameter of nanofibers was less than 500 nm, which is adequate for complete penetration in XPS depth profile experiments. The phase conversion of nanofibers was identical to that of particles, indicating that various LLZO morphologies cannot affect the phase conversion mechanism.

LLZO thin films are crucial for flexible electronic device applications. High ionic-conductivity, high-density, homogeneous, and crack-free tetragonal LLZO thin films were grown on Pt via a novel CO₂ laser-assisted chemical vapor deposition (CVD) technique.¹⁴⁹ The presence of a Li₂PtO₃ layer between LLZO and Pt was proven by XPS depth profile analysis. The presence of such a layer affects total impedance but not the ionic conductivity of the LLZO thin film system. Therefore, the thin film grown via CVD is an ideal candidate for solid-state thin-film lithium batteries. Si nanoparticles filled in polymers help absorb and anchor Li dendrites.¹⁵⁰ Li–Si–O and Li_xSi species formed by reactions between Si nanoparticles and Li dendrites were identified by depth profile XPS. Such reactions suppress Li dendrite formation at the interface and swallow tips of Li inside SSEs to avoid longitudinal growth. Under the condition that the fundamental properties of LLZO-based SSEs are not impaired, adding other nanomaterials that spontaneously react with Li should also produce similar effects. Lastly, the depth profile XPS revealed that LLZO, polyvinylidene fluoride-hexafluoropropylene copolymer (PVDF-HFP), PEO, and SN synthesized CE formed hierarchical SEI on the Li anode side, consisting of a LiF-rich outer layer and a Li₃N-rich inner layer.¹⁵¹ The assembled cell offered good electrochemical properties, flexibility, and interface adhesion, displaying application prospects in the field of flexible electronics.

Overall, in situ XPS qualitatively examines surface/interface elements via measurement of electron binding energy, which characterizes phase and elemental composition. Compared to in situ XRD measurement, where phase content is only reflected in certain amounts, in situ XPS offers a wider measurement range and higher sensitivity. However, it fails to visualize interface evolution and Li dendrite morphology as an imaging technique in real time. Furthermore, individual detection depth is less than 50 nm and requires integration with ion sputtering techniques to detect the whole composition. Therefore, the development of innovative XPS technology and compatible in situ cells is still a major challenge.

	Fig. 18 (a) Schematic diagram for XAS studies. (b) Ni L-edge and O K-edge XAS spectra of hybrid LLZO and NCM111 samples detected in the TEY mode and TFY mode at different stages. The inset shows the comparison of the Ni L3 edge. (c) Brief schematic illustration for the alteration of surface topography during the ball-milling and cosintering process. Reproduced from ref. 154 with permission from the American Chemical Society. (d) Co K-edge, Co L3-edge, and O K-edge X-ray absorption spectra of hybrid LCO and LLZO samples at different stages. Reproduced from ref. 133 with permission from the American Chemical Society. (e) Operando S K-edge spectra with first derivative mapping, Ni K-edge spectra, and charge/discharge profiles of bare NMC811-LGPS. Reproduced from ref. 155 with permission from the American Chemical Society.

Fig. 18b illustrates Ni L-edge and O K-edge XAS spectra of hybrid LLZO and LiNi_1/3Co_1/3Mn_1/3O₂ (NCM111) samples in TEY and TFY modes.¹⁵⁴ Results indicated that LLZO spontaneously coats NCM111 particles with a thickness of ∼100 nm after ball milling. The thickness of the LLZO layer was reduced to ∼10 nm and an interlayer of ∼3 nm was seen after cosintering at 600 °C (Fig. 18c). The interlayer was attributed to Ni/La and Ni/Li exchange resulting in the formation of LZO and LaNiO₃ species. The structure, chemical information and charge transfer resistance of the LCO/LLZO interface after cosintering at 500 °C were also investigated by in situ XAS.¹³³ As shown in Fig. 18d, Co K-edge, Co L-edge, and O K-edge XAS spectra revealed Co elemental oxidation and LLZO decomposition, causing the formation of La-Co-O species with a thickness of 20 nm at the interface. Hence, temperature, time, and atmosphere control in cosintering are vital to realize a superior cathode/LLZO interface. Besides, operando XANES spectroscopy was conducted to investigate the interface behavior between the NCM811 cathode and Li₁₀GeP₂S₁₂ (LGPS) electrolyte.¹⁵⁵ S K-edge and Ni K-edge spectra shown in Fig. 18e suggested that NCM811 and LGPS are unstable during the electrochemical process. Partial LGPS is decomposed to a Li₂S contaminant phase. Micro-cracks and phase evolution from a layered to rock-salt structure occur on the NCM811 surface. Operando XANES spectroscopy further proved that the LiNbO_x coating layer protected the integrity of the NMC811 cathode and stabilized the NCM811/LGPS interface. Therefore, in situ/operando XAS exhibits considerable potential for probing reactive intermediate states, interface products and reaction mechanisms. Notably, the in situ/operando cell setup poses several challenges during experimental manipulation and requires further development and optimization.

	Fig. 19 (a) Schematic diagram showing AFM microscopy technique. (b) Cross-sectional schematic diagram of an EC-AFM cell. Reproduced from ref. 160 with permission from Elsevier. (c) A digital image and schematic diagram of an in situ EC-AFM cell used for cross-sectional measurement. Reproduced from ref. 40 with permission from Elsevier. (d) Schematic illustration of ionic and electronic co-migration or electronic migration imaging devices based on conductive AFM. Reproduced from ref. 161 with permission from Elsevier.

In contrast to OM or EM, AFM provides surface information (including morphology, mechanical properties such as stress or Young's modulus and electrical properties such as electrical conductivity or diffusion resistance) of numerous materials (e.g., conductors, semiconductors, and insulators) in diverse environments (e.g., ambient air, liquids, or high temperatures).¹⁵⁹ Electrochemical AFM (EC-AFM), frequently used in the field of battery research, is mainly employed in the contact mode. Incorporating the benefits of high-resolution surface/interface analysis and electrochemical reactors, EC-AFM provides the evolutionary principles and electro-mechanical properties of surfaces/interfaces in real space during the charging and discharging process.

A fundamental architecture of a three-electrode electrochemical AFM liquid cell for surface/interface probing is shown in Fig. 19b.¹⁶⁰ High-resolution images are obtained from variations of phase difference between the driving signal and cantilever response. In LLZO-based SSLBs, given the solid–solid interface of buried electrodes/LLZO-based SSEs and high-activity Li metal, it is critical to refine the fundamental architecture and condition of liquid cells for friendly and non-destructive in situ AFM measurements. Based on this, the design of AFM electrochemical in situ LLZO-based SSLBs should be as follows: (1) cell assembly should be placed in an argon-filled glovebox since components are sensitive to air; (2) there must be excellent electrical connections among the electrode, the current collector and the external circuit; (3) distance between the counter electrode and working electrode inside the cell should be minimized to shorten Li⁺ migration distance, that is, LLZO-based SSEs should be minimized; (4) avoidance of internal short-circuits during cell assembly; (5) cell assembly materials and cantilever probes should be electrochemically inert to electrodes and LLZO-based SSEs; and (6) experiments should be conducted under the protection of an inert gas such as argon.

Fig. 19c illustrates a digital image and schematic diagram of an in situ EC-AFM cell used for cross-sectional measurement.⁴⁰ The EC-AFM apparatus is placed in an Ar glovebox and the in situ cell is scanned using an insulated silicon AFM tip. The Li anode, electrolyte, and cathode are sliced with a scalpel to obtain a high-quality cross-section. The layered films are compressed with polyamides and stainless-steel plates. The in situ electrochemical cell was assembled on the corresponding spectroscopy and imaging system coupled with a potentiostat to investigate real-time interface reaction processes. Fig. 19d illustrates the schematic illustration of ionic and electronic co-migration or electronic migration imaging devices based on conductive AFM (c-AFM).¹⁶¹ The Young's modulus, adhesion and current can be simultaneously obtained. The half-cell consists of LLZO-PEO electrolyte, a Li metal anode and a conductive AFM probe. The sample table is a current collector to drive Li⁺ migration with a certain voltage. When a positive voltage is applied to the sample table, Li⁺ from the Li metal anode passes through electrolyte and transforms into Li atoms by reduction of the AFM probe after reaching the sample surface. Since the probe range is significantly small, reduced Li⁺ is removed from the surface by the probe to prevent bulk deposition. Meanwhile, e⁻ migrate in the opposite direction from the probe to the current collector. As a result, current collected by c-AFM is the sum of electronic and ionic currents under a positive voltage. When a negative voltage is applied to the sample table, Li⁺ does not migrate from the probe to the Li metal anode through electrolyte due to Li-depleted electrolyte. The current collected by c-AFM is the only electronic current. Although in situ AFM techniques have been used in a limited capacity in LLZO-based SSLBs so far, several investigations have been conducted to assess electrochemical–mechanical transformations in the electrode, electrolyte, or electrode/electrolyte interface during electrochemical reactions.

In the field of Li-based battery research, in situ AFM was originally employed in 1996 to investigate the exfoliation/deposition behavior of various types of Li salts (including LiClO₄, LiPF₆, and LiAsF₆) in propylene carbonate (PC) solvent on the Cu current collector surface.¹⁶² In an electrolyte with LiAsF₆, Li⁺ deposition on the Cu current collector surface is seen to form a smooth layer. In the field of LLZO-based SSLBs, the potential differences between NCM811 particles were measured using AFM on the cathode side.¹⁶³ The average potential of NCM811-Li₃PO₄-LLZO composite cathode particles (−144.3 mV) was lower than that of NCM811-LLZO (−42.1 mV). The results indicate that Li₃PO₄ interlayer modification weakens the space charge layer. As a result, charge transfer between LLZO and cathode active material is boosted. On the Li anode side, both growth mechanism and mechanical properties of Li whiskers under mechanical constraints were investigated by in situ AFM coupled with environmental TEM (AFM-ETEM).¹⁶⁴ The Li₂CO₃ layer on the Li metal surface was crucial to stabilize Li whiskers and avoid electron damage during measurement. With an increasing applied voltage, the growth process of Li whiskers consists of three stages: (1) A spherical Li dendrite is generated and the size is increased. (2) Li whiskers develop axially due to accumulation of inner stresses. (3) Li whiskers collapse suddenly by the release of axial stress. Moreover, the growth of individual Li whiskers generates a stress of up to 130 MPa and yield strength as high as 244 MPa at room temperature. By utilizing mechanical characterization of AFM to monitor the stress during Li dendrite growth in real time, it can provide a novel idea for suppression of Li dendrite growth. Afterwards, morphological evolution of Li deposition in Au, Ag, and Si interlayers during cycling was evaluated by AFM.³⁵ As shown in Fig. 20a, Li deposition in Ag interlayers exhibited a smaller and more homogenous Li nucleation distribution and lower roughness compared to that in Au and Si interlayers. The cycle efficiency with the Ag interlayer system was also higher than that with the Au and Si interlayer system. Notably, an Ag interlayer was deformed during Li deposition/exfoliation due to the formation of Li–Ag alloy, which resulted in lower Coulomb efficiency. Overall, the variations in cell performance are determined from varying electrodeposition and stripping kinetics of interlayers. Furthermore, the flexible and electronically-insulated Li-inserted polyacrylic acid (LiPAA) layer automatically adjusted the stress intensity according to strain evolution, thereby suppressing dendrite growth during the Li plating/stripping process.¹⁶⁵ These artificial interlayers not only resolve the issue of dendrite growth, which can cause cell failure during cycling, but also effectively mitigate the side reactions between Li metal and air, thereby significantly enhancing safety.

	Fig. 20 (a) AFM images of the surface of silver and gold interlayers at various charge and discharge states (Rrms = root mean square roughness, and the scale bar in each image = 10 μm). Reproduced from ref. 35 with permission from Wiley. (b) In situ AFM images of CE with LiTFSI, LiFSI and LiTFSI-LiFSI bi-Li salts during initial charging and discharging. The scale bars are 1 μm. Reproduced from ref. 40 with permission from Elsevier. (c) In situ conductive AFM characterization of Li+ migration in CE with 50 and 75 wt% LLZO at 55 °C. Reproduced from ref. 161 with permission from Elsevier.

For CE, the effect of various PEO molecular weights on electrochemical and interface mechanical properties was investigated using AFM.¹⁶⁶ CE with 300 K exhibited high adhesion resulting in higher capacity and retention. The joint variation of adhesion and Young's modulus indicated that surface mechanical properties were decided by the sub-surface distribution of ceramic particles in CE. The morphological evolution of CE (PEO and Ta-LLZO) during cycling in Li–S cells was real-time monitored on a nano-scale via in situ EC-AFM.³⁹ With dissolution of PSs into CE, PEO is progressively transformed into an amorphous state. Interactions between the PEO matrix and Ta-LLZO particles were continuously weakened, which caused Ta-LLZO to become unstable and movable. Eventually, the volume of CE was increased as Ta-LLZO particles separated from the PEO matrix. In situ OM and in situ Raman results revealed that PS shuttling had a critical influence on the mechanical stability of CE and cycling stability of cells. In situ EC-AFM further explored the modulating effects of different Li salt anions in CE on PS shuttling and interface stability.⁴⁰ As shown in Fig. 20b, CE with LiTFSI, LiFSI, and LiTFSI-LiFSI bi-Li salts exhibited differences in structural evolution and interfacial behaviors during charging and discharging. In the LiTFSI system, the reaction kinetics were accelerated by severe dissolution of PSs, resulting in structure collapse within the interior of CE and weakening of interfacial contact. In contrast, the LiFSI system delayed the dissolution of PSs, which prevented structure deformation and generated a LiF protective layer for enhancing strain tolerance. However, FSI⁻ reacted with PSs to form Li-sulfite salts, resulting in constant and irreversible consumption of active S. The LiTFSI-LiFSI bi-Li salt system provides high ionic conductivity and wettability through LiTFSI and a LiF protection layer through LiFSI, achieving an optimal balance between interface compatibility and PS solubility. The bi-Li salt system is beneficial for realizing a superior interface and electrochemical properties. Therefore, real-time observation of in situ EC-AFM offers straightforward insights into interface morphology evolution mechanisms and reaction kinetics, guiding the direction for material design and property enhancement in LLZO-based SSLBs. In situ c-AFM was also performed to reveal the effect of different LLZO contents and temperatures on Li⁺ and e⁻ migration in CE.¹⁶¹ The images revealed that at a low temperature of 30 °C, Li⁺ only migrates along amorphous PEO regardless of the LLZO content. At a high temperature of 55 °C, Li⁺ migrates mainly along amorphous PEO in CE with 50 wt% LLZO, while Li⁺ can migrate along LLZO particles in CE with 75 wt% LLZO (Fig. 20c). Results suggest that Li⁺ migration pathways in CE are jointly affected by LLZO content and temperature. Crucially, in situ c-AFM showed that the introduction of LLZO particles enhanced CE electronic insulation. Hence, identification of ion migration pathways in CE via morphology evolution paves the way for rational design in future LLZO-based SSLBs.

Overall, in situ AFM provides better resolution and a vacuum-free working environment compared to in situ OM or in situ SEM. More importantly, AFM can examine insulative materials and mechanical properties, which are available to instruct native interface enhancement and artificial interlayer design. Although in situ AFM enables visualization of Li dendrite growth and surface/interface structure evolution, it suffers from a slower imaging rate. At high current density, the temporal resolution is inferior, resulting in failure to capture the whole evolution process and omitting vital procedures. Consequently, the improvement in the temporal resolution of in situ AFM is an issue to be addressed.

In the solid-state battery field, SSNMR is a critical approach to characterize ionic dynamics and local structures at the atomic level via detection of certain nuclei such as ⁷Li or ⁶Li. MAS NMR is used for a precise local structure and site occupancy of SSE or information about interface species. PFG NMR and NMR relaxation time measurements have been used for ionic movements on various length scales or time scales. PFG NMR is available for a time scale of 10⁻² to 10⁻¹ s, corresponding to a length scale of ∼1 μm. In contrast, NMR relaxation time measurements, including spin–lattice relaxation (T₁), spin–spin relaxation (T₂), and spin–lattice relaxation in a rotating frame (T_1ρ), are available on a time scale of 10⁻¹⁰ to 10⁻³ s.¹⁷² Specifically, the time scales of T₁, T₂ and T_1ρ are approximately 10⁻¹⁰–10⁻⁷, 10⁻⁶–10⁻⁴, and 10⁻⁴–10⁻³ s, respectively. Ionic dynamics in materials are qualitatively analyzed since the line width of NMR signals is inversely related to T₂.¹⁷³ 2D EXSY NMR is employed to explore the transport pathway and transport rate between two different chemical sites (either in a single material such as SSEs or in two materials such as SSEs and cathodes), i.e., ionic transport pathways both in SSEs and at interfaces can be investigated.¹⁷⁴ MRI, a non-invasive imaging technique, is utilized to visualize interface regions to provide information on the structure, morphology, and elemental distribution. In situ NMR technique is generally adopted for real-time monitoring of the evolution of structural and compositional information about the electrode, electrolyte, and electrode/electrolyte interface in lithium batteries. Li exhibits diamagnetism either in the electrolyte or interface film while the Li metal anode shows paramagnetism, resulting in distinction in identification due to difference in resonance frequencies. In addition, radio frequency signals from Li metal anodes are constrained at a surface depth of ∼12 μm (called the skin effect), while characteristic size of Li deposits is smaller than such depths.¹⁷⁵ The radio frequency signal can completely penetrate Li deposit microstructures; therefore, in situ NMR technique can qualitatively analyze the evolution of Li microstructures at the interface.

Design of in situ cells in NMR experiments is considerably critical. The plastic bag cell illustrated in Fig. 21a is favored by researchers due to its flexible design.¹⁷⁶ The two electrodes in an in situ plastic bag cell are connected to Cu mesh and Cu or Al mesh, respectively. The metal mesh helps overcome skin depth issues that are related to radio frequency penetration through the metal foil. In specific, metal foil shields the material inside the cell from radio frequency (RF) pulses. Moreover, the metal mesh can create better adhesion to electrode materials. Between the two electrodes is an electrolyte-soaked separator (for liquid cell) or a solid electrolyte (for solid cell). Then the in situ cell is placed in the coil of an NMR probe and cycled in the magnet. However, when the cell is inserted in an NMR coil, there is lack of stable pressure in the whole cell. The resistance inside the cell is increased and the contact between the particles and the current collector is reduced. Based on this, pressure is applied to the cell by placing polytetrafluoroethylene (PTFE) plates outside the in situ cell and tightening the cell with Teflon tape (a stronger string/tape can be used) or by applying small clamps to plates. A vacuum sealer is available to seal bag cells to produce them with a more consistent pressure. Besides the plastic bag cell, coin cells and cylindrical cells are also employed in in situ NMR experiments.¹⁷⁷ Due to sufficient temporal and spatial resolution, in situ MRI provides 3D images of species with specific concentration and NMR activity on the time scale of electrochemical reactions.

	Fig. 21 (a) Schematic illustration of the in situ NMR setup. Reproduced from ref. 176 with permission from Elsevier. (b) Pictures and schematic illustration of a symmetric Li|LGPS|Li cell placed in a home-made cylindrical cell for MRI. Reproduced from ref. 178 with permission from the American Chemical Society. (c) A fixed gap cell for synchronized operando characterization of acoustic transmission and ssNMR spectroscopy. Reproduced from ref. 179 with permission from Springer.

Fig. 21b illustrates pictures and schematic diagrams of an in situ cell placed in a home-made cylindrical cell for MRI.¹⁷⁸ Two pieces of Li foil are pressed closely onto both surfaces of electrolyte (LGPS) to ensure intimate contact. The overall setup is transferred into a home-made cylindrical cell designed for MRI. The main body of the cylindrical cell is fabricated from electrochemically inert polyether ether ketone (PEEK), which allows penetration of radio frequency pulses for NMR and MRI experiments. High purity stainless-steel wires and Cu foil act as current collectors. Rubber O-rings are used to ensure air tightness. The assembled cylindrical cell is placed in an NMR probe to enhance sensitivity and minimize noise for both NMR and MRI acquisitions. Interestingly, a fixed gap cell for synchronized operando characterization of acoustic transmission and SSNMR spectroscopy is designed, as shown in Fig. 21c.¹⁷⁹ Acoustic transmission probes change in interface contact mechanics, while ssNMR measures rates of microstructure formation. Such a cell is useful for characterizing mechanics and microstructures of the same cell under operation without noise from cell-to-cell variation. Two wide-band ultrasonic transducers are firmly bonded using epoxy to either side of a machined PTFE cell holder. An acoustic gel coupling agent is placed between the transducer face and PTFE cell surface. The fabricated cell is mounted in a cell holder. A cell-grade Cu mesh current collector replaces Ni foil. Ultrasonic transducers are attached to shielded electrochemical cables and cables are attached to protruding Cu mesh current collectors by using a small amount of Sn solder between the exposed wire and edges of Cu mesh. The cable connected to the bottom electrode is passed through the magnet, while the cable connected to top electrode is passed through top of the magnet. Since no cable crosses the electrolyte area, the signal-to-noise is lowered. Before operando tests, the cell potential is checked with a multimeter to verify that good contact is achieved.

Ex situ NMR technique offers distinct advantages in investigating ionic movement in LLZO-based SSEs. It provides an in-depth comprehension of the microstructure and ionic transport characteristics (including the transport rate and pathway) in SSEs on an atomic scale. The ionic conductivity is a vital parameter in the design of high performance LLZO. It is well known that the ionic conductivity of cubic LLZO is two orders of magnitude higher than that of tetragonal LLZO. The stabilization of the cubic phase at room temperature is needed to dope various elements. NMR spectroscopy can reveal doping mechanisms and site occupancy by doped elements. ²⁷Al or ⁷¹Ga NMR spectra indicated that the Al or Ga element is preferentially occupied at tetrahedral 24d sites instead of octahedral 96h or La sites in the LLZO structure.¹⁸⁰¹⁷O NMR spectra showed distinct and well-resolved resonance signals which were attributed to O bound to the doped elements (Al or Ga). Consequently, the occupation of Al or Ga at Li sites instead of its presence at grain boundaries is mainly responsible for stabilizing cubic LLZO. Introduction of heterovalent cations generally causes charge imbalance and Li redistribution. Doping of high-valent cation reduces the total amount of Li⁺ in LLZO, while doping of low-valent cation enhances the Li⁺ content. ⁶Li NMR spectra displayed that replacing Zr with W reduces Li⁺ occupancy at octahedral 48g sites.¹⁸¹ Replacing La and Zr with Ca and W, respectively, increased Li⁺ occupancy at 96h sites.¹⁰¹ Replacing Li, La, and Zr with Ga, Ba and Ta, respectively, enhanced Li⁺ occupancy at 24d sites and maintained constant occupancy at octahedral 96h sites.¹⁸² Such examples demonstrate that doping with ions of various valence states and at different contents of cations can change the Li⁺ occupancy at Li sites, resulting in lower migration barriers, lower activation energy, and higher ionic conductivity. Notably, doping with different cations modulates the occupancy of Li sites in a distinct manner. The systematic elaboration of the modulation mechanism in the cases of doping with different cations required further investigation. Moreover, Li⁺/H⁺ exchange occurring on the LLZO surface also changes the Li⁺ occupancy at Li sites and total content.¹⁸³

For anion doping, calculations suggest that F⁻ is favored to replace O2 and O3 sites in energy stabilization compared to Cl⁻ and Br⁻.¹⁸⁴ However, SSNMR only confirmed the presence of F in LLZO. The precise position of F in the structure could not be located. Importantly, the optimal doping content was determined via⁷Li MAS NMR.¹⁸⁵ The smaller the full width at half maximum (FWHM) in spectroscopy, the higher the ionic conductivity. The ionic conductivity is also affected by transport pathways and transport rates. For Li⁺ transport pathways within LLZO, NMR relaxation time measurement and stimulated echo NMR technique indicated the lowest activation energy for Li⁺ diffusion following 96h–24d–96h' pathways.¹⁸⁶ Moreover, 2D EXSY NMR demonstrated an ionic exchange between 24d and 96h sites and suggested that the Li⁺ transport pathway is 24d–96h–48g–96h–24d.¹⁸¹ Both transport pathways are essentially identical. The Li⁺ migration rate is slowest at 48g sites, which is considered as the deciding step to constrain ionic conductivity. Nonetheless, the NMR technique was unable to exclude direct jumps between two neighboring 96h sites with lower frequency and higher activation energy.¹⁸⁷ Li⁺ transport rates in LLZO can be accessed using dynamic NMR at different temperatures.¹⁷² The line width of ⁷Li signals in Ga- or Al-doped LLZO begins to narrow at a temperature of 180 K compared to 280 K for undoped LLZO, indicating rapid ionic movement in Ga- or Al-doped LLZO.

Besides LLZO substances, Li⁺ transport pathways in CEs consisting of LLZO and organic polymers are not clear. To resolve this issue, NMR technique has been used to reveal Li⁺ transport pathways in several CEs. Introduction of ionic liquids generally enhances both ionic conductivity and mechanical properties in CE.^188,189 Unexpectedly, variable temperature NMR and 2D EXSY NMR techniques indicate that chemical exchange between ionic liquids and LLZO particles is absent.^188,189 Enhancement of ionic conductivity is attributed to Li⁺ migration preferentially in ionic liquids while LLZO particles provide mechanical strength. Consequently, CE consisting of inorganic ceramic particles and ionic liquids is possibly not essential for expensive active ceramics such as LLZO. When LLZO primary particles are uniformly dispersed in PEO, Li⁺ migration is transferred from PEO polymers to LLZO ceramic particles with increasing LLZO content as confirmed by ⁶Li NMR and 2D EXSY NMR.¹⁹⁰ In specific, Li⁺ migration occurs mainly through LLZO ceramic particles rather than the LLZO/PEO interface or PEO polymer as LLZO content is 50 wt%. Moreover, a spontaneous Li⁺ exchange procedure between PEO polymers and LLZO particles resulted in a homogeneous Li⁺ diffusion and minimized the current gradient in CE.¹⁹¹ The protonated LLZO surface with the LiOH phase also exchanged Li⁺ with bulk LLZO and PEO polymers.¹⁹² Besides PEO polymers, several polymers such as polyvinylidene fluoride (PVDF), PAN and poly(ε-caprolactone-co-trimethylene carbonate) (PCL-PTMC) have been tested for interactions with LLZO particle surfaces for initiating rapid Li⁺ exchange.¹⁹³ Since LLZO particles in the CE system exhibit intense alkalinity, partial dehydrofluorination of PVDF matrix was revealed by ¹H NMR, which reinforces the interactions between PVDF and LLZO. Analogously, partial dehydrocyanation of PAN polymer skeletons formed local conjugated structures, resulting in continuous Li⁺ rapid conduction pathways at the LLZO/PAN interface. Interestingly, NMR studies suggest that preferential Li⁺ transport pathways in CE can be artificially controlled by utilizing several artificial modification approaches. Etching of LLZO particle surfaces to eliminate inherent resistive layers facilitates Li⁺ transport along locally modified interfaces and neighboring environments.¹⁹⁴ When LLZO is present in nanowires rather than particles, Li⁺ diffusion preferentially occurs in the LLZO/polymer interface region.¹⁹⁵ Construction of novel polymer skeletons such as fluoroboron-centered Li-conductive polymer frameworks (LiBFSIE) enabled the Li⁺ transport to be mainly dependent on LLZO particles and LiBFSIE/LLZO interfaces.¹⁹⁶ With the incorporation of additives such as plasticizer tetraethylene glycol dimethyl ether (TEGDME) or organic plastic salt (OPS), Li⁺ was principally transported via additive-associated phases instead of LLZO or polymers (Fig. 22a).¹⁹⁰

	Fig. 22 (a) 6Li NMR comparison of pristine and cycled LLZO with various contents or additive TEGDME. Reproduced from ref. 190 with permission from the American Chemical Society. (b) Li dendrite growth in LLZO visualized by 7Li NMR CSI. Reproduced from ref. 198 with permission from the American Chemical Society. (c) Voltage profile normalized peak intensities and 7Li chemical shift (ppm) of the Li metal region in operando MRI testing. Reproduced from ref. 179 with permission from Springer.

Incompatibility and electrochemical instability at the electrode/electrolyte interface are the main reasons that have limited practical applications of LLZO-based SSLBs. In order to comprehend the root causes of complex and transient reactions at the interface, in situ or ex situ NMR techniques have been developed and applied. On the anode side, Li deposits with differing structures were identified using NMR resonance displacement¹⁹⁷ since the characteristic peak value for dendrite Li is ∼270 ppm, while that for mossy Li is ∼261 ppm. Furthermore, the reduction of peak intensity in the NMR spectrum indicates enhancement of Li deposition. Accordingly, NMR technique demonstrated that mossy Li dominates initially, while dendrite Li growth occurs in later stages.^58,197 The LLZO/Li anode interface is generally accepted to be a kinetically stable interface but the major issue with this interface is Li dendrite growth. Owing to different resonance displacements, the growth evolution of Li microstructures in the Li |LLZO| Li symmetric cell was visualized by ⁷Li NMR chemical shift imaging (CSI).¹⁹⁸ As shown in Fig. 22b, although the current density was lower than 0.5 mA cm⁻² and the voltage profile was not characterized by short-circuiting, Li dendrites were generated and progressively accumulated to connect both electrodes ultimately. Such a method can be further employed to investigate electrolyte thickness, current density or stacking pressure effects on Li dendrite growth. The effect of stacking pressure on interface contact and Li microstructure growth was examined by SSNMR CSI.¹⁷⁹ At a stacking pressure of 2 MPa, continuous void formation at the interface caused significant surface roughening. Above a stacking pressure of 7 MPa, Li microstructure growth was alleviated due to creep-driven interfacial dynamics. Although contact depletion was reversibly recovered at stacking pressure up to 13 MPa, the likelihood of fracture induced by local crack propagation was increased. Subsequent operando MRI illustrates irreversible Li microstructure (⁷Li resonance peak centered at ∼258 ppm) formation during cycling (Fig. 22c). The stacking pressure of an operando cell is <1 MPa as tested by EIS. Therefore, searching for an appropriate stacking pressure is necessary to avoid cell short circuits. In addition, Li/LGPS interface evolution during cycling was investigated using MRI technique.¹⁷⁸ Images show that Li⁺ is non-uniformly distributed at the Li/LGPS interface after cycling, resulting in increased interface resistance and local hotspot development. PEO-coated LGPS demonstrated a uniform distribution of Li⁺ at the interface, which lowered the interface resistance and suppressed dendrite growth. On the cathode side, NMR technique revealed that a continuous Li⁺ transport pathway was formed within the interior of the composite cathode and this consisted of LLZO particles and cathode active material, reducing the Li⁺ diffusion distance and lowering the interface resistance.¹⁹⁹³¹P MAS NMR spectra showed elemental diffusion in composite cathodes comprisingLi_3+xP_1−xSi_xO₄ electrolyte and LFP active material during heat treatment.²⁰⁰ Fe in LFP diffused into Li_3+xP_1−xSi_xO₄ leading to the formation of Li_3−zFe_zP_1−zSi_zO₄. The interface transport rate between Li₂S active material and Li₆PS₅Br electrolyte under various preparation conditions was quantitatively analyzed by ⁷Li 2D EXSY NMR.¹⁷⁴ Preparation steps such as nanosizing and ball milling can accelerate Li⁺ exchange at the interface for improving ionic transport performance. However, electrochemical cycling remarkably changed such an exchange behavior, leading to lower ionic transport. Thus in situ NMR techniques are expected to be useful in investigating ionic transport at LLZO/cathode interfaces during cycling in future studies.

Overall, in situ and ex situ NMR techniques serve as powerful tools to research non-transparent samples, namely, LLZO-based SSLBs. In situ NMR technique with non-invasive character can simultaneously provide variation in ionic concentration at interfaces, Li dendrite formation and microstructure evolution. However, the resolution of NMR technique for solid substances was lower than for liquids due to the wider NMR spectrum.²⁰¹ In particular, MRI offers lower temporal and spatial resolution and is thus unable to image topographic changes in real-time as in situ EM, X-ray imaging or AFM. Improving the resolution will require a larger magnetic field gradient or the use of more advanced NMR pulse technique.

	Fig. 23 (a) In situ EPR spectra of the full cell at OCV and after charging to 3.6 V. (b) In situ EPR spectra of the Li symmetric cell. (c) Operando EPR image of the full cell, which is cycled between 2 and 4.6 V with images taken at voltages defined by letters on the voltage–composition curve by maintaining the cell at rest at these potentials. Reproduced from ref. 202 with permission from Springer.

Although in situ/operando EPR is a powerful analytical tool for cell electrode characterization, it is currently seldom employed in LLZO-based SSLBs. The LLZO surface state after molten Li treatment was investigated using ex situ EPR.²⁰⁵ Results suggested that the LLZO surface exhibits unpaired electrons trapped in oxygen vacancies, resulting in chemical coloration (from initial brownish white to grayish black) and enhanced electronic conductivity. As a result, prolonged molten Li manipulation increased the risk of Li penetration. Ex situ EPR also confirmed the oxidation of the Mn element in LLZO.²⁰⁶ The signal intensity of EPR peaks increased with increasing Mn content, and the calculated g-factor was consistent with the Mn²⁺ valence state. Moreover, in situ EPR was utilized to visualize the growth of Li microstructures in a liquid cell.²⁰² As shown in Fig. 23a, the EPR spectrum is featureless at OCV. By charging the cell to 3.6 V, a remarkably sharp and slightly distorted signal (g = 2.0023) of 1.5 G line width appeared. Such sharp signals are attributed to small metallic Li aggregates deposited on Li foil during charging. EPR spectra of a Li symmetric cell also revealed the presence of Li aggregates as a bias voltage was applied (Fig. 23b). Subsequently, operando EPR imaging tests were carried out. As shown in Fig. 23c, each pixel in the image is an EPR spectrum. Red, green, and blue colors indicate zones of high, medium, and no-spin electron concentration, respectively. Red regions correspond to Li aggregates, suggesting inhomogeneous Li microstructural deposition. Li dendrite formation was attributed to excessively inhomogeneous Li deposition. Incorporation of fluoroethylene carbonate (FEC) additives into the electrolyte dramatically reduced Li microstructural growth.²⁰⁴

Owing to limitations in terms of cell design and compatibility, the application of in situ EPR to SSLBs is at an early stage. In situ EPR technique is expected to investigate surface/interface issues in LLZO-based SSLBs with continuous development and improvement in the future. In particular, in situ EPR can provide semi-quantitative information on time-resolved Li metal deposition and can be an important supplement to in situ OM, EM, and NMR. This technique shows potential for real-time monitoring of Li plating/stripping. In addition, EPR imaging is still limited to micrometer resolution, and requires long times of >5 days for image collection. With further improvement in resolution, interface evolution can be imaged in real time to explore unknown information on Li dendrite nucleation sites.

	Fig. 24 (a) Schematic diagram for NDP investigations. (b) Schematic of the in situ NDP system. (c) Configuration and digital image of an in situ cell. Reproduced from ref. 209 with permission from the American Chemical Society. (d) Setup and sample configuration for in situ NDP measurements with top and bottom dual Si detectors. Reproduced from ref. 212 with permission from Elsevier.

Fig. 24b illustrates the setup for in situ NDP measurement.²⁰⁹ An in situ cell is connected to a temperature-controlled aluminum plate and disc in a vacuum chamber. ⁴He and ³H particles formed by neutron beam and ⁶Li reaction are detected using a Si detector. In general, the in situ NDP measurement setup is tested at higher temperatures (such as 90 °C) to ensure large Li⁺ transfer during short timespans, which permits higher applied current densities. Fig. 24c displays the configuration and digital image of a symmetric/asymmetric in situ cell. The symmetric cell mimicked the structure with a Li metal anode and a sulfur-type Li-free cathode, while the asymmetric cell mimicked the structure with a Li-free anode and a Li-rich cathode. Li metal was melted onto the LLZO surface. During melting, a Ti strip with a punched hole was attached to the Li electrode, functioning as both a mask and a current collector. CNT films were coated on the LLZO surface facing the detector by a solution-based process. Approximately 50 nm thick nickel was deposited by electron beam evaporation on the electrode to ensure favorable electrical contact. Meanwhile, an approximately 7.6 μm thick Kapton film blocks ⁴He particles and improves depth resolution. Without the blocking effect of Kapton films, ⁴He and ³H signals would overlap in the low channel range, and significantly affect depth resolution of NDP measurements. Fig. 24d illustrates the setup and sample configuration for in situ NDP measurements with top and bottom dual Si detectors.²¹² Thin Cu/Au composite bilayers of current collection were deposited on both sides of SSEs by magnetron sputtering with thicknesses of Cu and Au being 30 and 10 nm, respectively. In a vacuum chamber, the in situ cell was fixed to a sample holder made of polyethylene terephthalate glycol (PETG). The inclination angle of the sample holder was ∼5° with respect to the neutron beam plane. Dual detectors enable simultaneous measurement of NDP spectra on both sides of the cell.

The in situ NDP technique is mainly used to investigate the electrolyte/electrode interface behavior. It is expected to address the issues of high interface resistance and low interface stability in LLZO-based SSEs. At first, in situ NDP quantitatively analyzed the spatial heterogeneity during Li plating/stripping.²¹³ Results indicated that initial plating current density significantly affects subsequent cycle processes. Li deposition density is high at higher initial plating current density, while low density and thick surface Li deposition cause massive “dead Li” generation at lower density. The plating/stripping behavior of Li in the Li |LLZO| Li symmetric cell and Li |LLZO| 3D CNT asymmetric cell was monitored in real time using in situ NDP technique.²⁰⁹ CNT was adopted as a Li storage structure for mimicking Li-free metal anodes. As shown in Fig. 25a and b, the counts of Li significantly increased only near the end of cycling in a symmetric cell, while the counts increased and gradually accumulated at the beginning of cycling in an asymmetric cell. Such a behavior is attributed to the formation of a reversible layer around the LLZO/CNT interface, where reversible Li plating/stripping occurs. However, beyond this layer, the reversible behavior is lost, and Li becomes “dead Li” that constantly accumulates and grows in CNT. Since CNT surface energy ranges between ∼2150–2500 keV, the counts around 2375 keV increased significantly. Moreover, the short-circuit behavior of the symmetric cell was predicted from changes in in situ NDP spectra. The integrated NDP counts close to zero indicated that Li is reversibly plated/stripped in the cell. In contrast, accumulation of Li was induced by an unstable electrochemical behavior, resulting in increased integrated NDP counts. An abnormal increase in NDP counts occurred before the voltage profile changed significantly, which led to a short circuit in the cell. Interestingly, the cell experienced unilateral short-circuiting if NDP counts decreased slightly at each stripping stage but increased significantly at each plating stage. The Li deposition behavior in Li |LLZO| 3D Ti cells was also monitored in real-time using in situ NDP technique.²¹⁴Fig. 25c displays the data that show a rapid increase in counts with time at 2560–2630 keV, indicating the growth of Li in voids. The results suggest that Li is preferentially deposited in the voids of the 3D Ti electrode, which lowers the interfacial stress induced by electrode volume variation and mitigates Li dendrite formation. Hence, an efficiently designed 3D mixed electronic/ionic conductive skeleton can buffer volume expansion, adjust the Li deposition behavior, enhance the contact area, reduce diffusion distance, and suppress Li dendrite formation. In addition to the work on LLZO-based SSLBs, the lithiation/delithiation behavior in LCO |Li₃PO₄| Si thin film cells was revealed by operando NDP.²¹⁵ A Li immobilized layer was formed at the electrolyte/anode interface during cycling as Si migrates from the anode into the electrolyte. As the cycle proceeded, the Li immobilized layer grew constantly at a lower speed, which resulted in a reduction in the amount of movable Li⁺. Accordingly, the cell was not capable of extracting a sufficient amount of Li during normal operation and the storage capacity thus significantly decayed. Interface modification strategies are beneficial to realize reversible and homogeneous Li plating/stripping. Depositing an Al₂O₃ layer on the LLZO surface by atomic layer deposition (ALD) technique resulted in the formation of an interface region containing high Li concentration compared to bare LLZO.²¹⁶ High Li concentration ensures consistent charge transfer at anode/electrolyte interfaces, enabling excellent cell-rate performance. Similarly, a ZnO layer has been deposited on the current collector surface using the ALD technique.²¹⁷

	Fig. 25 In situ NDP measurement of (a) the Li |LLZO| CNT asymmetric cell and (b) the Li|LLZO|Li symmetric cell while cycling. Reproduced from ref. 209 with permission from the American Chemical Society. (c) Contour map of in situ NDP spectra of the Li |LLZO| 3D Ti cell collected during cycling. Reproduced from ref. 214 with permission from Elsevier. (d) Comparison of the Triton edges in various states corresponding to in situ NDP spectra. Reproduced from ref. 212 with permission from Elsevier.

The electrochemical deposition behavior of Li in Cu (with/without the ZnO layer) |CE| Li cell was investigated using operando NDP technique. Li concentration and plating thickness on a bare Cu current collector were remarkably increased during the cycle, indicating that inactive Li species were accumulated on the CE side. In contrast, a ZnO layer-coated Cu current collector improved the lithiophobic behvaiour and mitigated the accumulation of inactive Li species. The plating/stripping efficiency calculated from Li concentration distribution was 45% for the Cu current collector and 80% for the ZnO layer coated Cu current collector.

Researchers tend to concentrate more on ionic conductivity of SSEs in SSLBs while neglecting the effect of electronic conductivity. Dynamic evolution of Li concentration distribution of SSEs in LCO |LiPON| Cu, Li |LLZO| Cu and Li |LPS| Cu cells during Li plating was monitored in real time with in situ NDP technique.²¹⁸ Massive Li deposits nucleated and grew into Li dendrites inside LLZO and LPS. In contrast, Li concentration in LiPON did not significantly vary, suggesting the absence of internal Li dendrite formation. The high electronic conductivity of LLZO and LPS is a major cause of internal Li dendrite formation. Therefore, identifying a source of high electronic conductivity and lowering the value is imperative. Furthermore, in situ NDP technique was utilized to track Li concentration in LCO |LiPON| Cu and LMO |LiPON| LNO thin film cells to understand Li migration.²¹⁹ In LCO |LiPON| Cu cells, LCO was synthesized from 100% ⁶Li, indicating that the intensity of ⁶Li in the cathode was higher than that in electrolyte. During the charge process, intensity of Li in LCO was dramatically reduced, which is consistent with the depletion of Li. Notably, Li depletion at the cathode/electrolyte interface was more intense, which was attributed to ⁶Li migration from LCO to electrolyte and charging reactions. The NDP spectra of charged and discharged LMO |LiPON| LiNiO₂ cells showed the largest variation in Li concentration at both electrodes and the smallest variation at the electrolyte. The distribution of Li concentration in the cell is non-uniform and concentrated around the sample center. Consequently, in situ/operando NDP technique enables understanding of Li⁺ migration pathways and spatial distributions inside the cell via determination of the changes in Li concentration. Based on these features, the route of Li⁺ migration pathways in CE along polymers and inorganic ceramics or both interfaces can be determined by in situ/operando NDP technology. Recently, Li transport in a novel and prospective Li conductive glass-ceramic electrolyte Li₂Al_0.5Ti_0.75Ge_0.75Si_0.5P_2.5O₁₂ (LICGC) was detected using in situ NDP technique.²¹² As shown in Fig. 25d, Li concentration at the interface between LICGC and Cu/Au bi-layer collector was dramatically reduced when +2.8 V was applied. Generation of a Li-depleted region was attributed to the formation of space charge layers at the interface and this was the first time that a practical distribution of Li atoms during space charge layer formation was observed. Interestingly, Li was gradually returned to a Li-depleted region when the bias voltage was eliminated. In addition, the changes in the Li migration and diffusion behavior with the application of reverse voltages required further investigation.

In summary, in situ/operando NDP technique achieves real-time monitoring of Li plating/stripping at electrode/electrolyte interfaces, contributing to an in-depth comprehension of the interfacial behavior in LLZO-based SSLBs. Moreover, thanks to high temporal and spatial resolution and high sensitivity, the Li deposition density can be quantified and Li migration pathways and spatial distributions can be analysed. However, NDP technique generally requires testing under vacuum and thus the reliability of the results in comparison to real-cell environments may not be accurate.

	Fig. 26 (a) Schematic diagram of an in situ NI electrochemical cell. (b) Experimental setup for NI measurements. (c) Schematic diagram of 2D radiography and 3D tomography. (d) 3D evolution of the Li distribution in the cell at different stages of charging and discharging. Reproduced from ref. 220 with permission from the American Chemical Society.

In situ NI technique revealed the growth mechanism of Li dendrites and distribution dynamics during cell short-circuits.²²⁰Fig. 26d shows the schematic diagram of an in situ cell setup and 3D evolution images of Li distribution during cycling. In situ NI images coupled with electrochemical profiles confirmed that internal short-circuiting of the cells was initiated from Li dendrite growth between the LMO cathode and Li anode. A competition mechanism between self-discharge and charging after a cell short-circuit is proposed. During the charging process, contact between Li and Li_1−xMn₂O₄ leads to a short-circuit. As a result, further charging is not possible and the voltage drops. Due to voltage differences, Li_1−xMn₂O₄ obtains electrons to form the LMO cathode, resulting in contact loss with the Li dendrite. Cell self-discharge is thus terminated and then cell undergoes recharging as indicated by the increase in voltage. Such a mechanism can reasonably explain the abnormal voltage drop/rise after the cell short-circuit. Thus the use of this technique provides an in-depth insight into short-circuiting induced by Li dendrite formation along with the associated effects which need to be monitored to ensure a safe design and operation of the cell. The effect of Li morphology evolution on cell electrochemical performance was clarified by in situ NI and XRT techniques.²²¹ Due to the electrochemical side reactions, irreversible structure transition of Li leads to performance degradation. A greater comprehension of cell potential degradation and failure mechanisms can provide relevant information for improving the future cell design. Unfortunately, in situ NI technique is at a very early stage for application in solid-state batteries. With further development of this in situ technique, it is expected that the growth mechanism of Li dendrites as well as migration pathways and spatial distributions of Li in LLZO-based SSLBs can be determined to a higher precision and detail in the future.

Besides NDP and NI techniques, NR technique has been utilized to explore surfaces/interfaces with the aid of total internal reflection phenomenon. Due to high penetration, non-perturbation, and high sensitivity of neutrons, in situ NR technique exhibits remarkable advantages for identifying thin film layers (such as the SEI or modified artificial layer) on surfaces/interfaces. For example, formation of SEI films at the solid–liquid interface was investigated by in situ NR.²²² SEI film generation was spontaneous and featured a bi-layer structure. Low scattering length density of the inner layer suggested that Li₂O was the major component. In contrast, scattering length density of the outer layer was proportional to that of solution indicating that solution infiltrates pores or layers containing solvated H. Therefore, in situ NR technique is expected to be employed for investigating the evolution of thin film layers on the surface/interface in LLZO-based SSLBs during electrochemical response. Notably, the requirements of intense signals, long analysis times (namely, low temporal resolution), and constrained characterization facilities limit its extensive application. SANS technique provides information on the size and morphology of nanostructures based on the scattering length density fluctuation being on the order of 1 nm–0.5 μm. The growth of Li filaments in LLZO was monitored in real-time on a nanoscale with operando SANS technique.²²³ Growth of Li filaments was not seen to occur just by accumulation but relies on competition between the growth rate and self-healing rate. Increasing the temperature can strengthen self-healing ability, which positively suppresses Li filament growth. Accordingly, heat treatment assists in enhancing the cell electrochemical performance. Overall, neutron based in situ techniques are critical in LLZO-based SSLB investigations (such as evolution of surface/interface structures, growth of Li dendrites, and spatial distributions of Li species) due to the high sensitivity to Li and high penetration depth. However, neutron flux is too small to analyze rapidly and the neutron source is too expensive for large-scale laboratory applications. Moreover, the low abundance of ⁶Li increases the experimental complexity. More importantly, neutron technique-analyzed samples exhibit radioactivity, posing threats to humans and the environment and these issues need to addressed in the future research.

Coupled ion etching and ToF-SIMS techniques for depth orientation measurement are employed most frequently. The application scenario and derived outcomes are comparable to those of XPS depth profiling tests. The preparation of LLZO thin films using CO₂-laser assisted chemical vapor deposition was studied using ToF-SIMS technique.¹⁴⁹ The real elemental distribution in thin films can be detected by monitoring the variation of positive/negative ionic fragments related to LLZO and substrate in-depth orientation. The homogenous elemental distribution and intensity that matches the LLZO stoichiometry ratio suggested that LLZO thin films were successfully prepared. The film thickness was quantitatively analyzed via multiplication of the sputter time and rate. In addition, the distribution of doped elements in LLZO, as well as composition and thickness of the surface contamination layer, were also investigated using ToF-SIMS technique. At the anode/LLZO interface, introduction of different artificially modified interlayers was observed using ToF-SIMS technique.^{138,165,225–227} As shown in Fig. 27a, a LiPAA layer was successfully formed on the LLZO ceramic pellet surface.¹⁶⁵ The good flexibility of the LiPAA layer relieves interfacial stresses and maintains interface contact. Meanwhile, the electronic insulation of the LiPAA layer shields LLZO against electronic degradation, resulting in suppression of Li dendrite formation in the interior of LLZO. ToF-SIMS tests revealed that the lithiophobic behaviour of LLZO is intimately correlated to surface contaminants of Li and LLZO.^146,228 Li surface contaminants mostly consisting of Li₂O, Li₂CO₃, LiOH or LiF are the main causes of poor interface contact. Although LLZO surface impurities increase the interfacial resistance, their effect is less significant compared to Li surface impurities. An effect of introducing an artificially modified interlayer is to determine the impact of Li and LLZO surface impurities. At the cathode/LLZO interface, elemental inter-diffusion and decomposition phenomena during high-temperature melting or annealing have been revealed by ToF-SIMS technique.¹⁰⁰ Near the interface, LLZO was converted from a cubic to tetragonal phase while interlayers such as LBO or lithiated Nb₂O₅ minimized elemental cross-contamination and stabilized the cubic phase.^229,230 In addition to depth orientation measurement, cross-sectional ToF-SIMS mapping images of CE-based cells before and after cycling indicate that Li⁺ preferentially migrates along the amorphous phase in the polymer matrix.²³¹ Furthermore, ToF-SIMS mapping images also reveal the presence of Li dendrites in LLZO ceramic pellets after cycling.²⁸

	Fig. 27 (a) ToF-SIMS depth profiles and 3D views of the sputtered volume for LLZTO@PAA pellets. Reproduced from ref. 165 with permission from Springer. (b) Pressure-dependent Nyquist plots show external force's large impact on the interface resistance and capacity values. Reproduced from ref. 234 with permission from the American Chemical Society. (c) Configuration for operando acoustic transmission. Reproduced from ref. 244 with permission from Elsevier. (d) Fixed pressure configuration for operando acoustic transmission experiments. (e) Normalized acoustic amplitude (intensities relative to first waveform intensity) at 2, 7.4 and 13 MPa stacking pressures. Reproduced from ref. 179 with permission from Springer.

When combined with FIB technique to investigate interior compositions, a xenon ion source should be used as it provides low ionic contamination, high sputtering yield, low scattering and greater precision for Li mapping compared to a Ga ion source.²³² However, milling of the surface obtained using a xenon ion source is not as smooth as seen when using a Ga ion source. Moreover, in situ ToF-SIMS technique that can measure under an electrochemical response is not yet available. Therefore, with continuous development, in situ ToF-SIMS technology is expected to be employed to monitor the real-time evolution of surfaces/interfaces during electrochemical cycling.

Recent in situ EIS research on LLZO-based SSLBs mainly concentrated on the interfacial properties between LLZO and Li anodes or cathodes. On the anode side, in situ EIS examination revealed that vacancy diffusion in Li metal constrains the rate capacity of the Li anode.²³⁴ Contact loss is induced by void accumulation and pore formation at the LLZO/Li interface, resulting in high interface resistance. As current density increases, both the increased rate of interface resistance and the growth rate of Li dendrites are enhanced.²³⁵ In order to lower the interface resistance, several strategies, including application of high stacking pressure, construction of Li alloy anodes, and introduction of artificially modified interlayers have been investigated using the in situ EIS technique. First, as shown in Fig. 27b, interfacial resistance decreased to nearly 0 Ω cm² at a high enough stacking pressure of ∼400 MPa.²³⁴ When the pressure was removed, the interface resistance was preserved. In addition, under high stacking pressure, a decrease/increase in interface resistance is reversible during plating/stripping. However, this cannot prevent Li dendrite formation intrinsically. Notably, high temperatures promote diffusion and plastic deformation of Li. Therefore, increasing temperature modestly will reduce the required stacking pressure, which will lower the cell manufacturing costs. Moreover, use of Li alloys such as Li–Mg alloy eliminates interface pores to lower the interface resistance.²³⁶ However, the Li–Mg alloy system suffers from kinetic limitations, namely, in terms of diffusion-controlled decrease of interface Li metal concentration. Analogously, in combination with in situ EIS and DRT techniques, a study of Li–In alloy system suggested that interface charge transfer and Li diffusion are the critical kinetic steps that assist in stabilization of the interface.²³⁷ Such kinetic characteristics are seen in almost all Li alloys. Increasing the Li content in Li alloys can induce anode potential ≥0 V and a low charge transfer barrier for Li rapid diffusion, which is vital to stabilize the interface. Lastly, introducing alloy elements to form an alloy interlayer helps to construct a fast charge transfer pathway at the interface by changing the critical kinetic step at the interface from Li diffusion to Li alloying.⁶² However the kinetics of Li plating/striping are not changed fundamentally. The interface resistance will still increase after long-term cycling, which results in cell failure.

Formation of inorganic substances such as SnF₂ at the interface in situ in the form of a conformal interlayer was seen in the interactions of Li–Sn alloy and LiF.²³⁸ Since this layer is able to reinforce the adhesion to Li metal and has the function of an alloy layer, interface resistance is minimized. Electronically-insulating and Li⁺-conducting LiF layers suppressed internal Li dendrite growth initiated by electrons from Li metal into LLZO. On the cathode side, trace amounts of Li₂CO₃ on LLZO surfaces are decomposed at high potential (>3.8 V).²³⁹ The interface resistance was substantially elevated due to the release of CO₂ and O₂ from the LLZO/cathode interface. Removal of Li₂CO₃ by soaking LLZO in a LiBF₄-based liquid electrolyte in a glovebox before cell assembly was demonstrated using EIS technique.²⁴⁰ The reaction products are CO₂ gas and soluble B₂O₃ and LiF, and the interface resistance is dramatically decreased. Therefore, assembling an effective and contaminant-free LLZO-based SSLB is necessary. In addition, at high temperatures, an apparent cycle-induced reaction between LMO cathode and LLZO was revealed by utilizing in situ EIS technique.¹³⁴ Due to migration of the Mn element from LMO to LLZO, the interface resistance enhanced and the capacity decayed. The introduction of an artificially modified interlayer is thus an effective approach to mitigate element diffusion. Since EIS technique cannot be used to image the interface in real-time, in situ imaging techniques are required as complementary parts for analysing interfacial characteristics.

Fig. 27c illustrates a configuration utilized for in situ acoustic transmission.²⁴⁴ The LLZO sample is placed between two Li metal electrodes and a transducer is positioned on one side of the LLZO sample to ensure that only LLZO is affected. Variations in elastic properties and microstructures of LLZO are investigated using sound waves generated from a transducer. The results show that wave speed (LLZO stiffness) is reduced within a few minutes before failure, indicating occurrence of fracture in LLZO. Moreover, the lowering in the rate of LLZO stiffness is proportional to the applied current density. Consequently, in situ/operando acoustic characterization exhibits an ability to detect impending failure in LLZO-based SSLBs. Based on such characteristics, it can be adapted for on-line battery management systems to locate and isolate failed or failing cells. As shown in Fig. 27d, cell testing under different stacking pressures is realized by further modifying the configuration of operando acoustic transmission.¹⁷⁹ The stacking pressure is precisely controlled utilizing a double-piston pneumatic cylinder. At 2, 7.4 or 13 MPa stacking pressure, remarkable attenuation of acoustic amplitude occurred and this was attributed to microstructural changes (Fig. 27e). To be specific, amplitude attenuation at low stacking pressures is predominantly attributed to void formation, while that at high stacking pressure is attributed to fracture crack development. In contrast, a stable increase in acoustic amplitude prior to attenuation was attributed to improved contact and reduced cell thickness as Li foil plastically deformed over time. Given the intrinsic degradation mode at high stacking pressures and long duration of cycling, the cells operating at low stacking pressure are more favoured. Notably, in situ/operando acoustic characterization requires a non-vacuum and low-noise environment. Moreover, this approach is currently unavailable for 3D imaging of the interface structure.

	Fig. 28 (a) The schematic diagram of an experimental setup for strain monitoring. (b) The surface strain evolution of the anode-free pouch cell during cycling. Reproduced from ref. 247 with permission from Wiley. (c) Schematic illustration of the half-cell used for in situ curvature measurements. (d) Schematic illustration showing how the curvature evolution is measured during plating. Reproduced from ref. 251 with permission from Wiley. (e) Li, La, Zr, O and Ta atomic distribution curves derived from APT tests. (f) Site-specific sample preparation of the APT needle-shaped specimens. Reproduced from ref. 253 with permission from Elsevier.

Accordingly, this technique is available for use in battery management systems to understand the cell's internal health and electrochemical changes. In addition, an FBG integrated inside a Swagelok-type cell was employed to investigate Li–In |LPS| LTO or Li–In |LPS| Li–In cells.²⁴⁸ The viability of accessing orientation anisotropy in Li-driven stress field was proven by birefringence phenomenon. Interestingly, FBG failed to observe responses to stress changes at the electrode level, while it could successfully track stress changes at the electrode/electrolyte interface level. Such phenomenon indicates that internal stress monitoring is preferable for solid-state battery systems rather than external stress monitoring which is generally adopted for liquid batteries. Subsequently, the strain changes in CE-based Li symmetric cells have been monitored in real-time by using an internal stress monitoring system.²⁴⁹ The strain changes are attributed to Li dendrite generation at the interface. The evolution of side-view strain distributions during charging in LLZO-based SSLBs was measured in situ using a digital image correlation technique (DIC).²⁵⁰ The overall strain remarkably increased with SOC or current density increases. Based on principal strain analysis, the Li⁺ migration pathway was seen to be random and affected by mechanical or current density inhomogeneities.

Recently, a novel experimental configuration (Fig. 28c and d) was designed to in situ monitor the displacement induced by mechanical stresses through the measurement of wafer curvature.²⁵¹ This technique is capable of investigating the chemo-mechanical evolution during electrochemical cycling. A thin Al₂O₃ layer is plated between LLZO and melted Li utilizing atomic layer deposition to improve interface contact. A CCD detector is employed to detect a set of parallel laser beams reflected off the current collector. The in situ curvature value is derived from spacing variation of the parallel laser beam. During plating, the stress (displacement) gradually increased and reached a steady state over several hours. Moreover, cell short circuit was directly correlated with well-defined pressure variations. Results indicate that failure originated from inherent defects in LLZO and thus minimizing the inherent defects during manufacturing is essential. However, there are very few in situ analyses of the stress evolution mechanisms related to the field of solid-state batteries and thus research into the chemo-mechanical evolution of cells is expected to attract more attention in the future.

In summary, each in situ/operando technique is beneficial for researching the scientific and technical challenges faced by LLZO-based SSLBs at various temporal and spatial resolutions. However, each technique exhibits strengths and limitations and thus combining different testing methods together will be critical to obtain a comprehensive picture of the phenomena occurring at different sections of the battery system.