Revisiting the original 2003 garnet-like Li-ion conducting solid electrolytes Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta): a look into phase formation and the identification of carbonate and alumina contamination

Abstract

Garnet-type electrolytes are a promising class of solid-state oxide materials for next-generation lithium batteries. In this study, parent-phase garnets, Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta), were synthesised via solid-state reaction at 900 °C and 1100 °C, enabling comprehensive structural and chemical stability characterisation. Powder X-ray diffraction (PXRD) and Raman spectroscopy confirmed single-phase garnets with Ia3̄d symmetry, while scanning electron microscopy (SEM) imaging revealed improved densification at 1100 °C. Electrochemical impedance spectroscopy showed the high ionic conductivities of 4.8 × 10⁻⁵ S cm⁻¹ at 55 °C for Nb and Ta co-substituted Li₅La₃NbTaO₁₂ prepared at 1100 °C. X-ray photoelectron spectroscopy and thermogravimetric analysis identified a surface lithium carbonate layer formed under ambient conditions, which was not discussed in the original garnet reported in 2003. Solid-state nuclear magnetic resonance spectroscopy provided an insight into the lithium environment in the surface and bulk of the samples and confirmed aluminium contamination in samples sintered at the base of alumina crucibles at 1100 °C, with LaAlO₃ identified as the dominant secondary phase, corroborated by PXRD, SEM, and energy dispersive X-ray spectroscopy analysis. Nb-based garnet showed the most severe reaction with the alumina crucible. The use of a sacrificial mother powder of targeted garnet oxides, an approach commonly used for the preparation of Li-garnets, or sintering at 900 °C effectively reduced Al contamination. This work delivers a detailed evaluation of the parent-phase garnet, offering renewed insights into phase and structure two decades on from its initial development in 2003.

---

1. Introduction

The capabilities of lithium-ion batteries (LIBs) have advanced significantly since their first commercial use in the 1991 Sony camcorder. While conventional LIBs have continued to push the boundaries of energy storage, they still face several limitations that must be overcome to meet rising energy demands and support the global shift toward sustainable energy technologies. One promising route for LIB innovation lies in the development of new electrolyte materials, by replacing the traditional liquid electrolyte with an ionically conductive solid, which could enable next-generation battery technologies capable of higher energy and power densities.¹ Solid-state electrolytes (SSEs) offer numerous advantages over conventional organic liquid electrolytes, including improved safety. Inorganic oxide-based SSEs exhibit high thermal stability and significantly reduce the risk of thermal runaway, which is a critical safety concern in current LIB systems.² Additionally, SSEs reduce unwanted chemical reactions between electrodes, which can occur in liquid systems, where soluble species are able to migrate across the cell, leading to material loss, self-discharge, and long-term degradation.³ Over the years, extensive research has led to the exploration of many inorganic SSEs such as oxides, halides, and sulfides, each with distinct advantages.² Among them oxide-based electrolytes have emerged as one of the most favourable classes for future solid-state lithium battery development, including the NASICON, perovskite and garnet-type electrolytes.² Garnet-type materials with excess lithium compared to the ideal stoichiometry have gained significant attention in the solid-state battery field due to their high ionic conductivities. The most extensively studied Li-conducting garnets typically contain between five and seven lithium atoms per formula unit and are commonly known as Li-stuffed or Li-rich garnets.⁴

In 2003, Thangadurai et al.⁵ reported lithium-ion conducting original garnet-type oxides with nominal composition Li₅La₃M₂O₁₂ (M = Nb, Ta). This work highlighted that Li₅La₃Ta₂O₁₂ is chemically stable to lithium metal, which is a major advantage over many of the other promising solid electrolytes, such as La_0.67−xLi_3x□_0.33−2xTiO₃ (□ = A-site vacancy; x = 0.1 member exhibits bulk ionic conductivity of about 10⁻³ S cm⁻¹ at room temperature) and Li_1.3Ti_1.7Al_0.3(PO₄)₃.⁵ Cussen demonstrated through neutron diffraction study the three-dimensional network of Li diffusion within the garnet structure and highlighted the role of partially occupied octahedral sites in facilitating Li mobility.⁶ The combination of high ionic conductivity, chemical stability, and compatibility with lithium metal attracted significant interest, and over two decades on from their initial conception, the Li₅-composition has earned the title of ‘parent-phase lithium garnet’ electrolytes. Cerium-doping in the parent phase Li₅La₃Nb₂O₁₂ was shown to enhance ionic conductivity by nearly two orders of magnitude, representing the highest value reported among the Li₅ garnet family.⁷ The development of further garnet-like structures with enhanced conductivities was achieved by partially substituting the trivalent La³⁺ ions with divalent cations, leading to an increased lithium-ion concentration within the structure. This modification gave rise to a new subset of garnet-type electrolytes with the general formula Li₆ALa₂M₂O₁₂ (A = Mg, Ca, Sr, Ba, Sr_0.5Ba_0.5).⁸ A breakthrough in highly conducting garnet-type realised in 2007 when Murugan, Thangadurai and Weppner reported the cubic Li₇La₃Zr₂O₁₂ (LLZO), which showed a bulk conductivity of 3 × 10⁻⁴ S cm⁻¹ at 25 °C, over an order of magnitude higher ionic conductivity than that of parent Li₅La₃M₂O₁₂ garnets.⁹ Karasulu et al.¹⁰ used experimental and computational ab initio analysis to understand the structure of doped and undoped cubic LLZO and demonstrated that Al and Ga dopants preferentially occupy Li1 sites in garnet, giving rise to distinct NMR signatures associated with local distortions.

Following this discovery, subsequent studies reported the formation of a less conductive tetragonal phase when sintering was performed at lower temperatures and in Pt crucible,¹¹ indicating that the garnet-type structure is highly dependent on both the synthesis and processing conditions. The development of the Li₇-phase marked a significant advancement, offering enhanced ionic conductivity and sparking a wave of research into the optimisation and application of garnet-type electrolytes, a field that is actively explored today.^12–14 The surface chemistry of an SSE directly impacts the solid–solid interfacial properties in a solid-state battery, which in turn can significantly impact the battery performance. Initially, garnet-type electrolytes were thought to be chemically stable in ambient conditions; however, it has now been highlighted that surface layers of Li₂CO₃ formation in air results in low ionic conductivities and low densities.¹⁵ Further, the Li₂CO₃ layer is known to affect the wettability of Li metal on the electrolyte surface,¹⁶ which creates challenges during cell assembly.

The aim of this revisitation is to conduct a comprehensive investigation of the parent-phase garnet electrolyte Li₅La₃M₂O₁₂ to build a deep fundamental understanding of its structural, chemical, and electrochemical properties. This study offers a renewed evaluation of this emerging material through the synthesis of original Li₅La₃M₂O₁₂ and 1 : 1 solid solution of Li₅La₃Nb₂O₁₂ and Li₅La₃Ta₂O₁₂ and revisits the material purity through state-of-the-art solid-state characterisation methods. Whilst investigating surface contamination that arises during processing, particularly Li₂CO₃ formation due to air exposure, and examining intrinsic Al-doping introduced from alumina crucibles during high temperature processing, which have not been as widely studied for the parent phase Li₅-composition garnet, unlike Li-stuffed LLZO and its derivatives.

2. Experimental section

2.1 Synthesis

Standard solid-state synthesis procedure was followed to prepare the garnet materials. High purity (>99%) Li₂CO₃, La₂O₃ (pre-dried at 900 °C for 10 hours), Nb₂O₅ and Ta₂O₅ were obtained from Thermo Scientific (UK) as precursor materials. The required stoichiometric amounts of each material, including a 10 wt% excess of Li₂CO₃ (to account for lithium volatilisation upon heating), were mixed in propan-2-ol and roller milled with 5 mm diameter zirconia balls overnight. After evaporation of the solvent at room temperature, the powder was heated to 750 °C in air for 10 hours, then cooled down to room temperature. The calcined powder was roller milled again in propan-2-ol with 5 mm zirconia beads, and the resultant powders were further heated in air (900–1100 °C) with a heating and cooling rate of 5 °C per minute for 24 h. An intermediate milling step was conducted between any successive heating steps and powders were pressed into 13 mm pellets by uniaxial press before the final sintering step.

2.2 Characterisation by diffraction

Ex situ powder X-ray diffraction (PXRD) measurements at room temperature were performed on PANalytical Empyrean diffractometers, equipped with a solid-state X'Celeration detector, using monochromatic Cu-radiation. Measurements were made in the reflection geometry using flat sample holders. The samples were in either powder or pellet form. In situ high-temperature PXRD measurements were performed at variable temperatures using monochromatic Mo-radiation at STOE diffractometer, in the transmission geometry with a linear single-photon counting detector Mythen (DECTRIS). A high-temperature reaction furnace HT2 (STOE) was used with sapphire capillary tubes, containing up to 1 cm length of sample in the middle of the tube (ID approx. 1 mm), which was restricted on both sides by high-temperature quartz wool to prevent the movement of the sample. The measurements were performed under a very low Ar flow to create an almost static inert atmosphere. Program Topas Academic V7 ¹⁷ was used for Rietveld refinement analysis.^18,19

2.3 Characterisation by spectroscopy methods

Room-temperature Raman spectra were collected with a 532 nm wavelength laser in the Renishaw inVia™ confocal Raman microscope, with 1800 lines per mm holographic grating. Optical microscope with Leica 50× objective lens was used to focus on the samples, and the spectra were collected at 20 random spots on the surface of the sample with various μm distance between them and the average of these points was used in the graphs. Laser exposure time on the samples was 10 seconds, 5% of the laser power was used and 2 accumulations for each point were conducted to increase signal to noise ratio.

Scanning electron microscopy (SEM) images were obtained using the JSM-IT800 Schottky Field Emission Scanning Electron Microscope by JEOL coupled with JEOL Energy Dispersive X-ray Spectrometer (EDS). Images captured in both secondary and back-scattered electron detectors were made using 5 kV and 10 mm working distance. For elemental composition analysis, 15 kV and analysis current mode were used.

X-ray photoelectron spectrometry (XPS) measurements were collected in an ultrahigh vacuum, with base pressure ca. 1 × 10⁻¹⁰ mbar, chamber using a Kratos Axis Ultra DLD photoelectron spectrometer with monochromatic Al Kα (1486.6 eV) anode operating at 4 mA × 15 kV. The system was operated in hybrid mode with the slot aperture, giving a measuring area of ca. 700 micron × 300 micron. Charge compensation was used during data collection; the binding energy (BE) scale was referenced to adventitious carbon species at 285.0 eV and BE values are reported ± 0.1 eV. Peak fitting was carried out using the CasaXPS software,²⁰ with Kratos supplied sensitivity factors, using Gaussian–Lorentzian functions, after subtracting a Shirely-type background.

Solid-state NMR spectra were recorded using a Bruker Avance III spectrometer equipped with a 14.1 T wide-bore magnet. Powdered samples were packed into 3.2 mm outer diameter ZrO₂ rotors from Bruker and rotated at a magic angle spinning (MAS) rate of 20 kHz. The ²⁷Al MAS NMR spectra were recorded using a short-flip angle pulse (shorter than π/20) with a radiofrequency field strength of ∼100 kHz to ensure that the spectra are quantitative and are the result of averaging 256 transients with a recycle interval of 5 s. The ⁷Li MAS NMR spectra were recorded using a short-flip angle pulse (shorter than π/12) with a radiofrequency field strength of ∼106 kHz to ensure quantification and are the result of averaging 32 transients with a recycle interval of 10 s. ²⁷Al NMR spectra and ⁷Li NMR spectra are shown relative to 1.1 M Al(NO₃)₃ in D₂O and 1 M LiCl in D₂O, respectively. For the ²⁷Al NMR spectra acquisition, Al(acac)₃ (δ_iso = 0.0 ppm) was used as a secondary reference.

Scanning transmission electron microscopy (STEM) measurements were performed on an FEI Titan Themis operated at 200 kV, equipped with a CEOS DCOR probe corrector, a SuperX energy dispersive X-ray spectrometer (EDX) and a 4k × 4k Ceta CMOS camera. High-Angle Annular Dark-Field (HAADF) images were acquired with a probe convergence angle of 20 mrad and inner/outer collection angles of 60 and 200 mrad, respectively. Selected area electron diffraction patterns (SADPs) were indexed and fitted using Recipro.²¹ Samples for transmission electron microscopy were prepared by dispersion of powder in methyl isobutyl ketone (largely immiscible with water to avoid formation of lithium carbonate) before drop casting onto lacey C/Cu grids purchased from EM Resolutions Ltd, UK.

2.4 Characterisation by thermal and electrical measurements

TGA measurements were performed using a Netzsch STA449 simultaneous thermal analyser. Measurements were made under Ar atmosphere, with a heating rate of 5 °C per minute, and a background correction was applied using a prior baseline run to ensure accuracy of the measurements. The electrical conductivity of the samples was measured in air using a Li-ion blocking Au-electrodes. Au paste was coated on the pellet surface and cured in air at 700 °C for 2 hours. All measurements were made in the temperature range between 25 and 450 °C using an HP 4192A Impedance and Gain Phase Analyser (5 Hz to 13 MHz; 100 mV) interfaced with a PC. A two-probe measurement was used, and prior to each measurement, the samples were equilibrated for about 2–24 h at a constant temperature. All measurements were made during both heating and cooling cycles to check the reproducibility of the data. Dimensions of the garnet samples used for electrical measurements range from 0.11 to 0.15 cm in thickness and 1.2 to 1.32 cm in diameter.

3. Results and discussion

3.1 Ex situ powder X-ray diffraction

PXRD patterns of the parent Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) sintered at 900 °C (I) and at 1100 °C (II) together with the standard pattern Li₅La₃Nb₂O₁₂ reported on the Inorganic Crystal Structure Database (ICSD: 18586) are shown in Fig. 1. The PXRD peak positions are the same across the three compositions sintered at both 900 °C and 1100 °C, and all closely match the peak positions of the database reference without major impurity peaks. It is expected that the compositions with Nb, Ta, and the mixed Nb and Ta on the M-site have the same diffraction patterns as the six coordinate Nb⁵⁺ and Ta⁵⁺ have the same ionic radii of 0.64 Å,²² and each composition should have the same stoichiometric ratios of Li, La, M (Nb and/or Ta) and O. From the PXRD patterns, it can be concluded that at both sintering temperatures all samples form the Li₅-garnet, belonging to the Ia3̄d space group, as the main phase. Fig. S1–S3 (see SI) show the PXRD data for Nb, Ta and Nb/Ta phases synthesis and corresponding precursors compounds.

Fig. 1 Ex situ PXRD patterns of Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) prepared at 900 °C (denoted (I)) and at 1100 °C (denoted (II)) for 24 h with a heating and cooling rate of 5 °C per minute, together with the pattern of the inorganic crystal structure database for cubic Li₅La₃Nb₂O₁₂ (ICSD #: 18586) for comparison.

The unit cell parameters for all samples were found to be 12.73–12.77 Å. These values are comparable to the literature,^5,23 further confirming that cubic garnet structures were synthesised. Additionally, it was observed that the garnets containing Nb showed a slight decrease in lattice parameter at the higher sintering temperature, whereas the Ta only sample showed the same cell parameter at both sintering temperatures. The unit cell of the garnet may decrease due to the formation of small amounts of impurity as a secondary phase, preventing the formation of the garnet in the desired stoichiometric ratio or if a material decomposes at elevated temperatures, in turn changing the structure. However, the observed changes in the unit cell parameter are found to be rather small value, and therefore, the proposed compositional changes are only speculative, requiring further investigation.

3.2 In situ powder X-ray diffraction

A series of in situ PXRD experiments during solid-state synthesis was performed to clarify the influence of different parameters on the synthesis process. Interesting factors to consider were the variations in the number of intermediate annealing temperatures, their timing, and heating rates. As we used the laboratory PXRD for in situ measurements, the heating rates were mostly determined by the number of intermediate annealing temperatures, because the measurements took 30 minutes at each temperature step. Therefore, through careful experimental design, a series of multiple heating steps enabled the solid-state transformation to be followed in detail, thus helping to understand the solid state reaction mechanism that occurs between 700 °C and 1100 °C, and the necessity of the high temperatures during this synthesis.

Investigation of the phase formation of Li₅La₃Nb₂O₁₂ began using the mixture, which had been pre-annealed ex situ in the laboratory oven for 10 hours at 750 °C. Therefore, in the starting powder at room temperature (RT), a small amount of the targeted phase Li₅La₃Nb₂O₁₂ was present. Thus, even if considerable amounts of Li₅La₃Nb₂O₁₂ had formed during the ex situ pre-annealing step at 750 °C, it was largely decomposed to La₂LiNbO₆ (84 wt%) and Li₃NbO₄ (9 wt%) on cooling back to RT. The marginally small amount of targeted phase Li₅La₃Nb₂O₁₂ (4 wt%) as well as minor impurities La₂O₃ (1 wt%) of non-reacted high-temperature decomposition phase of La(OH)₃ and La₃NbO₇ (2 wt%, probably formed at 750 °C during ex situ pre-annealing and preserved during cooling to RT) were also observed in the starting pre-annealed ex situ material.

It is observed that the ex situ pre-annealed at 750 °C mixture quickly transforms before 700 °C (Fig. 2), giving 54 wt% of the targeted Li₅La₃Nb₂O₁₂ phase at 700 °C, while undesirable La₂LiNbO₆ completely disappears by 800 °C via decomposition to La₂O₃ and other possible amorphous phases. The target phase Li₅La₃Nb₂O₁₂ reaches 88 wt% at 1075 °C before unexpected complete decomposition into La₂LiNbO₆ and Li₃NbO₄ after 1075 °C. Although the very little impurity of La₃NbO₇ was formed during the pre-annealing step, it reacted with other phases before reheating during in situ PXRD to 700 °C. These results allow us to speculate that using a very high temperature of 1100 °C for the synthesis of Li₅La₃Nb₂O₁₂ may not be necessary, and extended annealing at a lower temperature can be preferable and more convenient.

Fig. 2 (a) 2D contour plot (logarithmic scale of intensities) for in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via slow heating of pre-annealed at 750 °C sample (heating rate 20 °C min⁻¹, 30 min per each scan); diffraction patterns were recorded during heating (25–1100 °C), isothermal at 1100 °C, and at 25 °C after cooling. Selected peaks are represented by the corresponding symbols as shown. (b) Weighted fractions of crystalline phases (determined by Rietveld refinement) during in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via slow heating of pre-annealed at 750 °C sample. The error bar of relative weighted fraction is smaller than the size of the symbol when not shown. The light blue shaded area is a guide to the eye to visualise the temperature range under which highest yield of the target phase was obtained.

Although the maximal amount of Li₅La₃Nb₂O₁₂ obtained during in situ PXRD experiment was 88 wt%, it is possible that the slightly different conditions, which are necessary for in situ measurement prevented 100 wt% transformation. During ex situ synthesis, the powder is pressed and a larger volume of the powder is used, while during in situ PXRD synthesis the powder was loose and a small amount of powder (cylinder of dia. 1 mm and of length 10 mm) was used. This scaling down of the material during synthesis may have prevented the full transformation because part of the powder may have stuck to the capillary tube walls and, therefore, was not involved in the reaction. To further understand whether the multiple annealing steps between 700 °C and 1100 °C (during which PXRD data were recorded) had a substantial influence on the transformation behaviour, the pre-annealed ex situ at 750 °C mixture was used for a different temperature profile: heating from RT to 700 °C for a short time to record one diffraction data at this temperature and subsequent heating up to 1100 °C and holding at that temperature for 4 hours while continuously recording diffraction data (Fig. 3).

Fig. 3 (a) 2D contour plot (logarithmic scale of intensities) for in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via heating to different temperatures and holding at 1100 °C for a prolonged time (heating rate 20 °C min⁻¹, 30 min per each scan); two heating, isothermal at 1100 °C for 4 hours, and cooling cycles are shown. Selected peaks are represented by the corresponding symbols as shown. (b) Weighted fractions of crystalline phases (determined by Rietveld refinement) during in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via heating to different temperatures and holding at 1100 °C for 4 hours. The error bar of relative weighted fraction is smaller than the size of the symbol when not shown. The light blue shaded area is a guide to the eye to visualise the temperature range under which highest yield of the target phase was obtained.

The room-temperature results for the freshly made ex situ pre-annealed mixture were almost identical to the previous mixture (Fig. 2 and 3), demonstrating 88 wt% of La₂LiNbO₆ and 8 wt% of Li₃NbO₄ with only 4 wt% of the targeted phase Li₅La₃Nb₂O₁₂. The first data point at 700 °C indicated that the transformation proceeded in the identical way to the previous experiment, with the targeted phase Li₅La₃Nb₂O₁₂ reaching 48 wt%. After that, the mixture was fast heated up to 1100 °C and measured at this temperature during hold. Very unexpected results indicate 91 wt% of targeted phase Li₅La₃Nb₂O₁₂ within 1 hour after reaching 1100 °C and subsequent complete decomposition of Li₅La₃Nb₂O₁₂ by the end of 3 hours hold at 1100 °C. The decomposition products were La₂LiNbO₆ (87 wt%) and Li₃NbO₄ (13 wt%), i.e. the same mixture was formed in situ after 3 hours hold at 1100 °C, as the starting mixture after ex situ pre-annealing at 750 °C. Using such a pre-annealed mixture allowed to synthesise the almost pure targeted phase Li₅La₃Nb₂O₁₂ (with restrictions imposed by scaling down the reaction for the in situ PXRD experiment). To check whether this mixture can again produce the targeted phase Li₅La₃Nb₂O₁₂, a second heating profile was used for in situ PXRD measurements (Fig. 3). Despite the same temperature profile for the second cycle, only decomposition products La₂LiNbO₆ and Li₃NbO₄ were observed without any variation in their amount. Attempt was made to synthesise the targeted phase Li₅La₃Nb₂O₁₂ during in situ PXRD measurements without an ex situ pre-annealing step at 750 °C in an ordinary laboratory oven. Two different experiments were performed with: (i) only 2 annealing steps (700 °C and 1100 °C) to simulate the standard ex situ synthesis process (Fig. 4) and (ii) multiple intermediate temperature steps to clarify the fine details of the solid-state synthesis process (Fig. 5).

Fig. 4 (a) 2D contour plot (logarithmic scale of intensities) for in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via heating to different temperatures and holding at 700 °C for 3.5 hours (heating rate 20 °C min, 30 min per each scan); two heating and cooling cycles are shown. Selected peaks are represented by the corresponding symbols (the same symbols as those for the corresponding relative weighted fractions). (b) Weighted fractions of crystalline phases (determined by Rietveld refinement) during in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via heating to different temperatures and holding at 700 °C for 3.5 hours. The error bar of the relative weighted fraction is smaller than the size of the symbol when not shown. The light blue shaded area is a guide to the eye to visualise the temperature range under which highest yield of the target phase was obtained.

Fig. 5 (a) 2D contour plot (logarithmic scale of intensities) for in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via slow heating (heating rate 20 °C min⁻¹, 30 min per each scan); diffraction patterns were recorded during heating (25–1100 °C) and at 750 °C and 25 °C during cooling. Selected peaks are represented by the corresponding symbols, as shown. (b) Weighted fractions of crystalline phases (determined by Rietveld refinement) during in situ PXRD during synthesis of Li₅La₃Nb₂O₁₂ via slow heating. The error bar of relative weighted fraction is smaller than the size of the symbol when not shown.

In situ PXRD measurements of as-prepared mixture (Fig. 4) contained at RT the mixture of starting materials Nb₂O₅, Li₂CO₃, La(OH)₃ (formed during preparation of the initial mixture from the correctly weighted La₂O₃ via hydration in air). Fast heating up to 700 °C indicated the fast formation of the targeted phase Li₅La₃Nb₂O₁₂ (25 wt%) with impurities La₂LiNbO₆ (7 wt%) and Li₃NbO₄ (9 wt%) and unreacted La₂O₃ (60 wt%). Subsequent annealing at 700 °C for 3.5 h had a minor impact on these formed phases. Ramping the temperature to 1100 °C allowed to increase the concentration of the targeted Li₅La₃Nb₂O₁₂ to 47%. Previous experiments (Fig. 2 and 3) demonstrated that longer annealing at 1100 °C would result in the complete decomposition of Li₅La₃Nb₂O₁₂, hence, the mixture was cooled to RT through one intermediate short temperature step at 700 °C to record one diffraction pattern. In situ diffraction data indicate almost complete decomposition of the targeted phase already on cooling to 700 °C (Fig. 4), which perfectly correlates with results from ex situ pre-annealing (first RT results on Fig. 2 and 3). Second identical heating cycle (Fig. 4) was performed for in situ PXRD with the same temperature profile, but the results indicated that the more stable phase of La₂LiNbO₆ does not participate in the reaction, and its amount remained almost unchanged during the second heating cycle. The amount of the targeted phase Li₅La₃Nb₂O₁₂ slightly increased during the second heating cycle to 5 wt%, but the complete decomposition of this phase was observed on the second heating to 1100 °C.

There is an acceptable correlation between completely in situ synthesis during PXRD measurements (Fig. 4) and in situ synthesis with ex situ pre-annealing step at 750 °C. The formation and decomposition of the targeted phase Li₅La₃Nb₂O₁₂ at 700 °C and the irreversible decomposition of the targeted phase Li₅La₃Nb₂O₁₂ during the second heating cycle, and similar decomposition phases and intermediate phase during all heat treatments were observed. The interesting observation is the possibility to synthesise the almost pure targeted phase Li₅La₃Nb₂O₁₂ from the mixture of La₂LiNbO₆ and Li₃NbO₄ after an ex situ pre-annealing step at 750 °C (Fig. 2 and 3), while using the same mixture after heating it up to 750 °C makes synthesis of the targeted Li₅La₃Nb₂O₁₂ phase impossible (Fig. 3 and 4).

To clarify the fine details of the solid-state synthesis process, the as-prepared mixture was used for in situ synthesis during PXRD measurements with multiple temperature steps (Fig. 5). As expected, the initial as-prepared mixture at RT (Fig. 5) contained the starting materials Nb₂O₅, Li₂CO₃, La(OH)₃ (formed during preparation from La₂O₃). The diffraction peaks from a crystalline Nb₂O₅ disappear after 500 °C. The formation of LiNbO₃ can be assigned to the reaction between Nb₂O₅ and Li₂CO₃ around 500 °C. At 650 °C, the transformation of LiNbO₃ into Li₃NbO₄ is observed; this transformation was previously observed during the synthesis of lithium niobates doped by transition metal.²⁴ An important observation is a substantial amorphisation of the starting materials (Li₂CO₃, La(OH)₃, Nb₂O₅) around 400 °C: all diffraction peaks at this temperature are very broad; La(OH)₃ is undergoing the transformation to crystalline La₂O₃ via a substantially amorphous intermediate state (Fig. 5). At 500 °C, the highly crystalline La₂O₃ and LiNbO₃ with low-crystalline Nb₂O₅ are observed. Subsequent heating up leads to decreasing the amount of LiNbO₃ at 650 °C (with subsequent disappearance after heating to 875 °C), formation of Li₃NbO₄, La₂LiNbO₆, and Li₅La₃Nb₂O₁₂. The targeted Li₅La₃Nb₂O₁₂ is formed only within a temperature range 650–750 °C. During this slow heating process, the amount of this phase remains very small due to the competitive formation of more stable Li₃NbO₄ and La₂LiNbO₆ phases. Thus, the very slow heating via multiple annealing temperature steps, to record the diffraction patterns at each temperature, changed the synthesis conditions considerably and prevent the detailed analysis of the synthesis process by in situ PXRD. It is important to underline that no quantifiable amount of LaAlO₃ was observed during in situ PXRD experiments.

To analyse the possible influence of the Ta substitution, additional in situ PXRD measurements were performed during synthesis of Li₅La₃NbTaO₁₂ (Fig. 6). The initial mixture was pre-annealed ex situ in the laboratory oven for 10 hours at 750 °C, and the resulting powder at RT showed a significant amount of targeted phase Li₅La₃NbTaO₁₂ (54 wt%) in contrast to all ex situ pre-annealed experiments for synthesis of Li₅La₃Nb₂O₁₂ (Fig. 2 and 3). Subsequent heating to 700 °C during in situ PXRD measurements allowed for an increase in the amount of targeted phase to 85 wt%, and further heating allowed achieving an almost pure sample with 94 wt% of targeted Li₅La₃NbTaO₁₂ at 975 °C. Further heating up resulted in the partial decomposition into more stable La₂LiNb_0.5Ta_0.5O₆ and other possible amorphous phases. Li₅La₃NbTaO₁₂ and La₂LiNb_0.5Ta_0.5O₆ were observed by in situ PXRD as crystalline phases at 1050 °C and above, as well as during subsequent cooling to RT. Still, a substantial amount of the targeted Li₅La₃NbTaO₁₂ (85 wt%) was observed at RT at the end of the in situ synthesis (Fig. 6), which contradicts the results for the targeted phase Li₅La₃Nb₂O₁₂ (Fig. 2–4).

Fig. 6 (a) 2D contour plot (logarithmic scale of intensities) for in situ PXRD during synthesis of Li₅La₃NbTaO₁₂ via slow heating of pre-annealed at 750 °C sample (heating rate 20 °C min⁻¹, 30 min per each scan); diffraction patterns were recorded during heating (25–1100 °C) and at 25 °C after cooling. Selected peaks are represented by the corresponding symbols, as shown. (b) Weighted fractions of crystalline phases (determined by Rietveld refinement) during in situ PXRD during synthesis of Li₅La₃NbTaO₁₂ via slow heating of pre-annealed at 750 °C sample. The error bar of relative weighted fraction is smaller than the size of the symbol when not shown. The light blue shaded area is a guide to the eye to visualise the temperature range under which highest yield of the target phase was obtained.

3.3 Raman spectroscopy

The phase of the synthesised garnets can be further explored using Raman spectroscopy, displayed in Fig. 7. Raman spectra can be split into three regions: (i) low-energy region below about 300 cm⁻¹, (ii) intermediate-energy region between 300 and 550 cm⁻¹ and (iii) a high-energy region above 550 cm⁻¹.²⁵ The high-energy region shows the vibrational stretching modes of the octahedral units, here the peak at 720 cm⁻¹ and 738 cm⁻¹ are assigned to Nb and Ta, respectively. The broad intermediate region consists of vibrational bending modes of octahedral and tetrahedral Li, and finally the low-energy region is assigned to the La translational stretching modes.²⁵ In addition to the garnet peaks, a high energy peak is observed at around 1075 cm⁻¹, which is characteristic of a carbonate vibrational stretching mode, indicating the presence of surface layer lithium carbonate contamination on a sample.

Fig. 7 Room temperature Raman spectra from the surface of Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) pellet, sintered at either 900 °C (I) or 1100 °C (II). Spectra are obtained from an average of 20 points measured randomly across the surface.

The Raman spectra further confirm formation of a cubic garnet structure across both sintering temperatures for each composition. This is evidenced by the presence of broad, characteristic peaks associated with the garnet structure.²⁵ Additionally, due to Raman spectroscopy's sensitivity to local bonding environments, subtle variations in M-site coordination can be resolved, allowing differentiation between Nb and Ta samples. The spectra indicate the successful formation of a mixed M-site garnet, which is not easily distinguishable by PXRD. It should be noted that for all compositions after sintering at 1100 °C, a shoulder peak emerges at 110 cm⁻¹, indicating the existence of small amounts of lower symmetry tetragonal phase.

3.4 Scanning electron microscopy

Microstructure and surface topography were investigated using scanning electron microscopy (SEM). The set of back scattered electron (BSE) images of the top pellet surfaces for each Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) garnet synthesised at both 900 °C and 1100 °C (Fig. 8) show the difference in microstructure across the samples. Across all compositions, samples exhibit a homogeneous microstructure with poorer densification observed when sintered at 900 °C compared to those sintered at 1100 °C, this effect is most prominent in Li₅La₃NbTaO₁₂ (Fig. 8b and e). Given that all samples were subjected to the same dwell time, the observed differences in grain size and density could be attributed to the higher thermal energy at 1100 °C, which enhances atomic mobility and promotes particle growth. Densification strongly influences the ionic conductivity of the material, as lower density and higher porosity materials typically result in reduced conductivity due to the presence of greater grain boundary resistance.²⁶ Furthermore, in the context of lithium metal battery applications, a porous solid electrolyte contains voids which can accelerate dendrite formation, compromising performance.²⁷ Therefore, the SEM analysis demonstrates that higher sintering temperatures lead to improved densification and grain growth, which are expected to enhance the ionic conductivity and electrochemical stability of the material. Additionally, the BSE imaging clearly reveals contrast associated with secondary surface contamination, namely Li₂CO₃ since carbon is found in EDS analysis (Fig. S4–S6, see SI). From the EDS mapping, it is apparent that all the elements are distributed evenly in the particles. Heat treatment of the pellets in Ar atmosphere is proven to be effective in removing surface contamination, which is crucial to subsequent steps, especially for lithium melting when preparing either full or symmetrical cells, since lithium carbonate renders the garnet surface highly lithiophobic.²⁸ The back scattered electron images of the pellets after removal of the lithium carbonate are shown in Fig. S7 (see SI).

Fig. 8 Backscattered electron detector (BSED) SEM images of the top surface of garnet pellets sintered at 900 °C (a) Li₅La₃Nb₂O₁₂, (b) Li₅La₃NbTaO₁₂, and (c) Li₅La₃Ta₂O₁₂ and at 1100 °C (d) Li₅La₃Nb₂O₁₂, (e) Li₅La₃NbTaO₁₂, and (f) Li₅La₃Ta₂O₁₂. The dark spots on the garnet surface are lithium carbonate.

3.5 Transmission electron microscopy

To assess the ordering over nanometric length scales, scanning transmission electron microscopy (STEM), was performed on a subset of samples. Fig. 9(a–c) show atomic resolution high angle annular dark field (HAADF) images and Fig. 9(d–f) the corresponding selected area electron diffraction patterns (SADPs) data from samples Li₅La₃Nb₂O₁₂, Li₅La₃Ta₂O₁₂ and Li₅La₃NbTaO₁₂, respectively. Data obtained from the 〈001〉-type zone axes, a high degree of local ordering can be seen, with all samples exhibiting the typical garnet-like ordering,²⁹ with a central La cation surrounded by four Nb/Ta cations, as depicted in the inset of Fig. 9(c). Based on measurements of the d-spacings, all samples exhibit an essentially cubic structure, although the measurements suggest a very slight tetragonality, with an a/b ratio (based on the nominal hkl indexing) of ∼1.01, consistent with the high symmetry observed by PXRD. Selected area electron diffraction data confirmed the absence of any superstructure reflection in all samples and no diffuse scattering, indicating a lack of any short-range ordering in all investigated samples.

Fig. 9 Scanning transmission electron microscopy data from Li₅La₃Nb₂O₁₂, Li₅La₃Ta₂O₁₂ and Li₅La₃NbTaO₁₂ samples showing (a–c) atomic resolution HAADF images and (d–f) SADPs, respectively. Both HAADF and SADPs were acquired from the 〈001〉-type zone axes of grains.

3.6 X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) analysis on the sample pellets sintered at 1100 °C and allowed to cool to room tempertaure (RT) confirmed the presence of the expected elements in each sample, as summarised in Table 1 and Fig. S8, see SI. The binding energies observed for each element are consistent across the three compositions, indicating similar chemical environments. For all garnets, the La 3d_5/2 core level peak is fitted with a single component at 833.0 eV (FWHM 2.23 eV) and exhibits the shake-up feature typical of La(III). Although on the lower binding energy side of the typical range for La(III), 833.8–835.2 eV,³⁰ La(III) in a mixed metals oxide matrix shows lower values of the binding energy,³¹ this is also confirmed by a Δj ca. 16.7 eV and the position of the satellite peak at 837.3 eV, giving a multiplet separation of ca. 4.3 eV from the parent peak, close to the expected value for pure La(III) oxide, La₂O₃, (4.9 ± 0.2) eV.³² In the Ta-containing garnet, the Ta 4d_5/2 core level peak is fitted with a single component at 228.7 eV (FWHM 3.6 eV) and shows a single chemical environment, within the expected range for Ta(V).²⁸ In the Nb-containing garnet, the Nb 3p_3/2 core level peak is fitted with a single component at 363.7 eV (FWHM 2.7 eV), and shows a single chemical environment, within the expected range for Nb(V).^31,33 This is preferred to the more intense 3d transition because it overlaps with La 4p_1/2. Similar fitting parameters were used to model the Nb 3p_3/2 and Ta 4d_5/2 core level peaks in the garnet containing both Nb and Ta, obtaining consistent results.

Table 1 XPS binding energies (BE) and atomic surface concentration (ASC) % for the surface of Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) pellets sintered at 1100 °C, measured at room temperature. BE values ± 0.1 eV

Element and orbital	Li5La3Nb2O12		Li5La3NbTaO12		Li5La3Ta2O12
	BE (eV)	ASC (%)	BE (eV)	ASC (%)	BE (eV)	ASC (%)
La 3d5/2	833.0	1.3	832.9	0.4	833.1	0.9
O 1s	531.7	47.8	531.8	49.9	531.7	45.3
	528.7	7.1	528.7	3	529.0	6.9
Nb 3p3/2	363.7	0.8	363.8	0.2	—	—
C 1s	289.9	15.7	289.9	18.3	290.0	18.3
	285.0	2.7	285.0	2.5	285.0	2.5
Ta 4d5/2	—	—	228.5	0.3	228.7	0.8
Li 1s	55.0	24.7	55.1	25.4	55.0	27.6

---

When comparing the M-site elements (Nb and/or Ta), differences in both binding energy and atomic percentage are observed for the mixed composition vs. both the single substituted samples. In the pure Nb and Ta samples, the M-site elements are present at approximately 0.8 atomic %, whereas the mixed composition shows a lower value of 0.5 atomic %. This may indicate a reduced amount of the garnet phase exposed at the surface in the mixed sample, likely due a thicker layer of lithium carbonate. Interestingly, the mixed M-site sample also exhibits slightly lower binding energies for both Nb and Ta compared to the single-substitution samples. This could indicate that Nb and Ta may reside in a more electron-rich environment within the mixed-phase garnet, potentially due to modified local bonding environments or variations in oxygen coordination. However, the observed binding energy shifts of 0.1 eV for Nb and 0.2 eV for Ta, are relatively small, suggesting that these changes are not significant, and perhaps just within the experimental error.

The O 1s core level peaks show a profile which is best fitted with two components (FWHM ca. 1.7 eV): the higher BE component, ca. 531.7 eV, is assigned to oxygen belonging to the carbonate anion, whereas the lower BE component, ca. 529 eV, is assigned to the garnets lattice oxygen.³⁴ The C 1s core level region is fitted with two components: the higher BE component, ca. 289.9 eV (FWHM ca. 1.2 eV), is assigned to carbon belonging to the carbonate anion, whereas the lower BE component, 285 eV (FWHM ca. 1.6 eV), is assigned to adventitious carbon,³⁵ and used as internal reference for the BE scale. For all the investigated garnets, the Li 1s core level peak is fitted with a single component with maximum at 55.0 eV (FWHM ca. 1.5 eV) and assigned to the presence of Li(I).^30,31 Overall, the presence of the garnet phase elements is confirmed, however, the measured atomic concentrations deviate from those expected for a bulk stoichiometric garnet phase, particularly showing an excess of lithium, carbon and oxygen. Since XPS is a highly surface-sensitive technique, probing in the range of about 30 nanometres, it is particularly susceptible to detecting surface composition. The presence of excess Li, C and O, suggests that the garnets' surfaces are coated in a layer of Li₂CO₃ which is likely due to surface reaction of the exposed Li with atmospheric H₂O and CO₂.

3.7 Solid-state NMR spectroscopy

Solid-state NMR spectroscopy of lithium in the 1100 °C base pellets provides a detailed insight into the local chemical environment of lithium ions within the garnet material. Fig. 10 presents the ⁷Li MAS NMR spectra of Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta), which display a single, well-defined resonance peak at ∼1 ppm. Cubic garnets are known to exhibit partial lithium disorder, where lithium ions are distributed across both tetrahedral and octahedral sites (Fig. 10b and c).^6,36 Due to the similar local environments of these sites and rapid lithium-ion hopping at room temperature, the ⁷Li MAS NMR spectra for it typically yield a single, averaged resonance.³⁷ A small shoulder at ∼0 ppm, that is more pronounced in the Nb-containing samples, can be assigned to the presence of minor amounts of Li₂CO₃ that is consistent with its formation on the surface of the material.

Fig. 10 (a) ⁷Li (14.1 T, 20 kHz MAS) MAS NMR spectra of powder Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta). (b) Tetrahedral and octahedral frameworks of oxygen (represented as red spheres at the polyhedral corners), which host Li atoms at specific sites: yellow (24 d, tetrahedral), blue (48 g, octahedral), and green (96 h, off-centre octahedral). (c) Illustrates the 3D network of connected oxygen polyhedron accommodating Li atoms (yellow and blue spheres).

3.8 AC impedance analysis

Typical impedance plots for the parent compound Li₅La₃Nb₂O₁₂ obtained at 55 and 100 °C in air, from samples prepared at two sintering temperatures, are shown in Fig. 11. Only pellets that were not in contact with the crucible during sintering were selected for ionic conductivity measurements, to eliminate the influence of undesirable reactions and garnet decomposition on the ionic conductivity. The 55 °C impedance plot for the Li₅La₃Nb₂O₁₂ sample sintered at 1100 °C features a large bulk (grain interior, R_b) resistance at the high-frequency region, alongside a grain boundary (R_gb) contribution and a tail at low frequencies (Fig. 11(c)). In contrast, the sample sintered at 900 °C exhibits primarily a single semicircle in the high-frequency region with a low-frequency tail. When ionically blocking electrodes are employed, the low-frequency tail on the AC impedance plots is characteristic of an ionically conductive material.²⁶ At temperatures (above ∼70 °C), the impedance response for all samples (Fig. 11, S9 and S10, see SI) becomes less distinguishable into separate contributions of bulk, and grain boundary contributions. Therefore, for comparative purposes, the total resistance (combining both bulk and grain boundary contributions) is used to represent the overall electrical conductivity across the full temperature range studied.

Fig. 11 Typical AC impedance plots obtained in the frequency range 5 Hz to 13 MHz using lithium ion blocking Au electrode for Li₅La₃Nb₂O₁₂ sintered at 900 °C (thickness: 0.15 cm; diameter: 1.3 cm), (a) 55 °C and (b) 100 °C and 1100 °C (thickness: 0.15 cm; diameter: 1.32 cm) (c) 55 °C and (d) 100 °C.

The resistance derived from the Nyquist plot at low-frequency intercept to the real axis, the thickness of the pellet, l, and the surface area of the electrode, A, were used to calculate the ionic conductivity, σ, of the investigated three garnet-type samples:

The activation energy for the lithium conductivity is calculated by fitting the data to the Arrhenius equation:

where A is the pre-exponential factor, k is the Boltzmann constant and T is the absolute temperature. The temperature-dependent pre-exponential factor of the garnet-type electrolyte takes the concentration, diffusion coefficient and entropy into consideration. Here, this was taken directly from the literature and was not calculated for these samples, as it has been assumed that the structures have the ideal stoichiometric concentration of lithium.

Lithium-ion conductivity of the parent garnets Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) can be obtained via Arrhenius plots. Fig. 12 presents the Arrhenius behaviour of Li₅La₃Nb₂O₁₂ sintered at both 900 °C and 1100 °C. The conductivity data, collected during both heating and cooling cycles, follow a linear trend (dashed lines in the plots), indicating that the measurements reflect equilibrium behaviour. The overall trends align well with those reported in literature for similar materials. The Li₅La₃Nb₂O₁₂ sample prepared in 2006 by Thangadurai et al. sintered at 1000 °C lies between the two sintering conditions used in this study, offering a useful point of comparison.³⁸ In general, the Nb- and Ta-only samples show higher conductivities when sintered at 900 °C compared to 1100 °C. Conversely, the mixed Nb/Ta sample displays improved conductivity when sintered at 1100 °C. These results highlight the intricate balance between phase purity and microstructure. While higher sintering temperatures promote densification, which should enhance ionic transport, in this study they also generate impurity phases, which can reside particularly at grain boundaries, increasing resistance and thus reducing overall conductivity. The samples sintered at 900 °C all exhibit comparable conductivities, as expected for samples with similar phase composition and fewer impurities. However, the 1100 °C samples show a broader range of conductivity values, indicating varying responses to high-temperature processing. The highest conductivity at 55 °C was observed for the Li₅La₃NbTaO₁₂ composition sintered at 1100 °C, with a value of 4.8 × 10⁻⁵ S cm⁻¹. Above 300 °C, the conductivities of all samples converge, this high-temperature effect is beyond the scope of the present study, and the analysis focuses on the Arrhenius behaviour between 25 °C and 200 °C.

Fig. 12 Arrhenius plots of the total conductivity of all Li₅La₃Nb₂O₁₂ (Nb-LLO); Li₅La₃Ta₂O₁₂ (Ta-LLO); and Li₅La₃NbTaO₁₂ (NbTa-LLO) sintered at 900 °C (a) and sintered at 1100 °C (b). For comparison, data from literature³⁸ sintered at 1000 °C is included in both (a) and (b).

From the Arrhenius plot, the activation energies for lithium-ion conduction were extracted (Table 2).^5,23,38–44 All samples sintered at 900 °C have activation energies of 0.50 eV or lower, reflecting favourable lithium-ion mobility. However, the Ta-substituted sample sintered at 1100 °C exhibits a significantly higher activation energy of 0.62 eV, highlighting a difference in the sample compared to the other samples, and suggesting that a secondary phase, such as surface contamination, may be present. A possible explanation for the reduced conductivity and increased activation energy could be surface contamination with Li₂CO₃, which can hinder lithium-ion transport and lead to poor overall conductivity. The conductivity study yielded results that are comparable to the literature. However, the observed differences in conductivity across the two sintering temperatures underscore the importance of optimising phase composition and structural features. To fully understand and control the conductivity behaviour in lithium garnet electrolytes, careful consideration of secondary phase (contamination) formation is required.

Table 2 Chemical composition, final synthesis temperature, Li ion conductivity and activation energy of Li₅-phase garnets from this study and the literature

Composition (final sintering temperature)	σLi^+(S cm^−1)	Ea (eV)	References
Li5La3Nb2O12 (900 °C)	2.9 × 10^−5 (55 °C; air)	0.43	This study
Li5La3Nb2O12 (1100 °C)	1.9 × 10^−5 (55 °C; air)	0.44	This study
Li5La3Nb2O12 (950 °C)	2.3 × 10^−5 (50 °C; air)	0.55	38
Li5La3Nb2O12 (950 °C)	8.0 × 10^−6 (50 °C; air)	0.43	5
Li5La3Nb2O12 (950 °C)	2.5 × 10^−7(25 °C; air)	0.51	39
Li5La3Nb2O12 (1100 °C)	5.1 × 10^−6 (22 °C; air)	0.60	40
Li5La3NbTaO12 (900 °C)	6.8 × 10^−6 (55 °C; air)	0.50	This study
Li5La3NbTaO12 (1100 °C)	2.6 × 10^−5 (55 °C; air)	0.45	This study
Li3.43La3Nb1.07Ta0.93O12−y (1000 °C)	7.1 × 10^−8 (25 °C; air)	0.59	23
Li5La3Ta2O12 (900 °C)	3.3 × 10^−5 (55 °C; air)	0.42	This study
Li5La3Ta2O12 (1100 °C)	8.6 × 10^−7 (55 °C; air)	0.62	This study
Li5La3Ta2O12 (850 °C)	5.0 × 10^−6 (40 °C; air)	0.55	41
Li5La3Ta2O12 (950 °C)	1.2 × 10^−6 (50 °C; air)	0.56	5
Li5La3Ta2O12 (1000 °C)	2.6 × 10^−6 (60 °C; air)	0.6	42
Li5La3Bi2O12 (775 °C)	4.0 × 10^−5 (22 °C; Ar)	0.54	43
Li5La3Sb2O12 (950 °C)	8.2 × 10^−6 (24 °C; Ar)	0.51	44

---

3.9 Lithium carbonate and alumina contamination

3.9.1 X-ray photoelectron spectroscopy. Initial evidence of lithium carbonate formation was observed through SEM and EDS imaging. Subsequent surface-sensitive XPS analysis confirmed an excess of carbon and oxygen on the surfaces of all samples, consistent with the presence of Li₂CO₃. However, XPS analysis of the Li 1s region could not effectively distinguish between lithium in the carbonate and lithium in the garnet lattice due to overlapping binding energies.⁴⁵ Additionally, if the carbonate layer thickness is greater than the photoelectron escape depth (ca. 3.43 nm at a kinetic energy of 1432 eV),⁴⁶ then only the Li from the carbonate layer will be measured. However, this value must be considered indicative only, as the uniformity of the layer is very likely inhomogeneous, which is why the signals from the metals can be detected. Therefore, to investigate the carbonate presence further, a Li₅La₃Nb₂O₁₂ pellet was analysed under different conditions: (i) after being stored in a desiccator for approximately one month, and (ii) following heat treatment at 800 °C for two hours, then cooled to room temperature in a desiccator before measurement. Additionally, the sample was measured after heat treatment at 300 °C within the XPS measurement chamber. The results (Fig. 13) highlight that a change to the garnet pellet surface occurs as the C 1s and O 1s binding energy regions.

Fig. 13 XPS spectra for (a) O 1s and (b) C 1s regions of the surface of Li₅La₃Nb₂O₁₂ pellet, at room temperature without prior heat treatment (red line), at room temperature after heating the sample to 800 °C in air for 2 hours pre-measurement (green line) and at 300 °C for after heating the sample to 800 °C in air for 2 hours pre-measurement (blue line).

In the O 1s spectra, the carbonate oxygen peak at ∼532 eV is significantly reduced after heat treatment, while the lattice oxygen peak associated with the garnet phase at ∼529 eV becomes more prominent. Similarly, in the C 1s region, the carbonate carbon peak at ∼290 eV and the atmospheric carbon peak at ∼285 eV both diminish after heating.⁴⁷ Initially, XPS measurements were restricted to room temperature due to gas evolution that occurred on heating, which disrupted the vacuum at elevated temperatures. However, partial removal of the carbonate layer via heating enabled higher-temperature measurements by minimising decomposition-related outgassing.

In the sample without additional heat treatment, the XPS spectrum was dominated by Li₂CO_3, effectively masking signals from key garnet elements such as La and Nb. After heating, the carbonate layer was substantially reduced, revealing signals from the underlying garnet phase (Fig. S11, see SI). Based on the surface sensitivity of XPS and comparison with literature, the thickness of the surface Li₂CO₃ layer is estimated to be approximately 30 nm–100 nm.^15,48,49

Although heating of the sample was effective in immediately reducing the carbonate layer, an order of tens of nanometres of carbonate remained, and the layer is expected to regrow over time due to continued surface reactions with atmospheric H₂O and CO₂. Since the carbonate layer was observed across all samples during XPS analysis on samples stored in the lab, it is likely that without removing this layer, the ionic conductivity was impacted due to the lithium-ion transport being impeded by the carbonate layer at the pellet surfaces. Fig. S12 (see SI) presents fractured cross-sectional SEM images of an Ar-treated pellet, stored in the dry room for 45 days, and an untreated pellet, stored for more than four months. The images reveal the extent of surface contamination, with a contamination layer of 50–120 nm in the former case and a thicker contamination layer of 500–800 nm in the latter (Fig. S12, see SI).

These findings emphasise the need for future focus on both the complete removal and prevention of surface carbonate formation. It was found that storing samples at room temperature in a desiccator in a dry room alone was insufficient. This area is growing interests, with strategies such as protective coatings, optimised atmospheres during synthesis, or post-synthesis carbonate layer control employed to mitigate excessive Li₂CO₃ formation and preserve the electrochemical performance of garnet-type solid electrolytes.^47,50–52

3.9.2 Thermogravimetric analysis. TGA was employed to investigate the impact of heat treatment on the surface chemistry of the garnet pellets. In ambient atmosphere garnet samples undergo protonic exchange with water, as described in eqn (3).

Li₅La₃M₂O₁₂ + H₂O → “Li_5−xH_xLa₃M₂O₁₂·xLiOH” (3)

The protonic exchanged garnet subsequently reacts with carbon dioxide and forms a surface layer of lithium carbonate, as proposed in eqn (4).

A TGA experiment was designed to replicate the thermal treatment applied to the XPS samples, by heating the previously synthesised (sintered at 1100 °C) Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) pellets to 800 °C for one hour, followed by cooling to 100 °C, with the cycle repeated once. All compositions (Fig. S13, see SI) exhibit three distinct weight loss events occurring at approximately 100 °C, 250 °C, and 550 °C, indicating a similar interaction with atmospheric species across all samples. These features are consistent with the behaviour reported in literature for garnet-type materials:^53,54

• The first weight loss (∼100 °C) corresponds to the desorption of surface-adsorbed water.

• The second (∼250 °C) is attributed to the release of structural H₂O arising from proton incorporation into the garnet lattice.

• The third event (∼550 °C) reflects the decomposition of surface Li₂CO₃, with the release of CO₂.

The ∼250 °C water loss is attributed to H⁺ ions substituting for Li⁺ within the garnet structure, supporting the concept of a Li⁺/H⁺ exchange mechanism.⁵³ This mechanism, when coupled with exposure to ambient CO₂, results in the preformed LiOH layer reacting to form Li₂CO₃ on the surface (eqn (4)). Upon heating, the decomposition of the surface carbonate species is believed to occur via loss of carbon dioxide (eqn (5)), and the back-insertion of the lithium from lithium carbonate into the garnet returns to the original “stoichiometry”.

Importantly, the garnet reactions in ambient conditions are reversible, and at the end of the cycle, the original garnet is recovered.⁵⁵ Interestingly, during both heating cycles, a slight mass increase is observed between 100 and ∼250 °C. This may reflect adsorption or formation of LiOH and Li₂CO₃ on the surface during heating from residual atmospheric amounts of water and carbon dioxide, which is subsequently lost at higher temperatures. This observation may suggest that carbonate formation is initiated during the cooling phase of synthesis,⁵⁶ particularly as the sample passes through the 100–250 °C range, where the reaction is thermodynamically favourable (Fig. 14). Additionally, the presence of unreacted lithium, intentionally added during synthesis to compensate for lithium volatilisation at high sintering temperatures, may contribute to the formation of surface Li₂CO₃. As the amount of excess lithium was not precisely optimised, residual reactive Li₂O within the structure may facilitate Li₂CO₃ formation upon exposure to ambient air. TGA analysis shows that heating the samples to 800 °C for one-hour results in a two-step decomposition of surface Li₂CO₃. This thermal behaviour aligns with the changes observed in the XPS spectra, confirming the progressive removal of carbonate species from the surface. Understanding the two-step Li₂CO₃ formation and decomposition process under ambient conditions is essential for guiding future efforts to optimise synthesis protocols, reduce surface contamination, and explore alternative strategies that mitigate carbonate formation more effectively than the current approach.

Fig. 14 Schematic representation of the two potential reaction pathways which describe the formation of Li₂CO₃ outer layer which forms on the garnet surface under ambient conditions. Top: direct reaction of both H₂O and CO₂ in one step, and bottom: two step exchange reaction, first with H₂O, followed by reaction with CO₂.

3.9.3 Reaction with the alumina crucible. Ex situ PXRD analysis was performed on the pellet which was in contact with alumina crucible to assess potential intrinsic Al-doping, which may have arisen from an interaction with crucibles during high-temperature sintering. Efforts were focused on conducting a more extensive study on the Nb-based garnet. The PXRD patterns of Li₅La₃Nb₂O₁₂ samples in Fig. 15 were synthesised under various conditions: with and without a sacrificial mother powder at 1100 °C for 24 hours, at 900 °C for 24 hours without a mother powder, and at 1100 °C for 40 hours without a mother powder. The ICSD reference pattern for Li₅La₃Nb₂O₁₂ (ICSD #18586) is included for comparison. Additionally, LaAlO₃ (ICSD #92554), marked with red asterisks, appears in the 1100 °C samples without mother powder, indicating the formation of this phase due to interfacial reaction with the alumina crucible.

Fig. 15 Ex situ PXRD patterns of Li₅La₃Nb₂O₁₂ pellet in contact with alumina crucible during sintering at (a) 1100 °C for 24 hours with sacrificial mother powder, (b) 900 °C for 24 hours without mother powder, (c) 1100 °C for 24 hours without mother powder, and (d) 1100 °C for 40 hours without mother powder. For comparison, ICSD Li₅La₃Nb₂O₁₂ (#18586) is given, and *denotes the LaAlO₃ (#92554) phase.

From the ex situ PXRD patterns of Li₅La₃Nb₂O₁₂, it is evident that the sample sintered at 1100 °C with a sacrificial mother powder and the sample sintered at 900 °C without a mother powder both yield a clean, single-phase garnet structure. In contrast, the sample sintered at 1100 °C without a mother powder shows the garnet phase along with multiple impurity phases, one of which is identified as LaAlO₃. Notably, when the sintering time at 1100 °C is extended to 40 hours without a mother powder, the Al-based secondary phase becomes the dominant phase, indicating extensive reaction between the garnet and the alumina crucible. The PXRD analysis of Li₅La₃Ta₂O₁₂ and Li₅La₃NbTaO₁₂ (Fig. S14 and S15, see SI) sintered without mother powder at 900 °C and 1100 °C, also shows the presence of a major secondary phase in the base pellet sintered at 1100 °C in the alumina crucible.

Al has been widely employed to stabilise cubic phase garnets, both doping intrinsically via alumina crucibles,^57–59 or doping extrinsically via the addition of alumina,^60–62 where Al substitutes into the structure in place of Li. This substitution creates charge-compensating lithium vacancies, which are essential for stabilising the high-conductivity cubic phase.⁵⁸ Al₂O₃ has been employed directly as both a sintering aid and phase stabiliser in the synthesis of garnet-type materials such as Li₇La₃Zr₂O₁₂ (LLZO).^62,63 Al₂O₃ promotes densification by forming a secondary phase at grain boundaries, which has been shown to suppress lithium dendrite growth and enhance overall ionic conductivity.⁶⁴ The process of intrinsic doping happens at high sintering temperatures, when reactive species, such as Li₂O or La₂O₃ can interact with the crucible, producing transient melt phases capable of dissolving Al₂O₃.⁵⁸ This process facilitates the diffusion of Al into the garnet, leading to the formation of secondary Al-containing phases.⁶⁵ In addition, it has been shown that one significant reason for reaction of garnet with the crucible during high temperature sintering is the Li volatilization inducing Al-doping to ensure charge balance.⁶⁶

A previous study on LLZO garnets reported that reactions between the garnet and alumina crucibles begin around at ∼1050 °C.⁶⁷ This aligns with our observations, where Al-containing phase LaAlO₃ emerges as a major phase in samples sintered without mother powder at 1100 °C but is only minor in the sample sintered at 900 °C. These results highlight the sensitivity of garnet materials to sintering conditions, particularly the role of temperature in driving Al contamination from the crucible. Careful control over sintering environments is therefore essential to ensure phase purity and material stoichiometry. To minimise Al contamination, two effective strategies are demonstrated in this study: (i) lowering the sintering temperature below 1100 °C, which reduces alumina crucible interaction, though it may compromise pellet densification, and (ii) utilising a sacrificial mother powder, which acts as a protective buffer layer, enabling clean garnet formation even at elevated temperatures. Overall, Al-related secondary phases were only observed in the base pellet in direct contact with the crucible, therefore strongly indicating that Al diffusion and contamination are surface-limited and driven by physical contact during sintering. Using single-crystal sapphire tubes for synthesis during in situ PXRD measurements allowed to avoid the Al contamination of the material due to the considerably lower reactivity of the alumina single crystal surface.

The BSE images of all the analogues sintered at 900 °C and 1100 °C without mother powder for 24 h are presented in Fig. 16. As also observed before from the PXRD analysis, the pellets in contact with the crucible during sintering at 900 °C show no evidence of secondary phases for all garnets. However, sintering at 1100 °C led to the formation of Al-rich phases, most severely in the case of Nb-doped garnets. Line and point EDS analysis of Li₅La₃Nb₂O₁₂ sintered at 1100 °C (Fig. S16 and S17, see SI) reveals the most significant reaction with the alumina crucible. Three distinct phases are observed: the garnet with the targeted stoichiometry, an Al- and La-rich phase, and a Nb-rich phase. This highlights the severe destabilisation of the garnet due to Al contamination from the crucible, forming a LaAlO₃ phase, as confirmed by XRD analysis.

Fig. 16 Backscattered electron detector (BSED) SEM images of the base pellets in contact with alumina crucible sintered at 900 °C (a) Li₅La₃Nb₂O₁₂, (b) Li₅La₃NbTaO₁₂, and (c) Li₅La₃Ta₂O₁₂ and 1100 °C (d) Li₅La₃Nb₂O₁₂, (e) Li₅La₃NbTaO₁₂, and (f) Li₅La₃Ta₂O₁₂. Increased density was observed at 1100 °C.

Fig. S18 and S19 (see SI), show similar analyses for Li₅La₃NbTaO₁₂ and Li₅La₃Ta₂O₁₂. For Li₅La₃NbTaO₁₂, three phases are present: the intended garnet, a Ta- and Nb-rich phase, and an Al-rich phase containing minor La. This indicates notable, though less severe, interaction with the crucible compared to the Nb-based sample. Li₅La₃Ta₂O₁₂ exhibits the least contamination, with only two phases detected: the garnet and an Al- and La-rich secondary phase. These findings demonstrate that Nb-based garnets are more prone to crucible interactions at temperatures above 900 °C, owing to the greater tendency of niobium to undergo reduction at high temperatures.

Combining the findings from PXRD, SEM, and EDS, it is evident that sintering at 1100 °C without sacrificial mother powder leads to the formation of secondary La–Al phases. This effect is intensified with prolonged sintering durations, as the 40-hours sample predominantly shows La–Al phase formation. As established earlier, achieving a single-phase garnet at 1100 °C requires the use of sacrificial mother powder. Beyond phase control, SEM analysis confirms that this approach also promotes better densification and reduces surface voids compared to samples prepared without a mother powder.

Future approaches to achieve highly dense solid electrolytes at elevated temperatures are the use of ZrO₂, MgO or Pt crucibles, as they provide higher control on elemental composition of produced garnet,⁶⁸ especially the latter two show lower reactivity with the Li₂O gas.⁶⁹ For example, it was shown by Jiang et al.⁷⁰ that a smaller amount of Mg was incorporated in the Ga-doped LLZO structure at the same conditions of heat treatment compared to Al. Alternatively, lower sintering temperatures or lower needed dwelling time at higher temperatures may be employed through advanced techniques such as ultra-fast high-temperature sintering,⁷¹ flash sintering,⁷² cold sintering⁷³ or spark plasma sintering^74,75 which can enable densification while minimizing contamination.^68,76

3.9.4 Solid-state NMR spectroscopy. The ²⁷Al MAS NMR spectra of Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) in Fig. 17 offer an insight into the Al environments within the base pellets sintered at 1100 °C. Aluminium incorporation is observed across the samples, with all three compositions exhibiting a prominent resonance at ∼11 ppm and a weaker resonance at ∼77 ppm.

Fig. 17 Quantitative ²⁷Al (14.1 T, 20 kHz MAS) MAS NMR spectra of powder Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta). The NMR spectra are shown normalised by the sample mass. The * symbol indicates spinning sidebands.

The high intensity resonance at ∼11 ppm, characteristic of octahedrally coordinated Al, can be assigned to LaAlO₃,⁴⁰ which is consistent with previous PXRD, SEM, and EDS results. Alongside this dominant peak, all samples exhibit a lower intensity resonance at ∼77 ppm, falling within the expected range for tetrahedrally coordinated aluminates (50–90 ppm).⁷⁷ A study by Vema et al.⁶² identified a similar resonance at ∼79 ppm when investigating Al-doped LLZO, attributing it to γ-LiAlO₂, therefore, the ∼77 ppm peak observed here could likely correspond to the same phase. Additionally, the Ta-only sample shows a low-intensity resonance at ∼67 ppm. Vema et al. also reported that a resonance at ∼68 ppm arises from Al substitution into the Li 24 d site in the garnet structure, this suggests that the resonance slightly above the noise in the Ta sample may reflect minor Al doping into the garnet lattice.⁶² Similar observations were reported by Karasulu et al.,¹⁰ whose DFT calculations demonstrated that Al preferentially occupies the Li1 site, giving rise to an NMR shift of ∼68.6 ppm and the observed NMR peak broadening originates from local structural distortions of the AlO₄ tetrahedra, where the absence of corner-sharing LiO₄ (Li2) neighbours increases the distortion index and thereby increasing the quadrupolar coupling constant.

The ²⁷Al MAS NMR spectra were normalised by the sample mass in the rotor, allowing qualitative conclusions about the composition with the most Al-secondary phase to be drawn. The peak at ∼11 ppm, associated with LaAlO₃, shows the greatest intensity in the Nb-only sample, followed by the mixed Nb/Ta, and the lowest in the Ta-only sample. This trend suggests that the Nb-only sample contains the highest concentration of LaAlO₃ secondary phase, which aligns with the SEM and EDS analysis of the base pellets. Interestingly, the trend does not hold for the low-intensity peaks. The mixed Nb/Ta sample exhibits the most intense resonance at ∼77 ppm, suggesting it contains the most γ-LiAlO₂. The solid-state NMR spectroscopy study indicates that aluminium incorporation from the alumina crucible is present across all compositions, predominantly in an octahedral LaAlO₃ environment, with the Nb-only sample containing the most secondary phase in the samples studied. This implies that Nb promotes greater reactivity with alumina crucibles. However, as the Li stoichiometry was not precisely determined and conclusions regarding Al incorporation mechanisms remain incomplete. As a study by Lan et al. indicates that Li content strongly influences Al uptake,⁶⁷ therefore, further analysis of the sample stoichiometry is required to determine whether M-site substitution is driving Al contamination.

4. Conclusions

The parent garnet-type Li₅La₃M₂O₁₂ (M = Nb, Ta, Nb/Ta) was successfully synthesised via solid-state reaction at 900 °C and 1100 °C. Powder XRD and Raman spectroscopy confirmed single-phase garnets with Ia3̄d symmetry. SEM imaging showed better densification of the samples sintered at 1100 °C, than those sintered at 900 °C. The in situ PXRD study of the Li₅La₃Nb₂O₁₂ garnet gave an insight into the route of the phase formation. Under the conditions studied, the Nb garnet phase proceeds via La₂LiNbO₆, Li₃NbO₄ and La₃NbO₇, and the product begins to form below 700 °C. The investigation clearly highlighted that at elevated temperatures, above 1075 °C, decomposition of the garnet phase begins, and on the in situ scale, the product was almost entirely decomposed. This study identifies that such elevated temperatures to achieve the garnet structure is not the most efficient and that a lower temperature for a longer dwelling time may be sufficient.

It was observed that samples sintered at 900 °C showed similar conductivities to each other, while those sintered at 1100 °C displayed more variation between the garnet compositions. The highest conductivity measured in this study was 4.8 × 10⁻⁵ S cm⁻¹ at 55 °C, for Li₅La₃NbTaO₁₂ sintered at 1100 °C. This is comparable to Li₅ phases previously reported in the literature.^{35 7}Li MAS NMR spectra showed a single resonance for all samples, indicating that all Li environments are equivalent due to fast Li-ion hopping, rendering the Li environments indistinguishable. ²⁷Al MAS NMR spectra revealed Al incorporation from alumina crucibles, in all samples sintered directly on the crucible base at 1100 °C. The main contaminant phase appears to be LaAlO₃, which was confirmed by PXRD, SEM, and EDS analysis. This work showed that to reduce Al contamination, sintering with a sacrificial mother powder or at 900 °C is recommended. This study provides a comprehensive analysis of the Li₅- garnet phase, a composition often overlooked in favour of the more commonly studied Li₇La₃Zr₂O₁₂ (LLZO). By revisiting the parent garnet structure, this work offers valuable insights that could inform the design and optimisation of next-generation garnet-based solid-state lithium-ion electrolytes.

Author contributions

Heather A. Ritchie: conceptualisation, synthesis, formal analysis, investigation, methodology, validation, visualisation, writing – original draft; Ioanna M. Pateli: Raman and SEM data curation, formal analysis, investigation, methodology, writing, review & editing; Oxana V. Magdysyuk: PXRD data curation, formal analysis, investigation, methodology, writing, review & editing. Aaron B. Naden: TEM data curation, formal analysis, investigation, methodology, review & editing. Heitor S. Seleghini: data curation, review & editing; Federico Grillo: XPS data curation, formal analysis, investigation, methodology, review & editing. Gavin Peters: data curation; Sharon E. Ashbrook: collaboration, review & editing. John T. S. Irvine: collaboration and review; Venkataraman Thangadurai: conceptualisation, supervision, methodology, writing, review & editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

The research data underpinning this publication, including X-ray diffraction, scanning electron microscopy, transmission electron microscopy, Raman spectroscopy, XPS spectroscopy, NMR spectroscopy, TGA analysis and EIS data are available at the University of St Andrews repository and can be accessed at https://doi.org/10.17630/ddf38307-f6c1-4e66-ac13-f42fbd8c775e.

Supplementary information: X-ray diffraction, scanning electron microscopy and elemental analysis, XPS spectroscopy, TGA analysis and electrochemical impedance data. See DOI: https://doi.org/10.1039/d5ta06905j.

Acknowledgements

We also acknowledge the support from EPSRC Core Equipment Grant (EP/V034138/1) for powder diffractometers and the support from the EPSRC Light Element Analysis Facility Grant EP/T 0192981 and the EPSRC Strategic Equipment Resource Grant EP/R 023751/1 for the Raman spectrometer and Scanning electron microscope. We gratefully acknowledge support from the EPSRC through grant numbers EP/L017008/1, EP/R023751/1 and EP/T019298/1 for scanning transmission electron microscope data. H. S. S. thanks the Allan Handsel Postgraduate Research Scholarship. V. T. also thanks University of St Andrews for Chair in Energy.

References

1. J. B. Goodenough, Energy Environ. Sci., 2014, 7, 14–18.
2. V. Thangadurai, ACS Energy Lett., 2025, 10, 5067–5074.
3. J. Janek and W. G. Zeier, Nat. Energy, 2016, 1, 1–4.
4. V. Thangadurai, S. Narayanan and D. Pinzaru, Chem. Soc. Rev., 2014, 43, 4714–4727.
5. V. Thangadurai, H. Kaack and W. J. F. Weppner, J. Am. Ceram. Soc., 2003, 86, 437–440.
6. E. J. Cussen, Chem. Commun., 2006, 412–413.
7. B. Dong, M. P. Stockham, P. A. Chater and P. R. Slater, Dalton Trans., 2020, 49, 11727–11735.
8. V. Thangadurai and W. Weppner, Adv. Funct. Mater., 2005, 15, 107–112.
9. R. Murugan, V. Thangadurai and W. Weppner, Angew. Chem., Int. Ed., 2007, 46, 7778–7781.
10. B. Karasulu, S. P. Emge, M. F. Groh, C. P. Grey and A. J. Morris, J. Am. Chem. Soc., 2020, 142, 3132–3148.
11. J. Awaka, N. Kijima, H. Hayakawa and J. Akimoto, J. Solid State Chem., 2009, 182, 2046–2052.
12. Y. Mao, J. Liu, W. Chen, W. Zhang and C. Sun, Mater. Horitz., 2025, 12, 6082–6123.
13. A. J. Samson, K. Hofstetter, S. Bag and V. Thangadurai, Energy Environ. Sci., 2019, 12, 2957–2975.
14. K. Ma, B. Chen, C. X. Li and V. Thangadurai, J. Mater. Chem. A, 2023, 12, 3601–3615.
15. H. Huo, J. Luo, V. Thangadurai, X. Guo, C. W. Nan and X. Sun, ACS Energy Lett., 2020, 5, 252–262.
16. L. Xu, J. Li, W. Deng, H. Shuai, S. Li, Z. Xu, J. Li, H. Hou, H. Peng, G. Zou and X. Ji, Adv. Energy Mater., 2021, 11, 1–24.
17. A. A. Coelho, J. Appl. Crystallogr., 2018, 51, 210–218.
18. H. Rietveld, J. Appl. Crystallogr., 1969, 2, 65–71.
19. H. Rietveld, Acta Crystallogr., 1967, 22, 151–152.
20. N. Fairley, V. Fernandez, M. Richard-Plouet, C. Guillot-Deudon, J. Walton, E. Smith, D. Flahaut, M. Greiner, M. Biesinger, S. Tougaard, D. Morgan and J. Baltrusaitis, Appl. Surf. Sci. Adv., 2021, 5, 100112.
21. Y. Seto and M. Ohtsuka, J. Appl. Crystallogr., 2022, 55, 397–410.
22. R. D. Shannon, Acta Crystallogr., 1976, 32, 751–761.
23. A. P. Kafle, W. Wong-Ng, V. Oleshko, G. R. Stafford, J. A. Kaduk, A. Eufrasio, I. L. Pegg and B. Dutta, Powder Diffr., 2024, 39, 191–205.
24. K. Hu, H. Tang, B. Zheng, L. Yu, F. Xiong, H. Li, L. Qiu, F. Wan, Y. Song, B. Zhong, Z. Wu and X. Guo, Angew. Chem., Int. Ed., 2025, 64, e202413563.
25. F. Tietz, T. Wegener, M. T. Gerhards, M. Giarola and G. Mariotto, Solid State Ionics, 2013, 230, 77–82.
26. J. T. S. Irvine, D. C. Sinclair and A. R. West, Adv. Mater., 1990, 2, 132–138.
27. Y. Huang, H. Yang, Y. Gao, G. Chen, Y. Li, L. Shi and D. Zhang, Mater. Chem. Front., 2024, 8, 1282–1299.
28. A. Sharafi, E. Kazyak, A. L. Davis, S. Yu, T. Thompson, D. J. Siegel, N. P. Dasgupta and J. Sakamoto, Chem. Mater., 2017, 29, 7961–7968.
29. H. Buschmann, J. Dölle, S. Berendts, A. Kuhn, P. Bottke, M. Wilkening, P. Heitjans, A. Senyshyn, H. Ehrenberg, A. Lotnyk, V. Duppel, L. Kienle and J. Janek, Phys. Chem. Chem. Phys., 2011, 13, 19378–19392.
30. C. D. Wagner, W. M. Kiggs, L. E. Davis and J. F. Moulder, Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer Corporation-Physical Electronics Division, Eden Prairie, MN, 1979.
31. NIST X-ray Photoelectron Spectroscopy Database (SRD 20), https://srdata.nist.gov/xps/%0A.
32. M. F. Sunding, K. Hadidi, S. Diplas, O. M. Løvvik, T. E. Norby and A. E. Gunnæs, J. Electron Spectrosc. Relat. Phenom., 2011, 184, 399–409.
33. A. Müller, F. Okur, A. Aribia, N. Osenciat, C. A. F. Vaz, V. Siller, M. El Kazzi, E. Gilshtein, M. H. Futscher, K. V. Kravchyk, M. V. Kovalenko and Y. E. Romanyuk, Mater. Adv., 2023, 4, 2138–2146.
34. H. Yamada, T. Ito, S. P. Kammampata and V. Thangadurai, ACS Appl. Mater. Interfaces, 2020, 12, 36119–36127.
35. T. Pang, S. Wu, H. Wu, X. Li, Y. Li, L. T. Yu, H. Zhang, Y. Han, Z. Guo and N. Zhang, Adv. Sci., 2025, 12, e04388.
36. E. J. Cussen, J. Mater. Chem., 2010, 20, 5167–5173.
37. L. Van Wüllen, T. Echelmeyer, H. W. Meyer and D. Wilmer, Phys. Chem. Chem. Phys., 2007, 9, 3298–3303.
38. V. Thangadurai and W. Weppner, J. Solid State Chem., 2006, 179, 974–984.
39. L. Truong and V. Thangadurai, Chem. Mater., 2011, 23, 3970–3977.
40. S. Narayanan and V. Thangadurai, J. Power Sources, 2011, 196, 8085–8090.
41. C. R. Mariappan, K. I. Gnanasekar, V. Jayaraman and T. Gnanasekaran, J. Electroceram., 2013, 30, 258–265.
42. H. El-Shinawi and S. M. El-Dafrawy, J. Solid State Chem., 2025, 343, 125154.
43. R. Murugan, W. Weppner, P. Schmid-Beurmann and V. Thangadurai, Mater. Sci. Eng., B, 2007, 143, 14–20.
44. R. Murugan, W. Weppner, P. Schmid-Beurmann and V. Thangadurai, Mater. Res. Bull., 2008, 43, 2579–2591.
45. M. Fingerle, C. Loho, T. Ferber, H. Hahn and R. Hausbrand, J. Power Sources, 2017, 366, 72–79.
46. S. Tanuma, C. J. Powell and D. R. Penn, Surf. Interface Anal., 1994, 21, 165–176.
47. P. Srivastava, B. Bazri, D. K. Maurya, Y. T. Hung, D. H. Wei and R. S. Liu, ACS Energy Lett., 2025, 10, 1725–1732.
48. L. Cheng, C. H. Wu, A. Jarry, W. Chen, Y. Ye, J. Zhu, R. Kostecki, K. Persson, J. Guo, M. Salmeron, G. Chen and M. Doeff, ACS Appl. Mater. Interfaces, 2015, 7, 17649–17655.
49. L. Cheng, E. J. Crumlin, W. Chen, R. Qiao, H. Hou, S. Franz Lux, V. Zorba, R. Russo, R. Kostecki, Z. Liu, K. Persson, W. Yang, J. Cabana, T. Richardson, G. Chen and M. Doeff, Phys. Chem. Chem. Phys., 2014, 16, 18294–18300.
50. N. Wang, M. Jia, Z. Bi and X. Guo, Adv. Funct. Mater., 2024, 34, 2401400.
51. X. Yi, Y. Guo, S. Chi, S. Pan, C. Geng, M. Li, Z. Li, W. Lv, S. Wu and Q. H. Yang, Adv. Funct. Mater., 2023, 33, 2303574.
52. Y. N. Yang, Y. X. Li, Y. Q. Li and T. Zhang, Nat. Commun., 2020, 11, 5519.
53. A. Sharafi, S. Yu, M. Naguib, M. Lee, C. Ma, H. M. Meyer, J. Nanda, M. Chi, D. J. Siegel and J. Sakamoto, J. Mater. Chem. A, 2017, 5, 13475–13487.
54. H. Huo, J. Luo, V. Thangadurai, X. Guo, C. W. Nan and X. Sun, ACS Energy Lett., 2020, 5, 252–262.
55. G. Larraz, A. Orera and M. L. Sanjuán, J. Mater. Chem. A, 2013, 1, 11419–11428.
56. R. Wagner, D. Rettenwander, G. J. Redhammer, G. Tippelt, G. Sabathi, M. E. Musso, B. Stanje, M. Wilkening, E. Suard and G. Amthauer, Inorg. Chem., 2016, 55, 12211–12219.
57. Y. Shimonishi, A. Toda, T. Zhang, A. Hirano, N. Imanishi, O. Yamamoto and Y. Takeda, Solid State Ionics, 2011, 183, 48–53.
58. C. A. Geiger, E. Alekseev, B. Lazic, M. Fisch, T. Armbruster, R. Langner, M. Fechtelkord, N. Kim, T. Pettke and W. Weppner, Inorg. Chem., 2011, 50, 1089–1097.
59. M. A. Limpert, T. B. Atwater, T. Hamann, G. L. Godbey, G. T. Hitz, D. W. McOwen and E. D. Wachsman, Chem. Mater., 2022, 34, 9468–9478.
60. S. Kumazaki, Y. Iriyama, K. H. Kim, R. Murugan, K. Tanabe, K. Yamamoto, T. Hirayama and Z. Ogumi, Electrochem. Commun., 2011, 13, 509–512.
61. J. Sastre, T. Y. Lin, A. N. Filippin, A. Priebe, E. Avancini, J. Michler, A. N. Tiwari, Y. E. Romanyuk and S. Buecheler, ACS Appl. Energy Mater., 2019, 2, 8511–8524.
62. S. Vema, A. H. Berge, S. Nagendran and C. P. Grey, Chem. Mater., 2023, 35, 9632–9646.
63. Y. Suzuki, K. Kami, K. Watanabe, A. Watanabe, N. Saito, T. Ohnishi, K. Takada, R. Sudo and N. Imanishi, Solid State Ionics, 2015, 278, 172–176.
64. X. Hu, S. Chen, Z. Wang, Y. Chen, B. Yuan, Y. Zhang, W. Liu and Y. Yu, Nano Lett., 2023, 23, 887–894.
65. E. Rangasamy, J. Wolfenstine and J. Sakamoto, Solid State Ionics, 2012, 206, 28–32.
66. B. Dong, L. L. Driscoll, M. P. Stockham, E. Kendrick and P. R. Slater, Solid State Ionics, 2020, 350, 115317.
67. W. Lan, D. Lu, R. Zhao and H. Chen, Int. J. Electrochem. Sci., 2019, 14, 9695–9703.
68. C. Chen, K. Wang, H. He, E. Hanc, M. Kotobuki and L. Lu, Small, 2023, 19, 1–33.
69. X. Huang, Y. Lu, H. Guo, Z. Song, T. Xiu, M. E. Badding and Z. Wen, ACS Appl. Energy Mater., 2018, 1, 5355–5365.
70. Y. Jiang, Y. Zhou, Z. Hu, Y. Huang and X. Zhu, Ceram. Int., 2020, 46, 3367–3373.
71. A. Kern and P. J. McGinn, J. Eur. Ceram. Soc., 2022, 42, 7501–7507.
72. Y. Yang, Z. Zhang, T. Ma, S. Jia and C. Huang, Ionics, 2025, 31, 1341–1350.
73. J. H. Seo, H. Nakaya, Y. Takeuchi, Z. Fan, H. Hikosaka, R. Rajagopalan, E. D. Gomez, M. Iwasaki and C. A. Randall, J. Eur. Ceram. Soc., 2020, 40, 6241–6248.
74. M. Kotobuki and M. Koishi, J. Alloys Compd., 2020, 826, 154213.
75. S. Palakkathodi Kammampata, H. Yamada, T. Ito, R. Paul and V. Thangadurai, J. Mater. Chem. A, 2020, 8, 2581–2590.
76. A. Sazvar, M. Hajibandeh, P. Vafaei, E. Hosseinzadeh and M. Jabbari, J. Energy Storage, 2024, 101, 113863.
77. M. Haouas, F. Taulelle and C. Martineau, Prog. Nucl. Magn. Reson. Spectrosc., 2016, 94–95, 11–36.

---