Stefan
Smetaczek
a,
Maximilian
Bonta
a,
Andreas
Wachter-Welzl
a,
Stefanie
Taibl
a,
Reinhard
Wagner
b,
Daniel
Rettenwander
c,
Jürgen
Fleig
a and
Andreas
Limbeck
*a
aInstitute of Chemical Technologies and Analytics, TU Wien, Vienna, Austria. E-mail: Andreas.Limbeck@tuwien.ac.at
bDepartment of Chemistry and Physics of Materials, University Salzburg, Salzburg, Austria
cInstitute for Chemistry and Technology of Materials, TU Graz, Graz, Austria
First published on 26th March 2020
Cubic Li7La3Zr2O12 (LLZO) garnets are among the most promising solid-state electrolytes for next-generation Li batteries. However, despite intensive research in recent years, little effort is spent on proper chemical analysis of the material. For reliable LLZO investigation, knowledge about the exact chemical composition, especially the Li content, is of uttermost importance. Herein, we present a method for the laterally resolved stoichiometry determination of Al stabilized LLZO (Li7−3xAlxLa3Zr2O12) using LA-ICP-OES. To ensure reliable signal quantification, matrix-matched standards are prepared and carefully characterized using sample digestion and liquid ICP-OES measurement. An internal standard-independent calibration strategy based on 100 m% normalization is applied, enabling the quantification of all cations within the material. By comparing the obtained high-precision LA-ICP-OES calibrations with analog LA-ICP-MS measurements, it is shown that ICP-OES is the superior choice for the analysis of Li. The developed method is applied to record quantitative distribution images of an Al stabilized LLZO pellet, revealing macroscopic stoichiometry variations within the sample. To verify the analysis, the average stoichiometry obtained by the LA measurements is compared with the bulk composition determined via liquid ICP-OES analysis after sample fusion. The obtained values show excellent agreement, confirming the accuracy of the developed method.
A variety of inorganic solid Li-ion conductors have been considered for use in all-solid state batteries.3–6 While some of the proposed solid electrolytes exhibit Li-ion conductivity comparable to organic liquid electrolytes, the performance of batteries based on these inorganic ion conductors is still inferior to that of commercially available Li-ion batteries.6 Several key challenges remain, such as poor electrolyte/electrode interface, instability in ambient environment, flexibility concerns, and poor cycling stability.6–9 One of the most promising solid electrolytes is the Li stuffed garnet Li7La3Zr2O12 (LLZO), which was first reported by Murugan et al. in 2007.10 Beside its high Li-ion conductivity (up to 10−3 S cm−1 at ambient temperature) and high electrochemical stability window, it also features chemical stability against elemental Li, which enables the use in Li-metal batteries.10–13
Pure LLZO crystalizes in two different polymorphs: a tetragonal and a cubic modification. For use as a Li-ion conductor, the cubic structure is much more desirable since its Li-ion conductivity is two orders of magnitude higher compared to the tetragonal variant.14,15 The highly conductive cubic modification is not stable at room temperature, however, stabilization is achieved by partial Al-substitution (3 Li+ → Al3+), resulting in the formula unit Li7−3xAlxLa3Zr2O12.16–18
Although a lot of research has been focused on LLZO in recent years, the reproducible synthesis of garnets with high Li-ion conductivity remains challenging. For example, pellets with the same nominal composition prepared at different laboratories show differences in ionic conductivity of more than an order of magnitude.11,19 Furthermore, local conductivity variations have been reported within one and the same sample.20 The reasons for these variations are still not completely understood, however, they probably relate to differences in sample stoichiometry.19 Since synthesis of LLZO normally involves high-temperature treatment, partial loss of Li2O due to evaporation cannot be avoided.21,22 In addition, Al incorporation from the commonly used Al2O3 crucibles further changes the sample composition.16,17,21 Thus, determination of the chemical composition after the synthesis is crucial for reliable LLZO investigation.
Chemical analysis of Al-stabilized LLZO pellets is mostly performed using sample-digestion and subsequent ICP-OES (Inductively Coupled Plasma-Optical Emission Spectroscopy) or ICP-MS (Inductively Coupled Plasma-Mass Spectrometry) analysis.14,17,22–24 While this approach enables reliable determination of the bulk composition, it requires time consuming sample preparation and does not provide any spatially resolved information. For elemental surface analysis of LLZO, mostly EDX (Energy-Dispersive X-ray Spectroscopy) is used.14,16,21,25,26 This method enables the determination of the Al, La, and Zr distribution, but its use is strongly limited by its incapacity to detect Li. Other applied surface analysis techniques are X-ray photoelectron spectroscopy (XPS)27–31 and secondary ion mass spectrometry (SIMS),32–35 however, both techniques usually only provide semi-quantitative information and are limited in terms of sampling depth (i.e., surface effects and not bulk properties are investigated). Additionally, the excellent spatial resolution of SIMS with spot sizes down to the sub-μm range makes the analysis of macroscopic areas very time-consuming.
Also LIBS (Laser Induced Breakdown Spectroscopy) has been used for the elemental analysis of LLZO. Depth profiles,36 2D cross-sectional maps21,27,37 and even 3D distribution images38 have been recorded, showing that LIBS is a very powerful tool for the investigation of sample inhomogeneity close to the sample surface. In these experiments, the bulk composition of the samples (determined by ICP-OES) served as standard for signal quantification, assuming that the chemical composition is constant over the whole sample after reaching a specific sampling depth. Since one of our previous studies revealed macroscopic inhomogeneities in Al stabilized LLZO pellets,19 this assumption does not seem to be correct for all samples, questioning the suitability of this quantification approach. Overall, there is still need for a laterally resolved method for LLZO analysis that provides reliable quantitative information.
A promising option for the chemical analysis of LLZO is coupling of ICP-MS or ICP-OES to LA (laser ablation). While the use of LA prevents time-consuming sample digestion and dilution, the subsequent analysis of the generated aerosol with ICP-OES or ICP-MS allows detection of all LLZO constituents including Li. Furthermore, these methods provide spatially resolved information and have already been used for the quantitative determination of bulk, trace and ultra-trace elements in various advanced materials including ceramics.39,40 While neither LA-ICP-MS nor LA-ICP-OES is widely adapted in the field of Li-ion batteries, both techniques have been successfully applied for the characterization of electrode materials.41–45
Since LA suffers from sample-dependent ablation behavior and elemental fractionation, reliable signal quantification is challenging. However, these issues can be overcome by using matrix-matched calibration standards in combination with appropriate signal normalization strategies.46
The aim of this study is to develop a method for the laterally resolved analysis of Al-stabilized LLZO (Li7−3xAlxLa3Zr2O12) by using LA-ICP-OES. To ensure reliable signal quantification, we prepared matrix-matched standards and characterized them by sample digestion and subsequent liquid ICP-OES measurement. We used these standards to obtain LA-ICP-OES calibrations, and compared the results with parallel LA-ICP-MS measurements. To show the capabilities of the developed method, we recorded quantitative distribution images of an Al stabilized LLZO pellet.
For signal quantification, univariate calibration using certified single element ICP-standard solutions (Certipur®, Merck, Germany) was used. We mixed Al, La, Li, and Zr standards to obtain elemental ratios corresponding to a Li6.4Al0.2La3Zr2O12 sample. By diluting the standard mixture, we prepared calibration standards with an LLZO concentration ranging from 3.6 to 36 mg kg−1. We added an Eu ICP-standard solution (Certipur®, Merck, Germany) to all standard and sample solutions to a final concentration of 0.2 mg kg−1 and used Eu as internal standard for the analysis.
An iCAP 6500 RAD (Thermo Fisher Scientific, USA) equipped with an echelle-type monochromator and a CID detector was used for the ICP-OES analysis. For data acquisition, we used Qtegra software provided by the manufacturer of the instrument. Introduction of the samples was performed using an ASX-520 autosampler (CETAC Technologies, USA), PTFE tubing, and a sample introduction kit consisting of a conventional Meinhard high-solids quartz nebulizer and a quartz cyclone spray chamber without ascension tube. We used a plasma torch containing a quartz injector tube with 1.5 mm inner diameter. Detailed information about the used instrument parameters can be found in Table 1.
Thermo iCAP 6500 RAD | |
---|---|
a Used for quantification. b Normalized to Eu 281.396 nm signal c Normalized to Eu 381.967 nm signal. | |
RF power | 1200 W |
Radial observation height | 12 mm |
Plasma gas flow (Ar) | 12 l min−1 |
Nebulizer gas flow (Ar) | 0.6 l min−1 |
Auxiliary gas flow (Ar) | 0.8 l min−1 |
Integration time | 5 s |
Replicates per sample | 5 |
Purge pump rate | 1.6 ml min−1 |
Sample flow rate | 0.8 ml min−1 |
Analytical wavelengths | |
Eu (internal standard) | 281.396 nm (b) 381.967 nm (c) |
Al | 396.152 nma,c |
La | 333.749 nmb 412.323 nma,b |
Li | 610.362 nmc 670.784 nma,c |
Zr | 339.198 nmb 343.823 nma,b |
For each analyte, we recorded the two most sensitive and non-interfered emission lines. Only one emission line per element was used for quantification, the other one was used for quality control (i.e., check for spectral interferences). In case of Al only one emission line with sufficient signal intensity was accessible (396.152 nm). Since this line suffers from spectral interference caused by a minor Zr emission line, we applied interference correction. For that purpose, we analyzed Zr single element standard solutions with various concentrations to determine the impact of the interference.
For the compensation of differences in sample introduction and instrumental drifts, we used Eu as internal standard and normalized all signals to a corresponding emission line. To further minimize errors in signal quantification, we took two additional measures: (a) we measured the calibration standards not consecutively but separately between the samples and in random order and (b) we repeated the whole analysis three times. The analyte concentrations obtained this threefold analysis were averaged for further data processing.
Laser ablation system | ESI NWR213 |
---|---|
a Used for quantification. | |
Average fluence | |
Pre-ablation | 2.50 J cm−2 |
Analysis | 4.25 J cm−2 |
Laser diameter | |
Pre-ablation | 250 μm |
Analysis | 100 μm |
Scan speed | |
Pre-ablation | 250 μm s−1 |
Analysis | 100 μm s−1 |
Repetition rate | 20 Hz |
Carrier gas flow (He) | 0.6 l min−1 |
Make-up gas flow (Ar) | 0.8 l min−1 |
ICP-OES instrumentation | Thermo iCAP 6500 RAD |
---|---|
RF power | 1200 W |
Radial observation height | 12 mm |
Plasma gas flow (Ar) | 12 l min−1 |
Auxiliary gas flow (Ar) | 0.5 l min−1 |
Integration time | 1 s |
Analytical wavelengths |
![]() |
Al | 309.271 nm 396.152 nma |
La | 261.034 nma 419.655 nm |
Li | 610.362 nma 670.784 nm |
Zr | 257.139 nma 274.256 nm |
ICP-MS instrumentation | Thermo iCAP Q |
---|---|
RF power | 1550 W |
Plasma gas flow (Ar) | 14 l min−1 |
Auxiliary gas flow (Ar) | 0.8 l min−1 |
Dwell time per isotope | 10 ms |
Cones | Ni |
Mass resolution | m/Δm = 300 |
Measured isotopes | 7Li, 27Al, 90Zr, 138La |
For the LA-ICP-OES experiments, the same ICP-OES instrument as for the bulk analysis was used (cf. Section 2.3). In contrast to the liquid measurements, a plasma torch containing a corrosion-resistant ceramic injector tube was used. Data acquisition was performed using iTEVA software (v.2.8.0.96) provided by the manufacturer of the instrument. For each analyte, we recorded two emission lines, where one emission line was used for signal quantification and the other one was used for quality control. Compared to the liquid analysis, we selected less sensitive La and Zr emission lines to avoid detector saturation. Furthermore, we reduced the integration time to 1 s in order to allow fast monitoring of the transient signals provided in imaging experiments. Similar to the liquid measurements, we applied interference correction for the Al 396.152 nm emission lines. In order to determine the impact of the Zr interference, we measured an Al-free YSZ single crystal (ZrO2 stabilized with 9.5 mol% Y2O3, 5 mm × 5 mm × 0.5 mm, Crystec, Germany) prior to the analysis using identical instrument setting.
In case of the LA-ICP-MS measurements, the LA system was coupled to a quadrupole ICP-MS instrument (Thermo iCAP Qc, ThermoFisher Scientific, Germany). For data acquisition, we used Qtegra software (v.2.10.3324.62) provided by the manufacturer of the instrument. Before every experiment, we optimized the measurement parameters concerning the MS instrumentation using NIST 612 trace metals in glass standard (National Institute of Standards and Technologies, Gaithersburg, MD) for maximum 115In signal. Additionally, we monitored the oxide ratio by the 140Ce16O/140Ce ratio, which was below 2.0% for all experiments.
LA-ICP-OES and -MS measurements were quantified using the prepared LLZO pellets as matrix-matched calibration standards. Analysis of these standards was performed using line scan ablation patterns with adjoining lines. Per standard, we placed four laser patterns (each having an area of about 0.4 mm2) on different locations on the pellets. Each laser pattern was ablated three times. For further data processing, we calculated average values for each pattern. In order to remove potential surface contaminations (e.g. Li2CO3 due to air exposure27), we employed a pre-ablation step consisting of a similar line scan patterns prior to the analysis.
Two different calibration strategies were used for signal quantification. On the one hand, we applied conventional univariate calibration in combination with the use of La as internal standard, assuming equal La content and -distribution in all samples. Additionally, we used an internal standard-independent calibration strategy presented by Liu et al.,48 which was adapted for LLZO samples. Based on the consideration that the sum of all metal oxides in LLZO adds up to 100 m%, this approach uses the intensities of all measured analytes expressed as the corresponding oxides (Al2O3, Li2O, La2O3, ZrO2) for signal normalization. The sample composition was calculated using the following eqn (1) and (2)
![]() | (1) |
![]() | (2) |
LA-ICP-OES settings described in the previous section were used for all measurements. To create elemental maps, we used line-scan ablation patterns with adjoining lines. A laser beam diameter of 100 μm and a scan speed of 100 μm s−1 were used, which, taking the washout-time of the ablation cell into consideration, results in images with a lateral resolution of 100 μm. As with the measurement of the calibration standards, we also employed pre-ablation step in the sample analysis for cleaning purposes (e.g., removing residues from SiC grinding paper).
Image processing was performed using the software ImageLab (v.2.41, Epina GmbH, Austria).
Fig. 1 shows the standard preparation and characterization process schematically. We synthesised LLZO pellets with different doping content ranging from 0.0 to 0.4 Al per formula unit (pfu) using a conventional high temperature sintering route. Since the high temperature is likely to affect the chemical composition of the pellets (e.g. due to loss of Li2O or diffusion processes) leading to sample inhomogeneities, the obtained pellets are not suitable for direct use as standard. To overcome this issue, we crushed and ground the sintered pellets and subsequently pressed the obtained powders into pellets. This results in LLZO pellets with constant chemical composition that are suitable for use as matrix-matched standard in LA-ICP-OES measurements. For the characterization of the obtained standards, we used the remaining powders that were not consumed by the pressing of the standards. We transferred the samples into liquid form by using a borax fusion and analysed the obtained liquids via ICP-OES. To obtain results as accurate as possible, we took multiple measures: (a) three replicate digestions for each sample, (b) internal-standard corrections using Eu, (c) random measuring sequence of standards and samples, (d) threefold repeat of the whole measurement including calibration, and (e) interference correction for the main Al emission line (396.152 nm, interfered by a minor Zr line). Although the Zr interference is rather weak and might not be relevant for samples with high doping contents, we still consider it necessary to prevent inaccurate results for samples with low Al concentration.
The determined bulk compositions of all standards are listed in Table 3. The results show significant but reasonable deviations from the intended values. The average Zr/La atomic ratio is 0.650, which is close to the expected value (0.667). Also the Al/La atomic ratios agree well with the intended ratios, except for the nominally Al-free standard. In this case, the standard shows a much higher Al/La ratio of 0.025, which indicates Al incorporation during the sintering in the Al2O3 crucibles. The reason that this can only be observed for the nominally Al-free sample could be related to the fact that Al-substitution stabilizes the cubic LLZO modification.16–18 The strongest deviations from the intended values are observed for the Li/La atomic ratio. This is not surprising, since it is known that Li2O evaporates during sintering, changing the Li content of the pellet. While usually Li excess is used in the starting material mixture to compensate this phenomenon,10,16,17,49 the assumed amount of lost Li might not be correct. Interestingly, the Li/La ratio is lower than desired for the standards with low Al contents, and higher than intended for the Al-rich standards. This means that less Li was lost during the synthesis of samples with a higher Al content, indicating that the Li2O loss depends on the Al content. Overall, the results highlight the importance of monitoring the chemical composition of LLZO samples after high-temperature treatment, not only if they are used as standards, but also for general purposes.
x | Al/La | Li/La | Zr/La | Formulaa |
---|---|---|---|---|
a Calculated, based on atomic ratios. Oxygen contents are estimated from charge balance considerations. | ||||
0.00 | 0.025 ± 0.006 (0.000) | 2.17 ± 0.04 (2.33) | 0.638 ± 0.003 (0.667) | Li6.51Al0.08La3 Zr1.91O11.7 |
0.05 | 0.013 ± 0.008 (0.017) | 2.23 ± 0.10 (2.28) | 0.651 ± 0.006 (0.667) | Li6.69Al0.04La3 Zr1.95O11.8 |
0.10 | 0.029 ± 0.007 (0.033) | 2.18 ± 0.03 (2.23) | 0.650 ± 0.003 (0.667) | Li6.55Al0.09La3 Zr1.95O11.8 |
0.15 | 0.048 ± 0.004 (0.050) | 2.17 ± 0.04 (2.18) | 0.653 ± 0.005 (0.667) | Li6.52Al0.15La3 Zr1.96O11.9 |
0.20 | 0.066 ± 0.004 (0.067) | 2.14 ± 0.06 (2.13) | 0.652 ± 0.004 (0.667) | Li6.42Al0.20La3 Zr1.96O11.9 |
0.25 | 0.083 ± 0.003 (0.083) | 2.10 ± 0.06 (2.08) | 0.651 ± 0.005 (0.667) | Li6.29Al0.25La3 Zr1.95O11.9 |
0.30 | 0.097 ± 0.004 (0.100) | 2.06 ± 0.06 (2.03) | 0.646 ± 0.003 (0.667) | Li6.19Al0.29La3 Zr1.94O11.9 |
0.35 | 0.116 ± 0.008 (0.117) | 2.02 ± 0.05 (1.98) | 0.653 ± 0.003 (0.667) | Li6.07Al0.35La3 Zr1.96O12.0 |
0.40 | 0.137 ± 0.006 (0.133) | 2.00 ± 0.03 (1.93) | 0.655 ± 0.004 (0.667) | Li6.00Al0.41La3 Zr1.96O11.9 |
Fig. 2 shows typical calibrations obtained by such an experiment. Conventional univariate calibrations are displayed where signal normalization was preformed using La as internal standard. A distinct linear correlation and an excellent coefficient of determination (R2 = 0.9988) can be observed for the Al signal (cf.Fig. 2a). The limit of detection (LOD) of the analysis (calculated based on the slope and the standard deviation of the response of the calibration curve) is 0.049 m%, which shows that the selected measurement parameters offer sufficient sensitivity for an accurate analysis of the Al content, even for the lowest dopant levels. In case of Li the coefficient of determination is lower (R2 = 0.9628), however, also here the signal shows a clear correlation (cf.Fig. 2b). It is important to notice that in case of Li the covered mass range is very narrow (highest and lowest standard only differ by 17% relatively) and obtaining a suitable calibration therefore difficult. This once more highlights the importance of a high-precision analysis for this material system.
Beside the LA-ICP-OES calibration, Fig. 2 also displays the results of analog LA-ICP-MS measurements for comparative purposes. While the calibration shows similar quality in case of Al (cf.Fig. 2c), no significant correlation at all can be observed for the Li signal when ICP-MS is used for the analysis (cf.Fig. 2d). This is mostly likely caused by the fact that the measurement of light elements such as Li suffers severely from matrix-induced suppression in the ion beam,50 limiting the precision of the ICP-MS analysis. The lack of precision in case of LA-ICP-MS confirms that LA-ICP-OES is more reliable for the measurement of Li and is therefore the preferred choice for the analysis of LLZO.
An important aspect for reliable signal quantification is the use of appropriate signal normalization, especially if the used calibration standards potentially show different ablation behavior due to differences in the mechanical properties (ablation rate compact samples vs. pressed standards). The most straightforward normalization approach for the analysis of LLZO is the use of La as internal standard, assuming that the same amount of La is present in all standards and samples. This assumption is most likely valid, since La2O3 is very heat resistant (melting point > 2300 °C) and occurrence of La loss during the LLZO synthesis is therefore very unlikely. Our experiments confirm this, since we obtain good LA calibrations when we use La for signal normalization (cf.Fig. 2). However, assuming a constant La content introduces an additional uncertainty, potentially decreasing the precision of the analysis. Furthermore, it is not possible to determine the full stoichiometry of LLZO samples using that approach, since the La content is not investigated.
To overcome this issues, we applied an adapted version of internal standard-independent calibration strategy presented by Liu et al.48 This approach is based on the consideration that the sum of all metal oxides in LLZO adds up to 100 m% and uses intensities of all measured analytes for signal normalization. Since the mass fractions of all cations (including La) are determined, this calibration strategy enables the calculation of all cation atomic ratios, giving access to the complete stoichiometry of the material.
To determine the accuracy of this quantification strategy, we used a cross-validation approach: in a first step, the stoichiometry (expressed as atomic ratios) of each standard is determined using all other standards for the calibration of the measurement. After that, the obtained values are compared with the corresponding target values (derived from the standard characterization via ICP-OES). Fig. 3 shows the results of this cross-validation. In general, excellent agreement between measured and target stoichiometry can be observed. The average deviations for the Al/La, Li/La, and Zr/La atomic ratio are 3.1%, 1.1%, and 0.5%, respectively. Solely the Al/La ratio for the sample with the lowest Al content shows a significantly larger deviation. The reason for this is the low absolute Al concentration in the material, which leads to low signal intensity and therefore large relative measurement uncertainties. This is in agreement with the relatively large error bars for the low-Al standards. Overall, the measurements confirm that the internal standard-independent calibration strategy based on 100 m% is reliable for the analysis of LLZO samples.
Since the aim of this experiment is to investigate the bulk of the material, we took two measures to avoid influences of surface effects: first, we mechanically removed near surface layers that could be affected by Al-enrichment38 or other surface-related composition variations. Second, we employed a pre-ablation step directly before the measurement (ablating ca. 2 μm), removing potential surface alterations caused by the grinding process or contact with air (e.g. Li2CO3 formation27).
The imaging experiment resulted in distribution maps with a lateral resolution of 100 μm in both vertical and horizontal direction. The ablation depth of the analysis was approx. 3 μm. Due to use of the 100 m% normalization strategy, we obtain the mass fractions of all cations (Al, La, Li, and Zr) by the analysis. We converted the mass fractions into the corresponding atomic ratios, which define the stoichiometry of the material.
Fig. 4 shows the obtained distribution images. The maps represent the (locally resolved) sample stoichiometry expressed in form of atomic ratios (Al/La, Li/La, and Zr/La). Significant composition differences between the inner and the outer part of the pellet can be observed. For a better visualization of these variations, the atomic ratios along a selected cross section are plotted next to the corresponding images. To verify that the selected cross-section is representative for the sample, additional cross-sections were investigated, leading to similar results (not shown). As expected, the sample shows a constant Zr/La atomic ratio across the whole pellet (cf.Fig. 4a). Since it is extremely unlikely that the Zr and La content changes in the same ratio, this confirms that Zr as well as La are evenly distributed within the pellet. This is not surprising, since both elements are not prone to evaporation of their oxide, and, moreover, their crystal sites are not affected by Li. Hence, they are most likely not changed by the sintering process. In contrast to that, significant Al and Li variations are observed. The Al/La map shows a strong Al-enrichment in the inner part of the pellet with an increase of Al/La atomic ratio up to 46% compared to the outer part of the sample (cf.Fig. 4b). Considering the fact that Al replaces Li in the LLZO crystal lattice, the Li distribution image should show an opposing trend. Indeed the Li/La image indicates a higher Li content in the outer part of pellet (cf.Fig. 4c), however, the differences are much smaller and hardly visible in the map. The reason why the variations are much harder to see in case of Li is the fact that LLZO contains a relatively high amount of Li. The changes of the Li-content that are induced by Al-doping are therefore relatively small, making the variations less pronounced. Interestingly, the images show that the Li-content is significantly lower than intended, indicating that more Li than expected was lost during the sintering process.
The obtained distribution images demonstrate the existence of stoichiometry variations within the bulk of LLZO samples. These composition changes are of great relevance since they might affect the electrochemical behavior of LLZO, which is crucial for its application as solid electrolyte material. Possible correlations were investigated in one of our recent studies, where we analyzed several LLZO pellets using the presented method and combined the findings with local conductivity measurements.51
In a further step, we determined the bulk composition of the pellet via conventional ICP-OES analysis. For that purpose, we analyzed the remaining pellet using the same procedure as for the characterization of the standard powders.
In order to validate the accuracy of the LA measurement, we compared the average stoichiometry derived from the distribution images with the bulk composition of the sample determined via ICP-OES. The results of the comparison are summarized in Table 4. The determined values agree very well with relative deviations below 2% for the Al/La, Li/La as well as the Zr/La atomic ratios. This excellent agreement confirms the accuracy of the developed LA-ICP-OES analysis.
LA-ICP-OES | ICP-OES (liquid) | Relative deviation | ||
---|---|---|---|---|
Atomic ratio | Li/La | 2.03 ± 0.12 | 2.05 ± 0.05 | −0.9% |
Al/La | 0.068 ± 0.016 | 0.067 ± 0.007 | 1.7% | |
Zr/La | 0.65 ± 0.03 | 0.661 ± 0.006 | −1.4% | |
Formula | Li6.09Al0.20La3Zr1.96O11.8 | Li6.15Al0.20La3Zr1.98O11.9 |
The recorded distribution images revealed macroscopic composition variations within the bulk of Al-stabilized LLZO samples. Beside strong Al inhomogeneities, also significant variations in the Li content were observed. The stoichiometry variations were most likely caused by the high-temperature treatment during the synthesis, leading to loss of Li2O and Al diffusion. Moreover, the Li2O loss depends on the Al content, with less loss for Al-rich samples.
The obtained results demonstrate the importance of spatially resolved sample characterization. Stoichiometry variations might strongly affect the electrochemical behavior of a solid electrolyte material and need to be monitored for that reason. To get a deep understanding of the material properties, combining electrochemical measurements with spatially resolved chemical analysis is necessary.
This journal is © The Royal Society of Chemistry 2020 |