Fast Li-Ion-Conducting Garnet-Related Li7–3xFexLa3Zr2O12 with Uncommon I4̅3d Structure

Fast Li-ion-conducting Li oxide garnets receive a great deal of attention as they are suitable candidates for solid-state Li electrolytes. It was recently shown that Ga-stabilized Li7La3Zr2O12 crystallizes in the acentric cubic space group I4̅3d. This structure can be derived by a symmetry reduction of the garnet-type Ia3̅d structure, which is the most commonly found space group of Li oxide garnets and garnets in general. In this study, single-crystal X-ray diffraction confirms the presence of space group I4̅3d also for Li7–3xFexLa3Zr2O12. The crystal structure was characterized by X-ray powder diffraction, single-crystal X-ray diffraction, neutron powder diffraction, and Mößbauer spectroscopy. The crystal–chemical behavior of Fe3+ in Li7La3Zr2O12 is very similar to that of Ga3+. The symmetry reduction seems to be initiated by the ordering of Fe3+ onto the tetrahedral Li1 (12a) site of space group I4̅3d. Electrochemical impedance spectroscopy measurements showed a Li-ion bulk conductivity of up to 1.38 × 10–3 S cm–1 at room temperature, which is among the highest values reported for this group of materials.


INTRODUCTION
The need for safe, powerful, and efficient energy storage devices has led to a high level of interest in the improvement of battery technologies. Currently, most commercially available Li-ion batteries contain organic polymer-based electrolytes, leading to multiple safety issues such as poor chemical and electrochemical stability, leakage, and flammability. These disadvantages have prevented this technology from reaching maturity. Therefore, next-generation Li-ion battery concepts are based on all-solid-state technologies using ceramic Li-ion conductors as electrolytes. 1−3 Li-stuffed oxide garnets containing more than three Li + ions per formula unit are among the most promising solid-state Li electrolytes. Cubic Li 7 La 3 Zr 2 O 12 (LLZO) is considered as an especially suitable candidate. LLZO shows a high Li-ion conductivity as well as superior chemical and thermal stability. Its electrochemical inertness over a wide potential window and its stability against Li metal makes LLZO an especially promising material for use as a solid-state Li electrolyte. 1,4 Pure LLZO occurs in two structural modifications. The tetragonal low-temperature phase in space group (SG) I4 1 /acd has a fully ordered Li + arrangement and shows a low ion conductivity of ∼10 −6 S cm −1 at room temperature (RT). The cubic high-temperature modification shows SG Ia3̅ d. The garnet-type Ia3̅ d structure consists of a framework of 8-fold coordinated La 3+ (24c) and octahedrally coordinated Zr 4+ (16a). Li + occupies interstitial sites with tetrahedral (24d), octahedral (48g), and distorted 4-fold (96h) coordination. In contrast to the tetragonal modification, the cubic modification with SG Ia3̅ d shows a disordered Li + distribution. The cubic modification can be stabilized at RT by the introduction of supervalent cations. 5−14 It was recently shown that the introduction of Ga 3+ leads to a different structural modification with acentric cubic SG I4̅ 3d (Figure 1). 15,16 The formation of this SG is attributed to the site preference of Ga 3+ , which causes a splitting of the 24d position of SG Ia3̅ d into two different sites, namely, Li1 (12a) and Li2 (12b).
This study demonstrates that LLZO stabilized with Fe 3+ also shows the acentric SG I4̅ 3d. Coarse-grained samples of Festabilized LLZO were prepared by solid-state reaction and characterized by scanning electron microscopy (SEM)/backscattered electron imaging (BSE) and energy-dispersive X-ray spectroscopy. The structural characterization was achieved by a rich portfolio of techniques, including X-ray powder diffraction (XRPD), single-crystal X-ray diffraction (SC-XRD), neutron powder diffraction (NPD), and 57 Fe Moßbauer spectroscopy. Electrochemical impedance spectroscopy (EIS) of Fe-stabilized LLZO reports a Li-ion bulk conductivity of up to 1.38 × 10 −3 S cm −1 at RT, which is among the highest values reported for this group of materials.

EXPERIMENTAL SECTION
Synthesis. A series of Li 7−3x Fe x La 3 Zr 2 O 12 compounds with intended Fe concentrations (x) of 0.10, 0.16, 0.18, 0.20, 0.25, and 0.30 atoms per formula unit (pfu) were prepared by high-temperature sintering in air. The preparation route is equivalent to the procedure described by Wagner et al. 15  containing the natural mixture of isotopes. The starting materials were weighed out in their intended stoichiometric proportions with an excess of 10 wt % Li 2 CO 3 to compensate for the loss of Li 2 O due to evaporation during sintering. The reagents were mixed and ground in an agate mortar and then cold-pressed into pellets. The sample pellets were placed on a pellet of pure LLZO and then put into an alumina crucible. After being heated at a rate of 5°C min −1 to 850°C and sintered for 4 h, the pellets were ground in an agate mortar and ball-milled under isopropyl alcohol for 1 h (FRITSCH Pulverisette 7, 800 rpm, 2 mm ZrO 2 balls). The resulting powder was pressed into pellets and put into an alumina crucible. To avoid the incorporation of Al 3+ from the crucible, the sample pellet was again placed on a pellet of pure LLZO. To reduce the Li 2 O loss of the samples, the sample pellet was covered with another pellet of pure LLZO. The samples were heated at a rate of 5°C min −1 to 1230°C and sintered for 6 h.
SEM/EDX. SEM investigations were performed with a Zeiss Ultra Plus device to study the grain size, morphology, phase composition, and chemical homogeneity, i.e., the distribution of Fe, La, and Zr, by using BSE imaging mode and standard-free EDX measurements with an acceleration voltage of 20 kV. Special regard was given to the detection of Al and Ga, as well. For SEM characterization, polycrystalline chips from the pellets were embedded in an epoxy holder, polished with diamond paste, and coated with carbon.
XRPD. For the preparation of XRPD samples, small fragments from the sintered pellets were ground in an agate mortar. XPRD measurements were taken with a Bruker D8 Advance DaVinci Design diffractometer with a Lynxeye solid-state detector using Cu Kα radiation. Data were collected at 2θ positions between 10°and 80°. Evaluation of XRPD data was performed by Rietveld refinement using Bruker DIFFRACplus TOPAS (version 4.2).
SC-XRD. SC-XRD measurements were performed on a Bruker SMART APEX CCD diffractometer. Single crystals from gently crushed pellets were selected under the binocular on the basis of their optical properties. Selected crystals were glued on top of glass capillaries and tested on the diffractometer. To obtain good statistics, full sets of intensity data were collected on several crystals for each sample composition, resulting in a total of 18 data sets. In general, all tested crystals were of high quality and showed sharp reflections. Graphite-monochromatized Mo Kα radiation was used for data collection. The crystal-to-detector distance was 30 mm. The detector was positioned at 2θ positions of −30°and −50°using an ω-scan mode strategy at four different φ positions (0°, 90°, 180°, and 270°) for each 2θ position. For each run, 630 frames with Δω = 0.3°were collected, so data were acquired in a large Q range up to minimum d values of ≈0.53 Å to obtain accurate anisotropic displacement parameters and to reduce the correlation between atomic displacement parameters and site occupation numbers. Three-dimensional data were integrated and corrected for Lorentz, polarization, and background effects by using the APEX2 software. 17 Structure solution with direct methods and subsequent weighted full-matrix least-squares refinements on F 2 were performed with SHELX-2012 as implemented in the program suite WinGX 2014.1. 18, 19 Further structural analysis and visualizations of the crystal structure were performed by using the VESTA 3 program. 20 NPD. Neutron powder diffraction data for samples with x = 0.18 pfu and x = 0.25 pfu were collected at the Research Reactor BER II at the Helmholtz-Zentrum Berlin. Both samples were stored under Ar directly after the synthesis to prevent the incorporation of H + and other reactions with H 2 O and CO 2 from air. Powder diffraction data were acquired at 1.8 and 298 K in constant wavelength mode using the fine-resolution powder diffractometer FIREPOD with λ = 1.7982(1) Å. 21 Data were treated with the FULLPROF suite of programs. 22 57 Fe Moßbauer Spectroscopy. The experimental spectra were recorded using the standard setup of the Moßbauer lab at the University of Salzburg. A 57 Co/Rh single-line thin source (initial activity of 50 mCi) was used to collect 57 Fe Moßbauer spectra of 57 Feenriched samples with x = 0.18 pfu and x = 0.25 pfu. Transmission powder spectra were recorded at RT with a multichannel analyzer (1024 channels) operating in conjunction with an electromechanical drive system with a symmetrical sawtooth velocity shape. For each composition, 50 mg of powdered sample was placed between plastic discs fitted into a Cu ring with a 10 mm inner diameter covered with Al foil on one side. The two simultaneously obtained spectra (512 channels each) were folded, calibrated against α-iron, and evaluated using RECOIL. 23 EIS. The ionic conductivity was measured by electrochemical impedance spectroscopy (EIS). Thin films of platinum (200 nm) and titanium (10 nm) were sputtered on the top and bottom sides of the sample and act as ionically blocking electrodes. Titanium is needed to improve the adhesion of the platinum layer. For the EIS measurements, a Novocontrol Alpha analyzer was used in the frequency range of 3 × 10 6 to 10 Hz. Cooling and heating of the sample were performed with a Julabo F-25 HE instrument, which is a thermal bath fluid-based cooling and heating device. The temperature, measured by a thermocouple at the sample, was in the range of −7 to 25°C.

RESULTS
Microstructure. For all compositions, coarse-grained samples with grain sizes of >100 μm were obtained. Individual Figure 1. Crystal structure of LLZO with SG I4̅ 3d. Blue dodecahedra represent 8-fold coordinated La 3+ (Wyckoff position 24d) and green octahedra 6-fold coordinated Zr 4+ (16c). Red spheres correspond to tetrahedrally coordinated Li + ions at the Li1 (12a) site; orange spheres represent tetrahedrally coordinated Li + ions at the Li2 (12b) site, and yellow spheres represent distorted 6-fold coordinated Li + ions at the Li3 (48e) site. grains show isometric shapes. Single grains separated by gaps are mainly found in the peripheral part of the pellets. In contrast to this, grains in the central part of the pellets do not show pronounced grain boundaries. Open voids are found as inclusions within single LLZO grains (see Figure S1). In some cases, extra phases are present along grain boundaries. SEM-BSE investigations did not show a zoning of LLZO grains. Chemical Composition As Observed by SEM-EDX. EDX measurements of LLZO grains show the presence of La, Zr, and Fe. For all samples, the La/Zr ratio is <3/2, which indicates a deficiency of La relative to Zr on the order of a few percent compared to the intended overall La content. Fe contents of individual LLZO grains within one and the same pellet vary slightly. The average Fe content of the LLZO grains is slightly lower than the intended Fe content. In some cases, an extra phase containing La, Fe, and O with La > Fe was observed along grain boundaries. EDX mapping did not show a zoning of LLZO grains with regard to La, Zr, and Fe. No evidence of contamination with Al or Ga was indicated by EDX.

Chemistry of Materials
XRPD Crystal Structure According to SC-XRD. Single-crystal Xray intensity data processing gives strong evidence of the cubic crystal system for all 18 data sets collected. For samples with a refined Fe content of x > 0.09, the analysis of intensity statistics and systematic extinctions yields the acentric SG I4̅ 3d (No. 220). Therefore, for the garnet family, atypical SG I4̅ 3d is found not only within the LLZO:Ga series, as described in great detail very recently by Wagner et al., 15 but also for the incorporation of Fe 3+ into the LLZO structure. 15 For more details about differences between common Ia3̅ d and the acentric I4̅ 3d structure, the reader is referred to ref 15. Only for samples with

Chemistry of Materials
Article a refined Fe content of x < 0.09 is the common LLZO SG Ia3̅ d observed. This is in accordance with the LLZO:Ga series, where for small Ga 3+ substitutions SG Ia3̅ d is observed. Higher degrees of substitution force a reduction in symmetry due to ordering of Ga 3+ onto one of the two possible tetrahedral sites. For structural refinement, the models of Wagner et al. 15 for Ia3̅ d and I4̅ 3d structures were used as starting points. 15 In the final structural models, all atoms could be treated with anisotropic atomic displacement parameters and data sets could be refined for all data to wR 2 values of <6%. Experimental data and results of structural refinement for selected samples are reported in Table S1; the fractional atomic coordinates, occupation numbers, and equivalent isotropic atomic displacement parameters are listed in Table 1 and Table  S2, while selected bond lengths and angles are compiled in Table S3.
To summarize in brief, the crystal structure of Fe 3+substituted LLZO, SG I4̅ 3d, exhibits seven different atomic positions: La 3+ occupies the 8-fold coordinated 24d position (site symmetry 2..), and Zr 4+ is located at the octahedrally coordinated 16c position (site symmetry .3.). These positions are similar to those of the Ia3̅ d structure. Li + is found at three different positions in I4̅ 3d. Two of them correspond to the regular tetrahedral coordinated site of the silicate garnets (24d in Ia3̅ d); however, they split into two positions, namely, Li1 (12a) and Li2 (12b), both with site symmetry 4̅ .., but they differ in bond lengths and polyhedral distortion. The third one is found at general position 48e, with site symmetry 1. This latter position is only partly occupied as is also common in LLZO compounds. Different from common silicate garnets, the acentric structure exhibits two independent oxygen atom positions, both in general position 48e. A graphical representation of the crystal structure of LLZO with SG I4̅ 3d is given in Figure 1.
For the Ia3̅ d and I4̅ 3d structure, site occupation refinements yield a small deficit of La 3+ ; thus, the site occupation factor was allowed to vary freely during the refinements. The finding of a small La 3+ deficit is consistent with results from EDX analysis and also was reported for LLZO:Ga. The number of La 3+ vacancies varies within the I4̅ 3d structure from 0.12(2) pfu in Fe 3+ -poor samples to 0.04(2) pfu in Fe 3+ -rich samples, and there is some good evidence that the number of vacancies steadily decreases with an increase in Fe 3+ content ( Figure 3). Interestingly the two data points for the Ia3̅ d structure are off of the trend and show a smaller La 3+ deficit.
During refinement, the occupation factor of the octahedral 16c site, hosting Zr 4+ , was allowed to refine freely. For samples with low Fe 3+ concentrations in the I4̅ 3d structure, the results show a very small number of Zr 4+ vacancies of <0.04 pfu on the 16c site when considering only Zr 4+ . This might be overcome by placing Fe 3+ at this site and assuming full occupancy. However, there is strong evidence from NPD data that the octahedral site is occupied by Zr 4+ only. Also, in Mossbauer spectra, no evidence of octahedral Fe 3+ was found.
For the Ia3̅ d structure, strong evidence has shown that Fe 3+ exclusively substitutes onto the tetrahedrally coordinated 24d position. To develop a site occupation model for the I4̅ 3d structure, a multiple-step process was used, which is described in detail in the Supporting Information and in Figure S2. It turned out that Fe 3+ distinctly prefers the Li1 (12a) site, and only for higher overall Fe 3+ concentrations are small amounts of ferric iron present at the Li2 (12b) site (Figure 3). This observation is fully in agreement with the results of Moßbauer spectroscopy, where indications of a second Fe 3+ tetrahedral doublet with low intensity were found. Generally, both sites Li1 (12a) and Li2 (12b) show some vacancy concentrations, which were estimated to be 30% for Fe 3+ concentrations of <0.15 pfu and decrease somewhat to 25% with an increasing overall Fe 3+ content. The distinct preference of certain trivalent substituents for the Li1 (12a) site was already observed for LLZO:Ga. Again, it is assumed that the strong preference of Fe 3+ for the Li1 (12a) site causes the reduction in symmetry from Ia3̅ d to I4̅ 3d.
For low overall Fe 3+ concentrations, a large number of vacancies is observed at the Li2 (12b) site. The electron density progressively increases at the Li2 (12b) site with an increase in x, most probably because of an increasing amount of Li + present at this site. This is different from the case for the LLZO:Ga series, where higher occupations were found also for low Ga 3+ concentrations. For higher Fe 3+ concentrations, the Li + content of the Li2 (12b) site reaches a saturation level at 1.1 pfu of Li + , assuming an additional electron density contribution from small amounts of Fe 3+ at this site.
The occupation of the Li3 (48e) site generally was refined freely and gives a partial occupation of this site with Li + . Even if the data show distinct scatter and the occupation number is highly correlated with the anisotropic displacement parameters, it is possible to deduce a decrease in the Li + occupation of the Li3 (48e) site with an increase in Fe 3+ content at the Li1 (12a) site. The available data tentatively suggest that the substitution mechanism may follow the reaction [48e] Li 4 Analysis of equivalent isotropic atomic displacement parameters shows that Zr 4+ generally has the smallest atomic displacement, followed by La 3+ . Comparing the two oxygen ions, the atomic motion of the O2 position is larger than that of the O1 site. Different from the case for O1, the O 2− ion on the O2 site is bonded to the less localized Li2 (12b) site, which might result in greater motion. Generally, the atomic displacement parameters decrease with an increase in Fe 3+ content, indicating that Fe 3+ stabilizes the I4̅ 3d structure more and more. Li + ions show large atomic displacements with distinct anisotropic character of the displacement ellipsoids, especially at low substitutional levels at the Li2 (12b) site. With  an increase in Fe 3+ content, the equivalent isotropic atomic displacements decrease especially at the Li2 (12b) and Li3 (48e) sites, meaning the Li + ions on these sites become more precisely located ( Figure S3). Crystal Chemistry. The incorporation of Fe 3+ into the structure of pure LLZO stabilizes the cubic structure. The lattice parameters thereby are reduced from 12.9831(4) to 12.9757(1) Å, and the decrease is almost linear (Figure 4). As for the LLZO:Ga series, there is no evident change in the slope when switching from Ia3̅ d to I4̅ 3d symmetry. However, the slope of the data trend for the Fe 3+ substitutional series is somewhat less steep than that for the LLZO:Ga series. For LLZO:Fe, a linear regression yields a 0 = −0.0556x Fe + 12.987 (R 2 = 0.9876), i.e., a slope of −0.0556 Å x Fe −1 , while for LLZO:Ga, the linear regression yields a 0 = −0.0608x Ga + 12.987 (R 2 = 0.9929), i.e., a slope of −0.0608 Å x Ga −1 . This trend is consistent with the ionic radii of tetrahedrally coordinated Fe 3+ (0.49 Å) and Ga 3+ (0.47 Å), replacing the larger Li + (0.59 Å). 24 The variations of bond lengths and distortion parameters for the LLZO:Fe series follow closely the trends observed and described in detail for the LLZO:Ga series. 15 As for the LLZO:Ga series, the Li1 (12a) and Li2 (12b) tetrahedral sites in I4̅ 3d are different in bond length, tetrahedral distortion, and volume ( Figures S4 and S5). The Li1 (12a) site shows smaller Li−O bond lengths and volumes; however, it displays distinct larger tetrahedral angle variance and quadratic tetrahedral elongation. It is worth noting that on the small substitutional level, the data points for both series plot well into the same trends. On the basis of site occupation refinements, the trivalent cations distinctly prefer the smaller tetrahedron. The distance between neighboring Li2 and Li3 decreases from 2.38 to ∼2.34 Å, while for the Li1−Li3 distance, this effect is not that clear ( Figure S6). For the Li3 (48e) site, the average Li−O distance decreases with an increase in Fe 3+ content and stabilizes at a value of ∼2.085 Å. The trends for the individual Li3−O bonds are diverse; some increase, and some decrease with an increase in trivalent cationic content. As shown in Figure S7, the overall polyhedral volume of the Li3 (48e) site (as LiO 6 ) decreases with an increase in M 3+ (M = Ga or Fe) level, which may be seen in light of the decrease in Li content.

Chemistry of Materials
NPD. Refinement of NPD data was performed on the basis of the structural model derived from SC-XRD and yields a very good coincidence (see Figure S8A and Table S4). The preference of Fe 3+ for the Li1 (12a) site is also indicated by the results of NPD data refinement. The comparison of NPD measurements of one and the same sample at different temperatures, 1.8 and 298 K, shows that the occupation of Li1 (12a) and Li2 (12b) is temperature-dependent. At very low temperatures, the Li2 (12b) site is almost fully occupied while the Li1 (12a) site shows very low occupation. At RT, the occupancy of the Li2 (12b) site is decreased whereas the Li1 (12a) site hosts more Li + than at 1.8 K. In addition to these observations, the residual nuclear density of the low-temperature measurement of the x = 0.25 sample suggests the presence of two additional Li + positions that were not observed by SC-XRD (see Figure S8B and Table S5). Results of crystal structure refinement at 298 and 1.8 K indicate a distinct distribution and movement of Li + even at low temperatures (see Figure S9). A detailed evaluation and interpretation of the NPD results is given in the Supporting Information.
Fe Distribution According to 57 Fe Moßbauer Spectroscopy. The results of 57 Fe Moßbauer spectroscopy are listed in Table 2 and shown in Figure 5. XRPD of 57 Fe-enriched LLZO:Fe samples with intended Fe contents of 0.18 and 0.25 pfu showed pure phase cubic LLZO. Moßbauer spectra of both samples are very similar to the LLZO:Fe spectra reported previously. 13,14 The high background is attributed to poorly crystalline Fe-bearing impurity phases that were also detected by SEM-EDX. A satisfying fit of the spectra is achieved by using tetrahedrally coordinated Fe 3+ only. These results are in full agreement with the findings obtained from SC-XRD. Aside from a broad doublet used to model the background, both samples can be fitted by using two doublets showing an isomer shift δ ≈ 0.20 mm s −1 and a quadrupole splitting ΔE Q ≈ 0.8− 1.0 mm s −1 . For LLZO and other garnet-like materials, similar values have been interpreted as tetrahedrally coordinated Fe 3+ on the 24d site. 13,14,25,26 Under consideration of the results of SC-XRD, the results of 57 Fe Moßbauer spectroscopy obtained by this study are interpreted as Fe 3+ on the Li1 (12a) and Li2 (12b) sites of LLZO:Fe with SG I4̅ 3d. As the quality of the obtained spectra is comparatively poor, the quantitative interpretation of the Moßbauer spectra must be treated with caution. However, the results indicate that Fe 3+ is unequally distributed over these two tetrahedral sites, which coincides with the structural model derived from SC-XRD. It has to be emphasized that attempts to include octahedrally coordinated Fe 3+ in the fit were not successful. Therefore, we conclude that no significant amounts of Fe 3+ are located on octahedral sites. In contrast to other studies of LLZO:Fe, neither indications of strongly distorted tetrahedrally coordinated Fe 3+ on the 48e site of SG I4̅ 3d, equivalent to the 96h site of SG Ia3̅ d, nor magnetic subpatterns were observed in this study. 13,14

Article
Li-Ion Conductivity As Seen by EIS. Impedance spectra of the Fe-stabilized LLZO garnets in Figure 6A display measurements at the lowest temperature (−7.3°C). The spectra show parts of a high-frequency semicircle and a strong increase of the imaginary part toward lower frequencies. The latter can be attributed to the ionically blocking electrode. The high-frequency part can be described by a resistor (R 1 ) in parallel to a constant phase element (CPE1) with impedance where ω is the angular frequency and Q and n are fit parameters. In series to R 1 −CPE1, an inductance due to wiring is required for data analysis. In these spectra, there is no indication of interface or grain boundary effects, which would cause a second semicircle in the Nyquist plots. 12 When the blocking electrode is represented by another constant phase element, the circuit in Figure 6A results, which allows excellent fits of all data (see Figure 6A).
Resistance R 1 is interpreted in terms of a bulk transport process. A strong argument in favor of this assumption is the capacitance associated with the CPE according to 27 On the basis of the obtained fit values for CPE1, the calculated capacitance is in the range of 10 pF. Using this value, sample area A, and thickness d, a relative permittivity can be calculated according to This leads to values on the order of 55−80, obtained for the measurements at the lowest temperature. These are realistic values for oxides and support the interpretation of the arc as a bulk property. 12 Therefore, resistance R 1 can be used to determine the Li-ion conduction in the bulk by As can be seen in Figure 6B, the high-frequency semicircle in the impedance spectra vanishes upon heating of the sample to room temperature. The reason for this change is the increasing relaxation frequency of the LLZO bulk ω = 1/RC due to the temperature-dependent resistance. Nevertheless, the equivalent circuit in Figure 6A can also be applied for higher temperatures, and bulk resistances can be extracted for all temperatures.
The Arrhenius graph in Figure 7 shows that the activation energy is ∼0.29 eV for the samples. Very high ionic

Chemistry of Materials
Article conductivities result at RT for the x = 0.25 pfu sample, reaching 1.38 × 10 −3 S cm −1 at 23.5°C (Table 3). This value is one of the highest bulk conductivities at RT measured for LLZO garnets so far (σ = 1.3 × 10 −3 S cm −1 for Li 6

DISCUSSION
It was shown that the substitution of Li + with ferric iron in LLZO induces a reduction in symmetry to SG I4̅ 3d. Thereby, Fe 3+ strongly prefers the Li1 (12a) site; this site preference is assumed to be a main reason for the change in symmetry. In light of recent observations for the LLZO:Ga and LLZO: (Al,Ga) series, the appearance of the I4̅ 3d symmetry also in the LLZO:Fe series might not be astonishing. 15,16 While the small Al 3+ keeps Ia3̅ d symmetry, the larger Ga 3+ and even larger Fe 3+ show the change in symmetry, indicating a strong influence of steric size effects on inducing the symmetry change. Indeed, Ga 3+ and Fe 3+ prefer the larger and more distorted Li1 (12a) tetrahedral site. One might speculate that a substitution with Sc 3+ for Li + should be suitable for achieving I4̅ 3d symmetry for LLZO, as well; however, until now, Sc 3+ -stabilized LLZO has not been synthesized successfully.
Even if refinement of Li site occupancies is delicate using Xray diffraction, it could be shown, in analogy to the LLZO:Ga series, that the incorporation of Fe 3+ scales with a reduction of the Li + content mainly at the Li3 (48e) interstitial site rather than with a reduction of the Li + content on the Li1 (12a) or Li2 (12b) tetrahedral site. This is in line with observations for SG Ia3̅ d. 29−31 A phase transformation from SG Ia3̅ d to SG I4̅ 3d caused by protonation as recently reported by several authors can be excluded for this study. 32−34 Samples were either immediately characterized after their synthesis or stored under Ar. Phase transformations due to protonation were reported either after a special treatment or after storage at RT for several years. In addition, a Li + /H + exchange leads to an increase in lattice parameter a 0 to values of >13 Å. As the lattice parameter values reported in this study are in the range of 12.97 Å and thus well in accordance with those of other unaltered LLZO samples, we do not expect any significant protonation of the LLZO:Fe samples used in this study.
It is conspicuous that both members of the LLZO group with SG I4̅ 3d show superior ionic conductivity compared to that of the LLZO:Al group with SG Ia3̅ d, even if the differently stabilized samples were synthesized by the very same preparation route. 16 Rettenwander et al. 16 already discussed possible explanations for the improved ionic conductivity of LLZO:Ga compared to that of LLZO:Al. Besides other factors, they also supposed that the phase transformation from SG Ia3̅ d to SG I4̅ 3d partially accounts for the enhanced electrochemical properties. The results of this study reinforce these assumptions, as the ionic conductivities of LLZO:Fe reported in this study are among the highest values reported for this class of materials. However, further electrochemical investigations will be needed to evaluate the aptitude of LLZO:Fe as a solidstate electrolyte, as the electrochemical stability of the transition element cation Fe 3+ might be inferior compared to those of other stabilizing cations such as Al 3+ and Ga 3+ .

CONCLUSIONS
This study reveals that Fe-stabilized LLZO shows the acentric cubic SG I4̅ 3d (No. 220). This crystal structure was recently reported for Ga-stabilized LLZO and is different from those of other members of the Li oxide garnet group that show SG Ia3̅ d (No. 230). Similar to that of Ga-stabilized LLZO, the phase transformation seems to be caused by the site preference of Fe 3+ . Results of 57 Fe Moßbauer spectroscopy confirm the preference of Fe 3+ for the tetrahedral Li1 (12a) site. Electrochemical impedance spectroscopy exhibits very high Li-ion bulk conductivity up to 1.38 × 10 −3 S cm −1 at RT. These results confirm that garnet-similar LLZO materials with SG I4̅ 3d show superior Li-ion conductivity compared to that of conventional LLZO garnets with SG Ia3̅ d. Again, it must be highlighted that understanding the structure property relationship is essential for understanding the electrochemical properties of these electrolyte materials.  Figure  S6), behavior of the LiO 6 polyhedral volume ( Figure S7), refinement of NPD data using different structural models ( Figure S8), residual nuclear density of LLZO:Fe at different temperatures ( Figure S9), experimental setup and results of the crystal structure refinement of SC-XRD data (Table S1), atomic coordinates and atomic displacement parameters of selected LLZO:Fe samples (Table S2), selected bond lengths and distortion parameters (Table S3), results of Rietveld refinement of NPD data (Table S4), results of Rietveld refinement of NPD data using a structural model with five Li positions (Table S5), a detailed description of the development of a site occupation model for tetrahedral sites, and a detailed discussion of NPD results (PDF)

Chemistry of Materials
Article