Oxygen Vacancies in Fast Lithium-Ion Conducting GarnetsOxygen Vacancies in Fast Lithium-Ion Conducting Garnets Markus Kubicek,*,† Andreas Wachter-Welzl,† Daniel Rettenwander, Reinhard

Oxygen Vacancies in Fast Lithium-Ion Conducting GarnetsMarkus Kubicek,*,† Andreas Wachter-Welzl,† Daniel Rettenwander,§ Reinhard Wagner,‡

Stefan Berendts,∥ Reinhard Uecker,⊥ Georg Amthauer,‡ Herbert Hutter,† and Jurgen Fleig†

†Institute of Chemical Technologies and Analytics, Technische Universitat Wien, Getreidemarkt 9/164EC, 1060 Vienna, Austria‡Department of Chemistry and Physics of Materials, University of Salzburg, Jakob Haringer Straße 2a, 5020 Salzburg, Austria§Center for Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge,Massachusetts 02139-4307, United States∥Department of Chemistry, Technische Universitat Berlin, Straße des 17. Juni 124, 10623 Berlin, Germany⊥Leibniz Institute for Crystal Growth (IKZ), Max-Born-Straße 2, 12489 Berlin, Germany

ABSTRACT: Fast Li-ion conducting garnets have shown excellent performance aschemically stable solid state Li electrolytes even at room temperature. However, becauseof phase formation and Li loss during preparation, reliably obtaining high Li-ionconductivities remains challenging. In this work, we show that an additional defect chemicalspecies needs to be considered, namely, oxygen vacancies. We prove the existence of oxygenvacancies in all six investigated sample types: Ta-, Al-, and Ga-stabilized cubic Li7La3Zr2O12(LLZO) polycrystals and Ta-stabilized LLZO single crystals. Isotope exchange three-dimensional analysis was used to characterize surface oxygen exchange (k*) and bulk oxygendiffusion (D*) enabled by the oxygen vacancies present in the LLZO variants. Remarkablyhigh k* values of 10−11−10−8 cm s−1 and D* values of 10−15−10−11 cm2 s−1 were found at350 °C in air. In a further data analysis, the differences between the compositions areinvestigated, the concentration of oxygen vacancies is estimated, and the possible effects onthe cation defect chemistry and phase formation of LLZO are discussed.

■ INTRODUCTION

Replacing today’s liquid-electrolyte Li-ion batteries with all-solidstate batteries is highly desirable to avoid safety and durabilityissues such as dendrite-driven short circuit or thermal runaway.Since its introduction by Murugan et al. in 2007,1 the garnetLi7La3Zr2O12 (LLZO) has received a great deal of attention as asolid electrolyte with significant Li-ion conductivity at roomtemperature in air.2−5 In numerous doping studies since then itwas attempted to optimize the cation composition of LLZO withrespect to Li-ion conductivity and stability.6,7 Essential is thestabilization of cubic phase(s) and avoiding the tetragonal phasebecause cubic LLZO shows Li-ion conductivity that is ∼2 ordersof magnitude higher.2,6,8 Also the influence of moisture and CO2on the stability of different LLZO compositions requiresattention, because LLZO can degrade or decompose in ambientair.9,10 A common strategy is to introduce substituents acting asdonors such as Al3+, Fe3+, or Ga3+ on the Li+ sites or Nb5+, Ta5+,Bi5+, or Mo6+ on the Zr4+ site and thereby reduce the Listoichiometry per formula unit from 7 to an optimum usuallybetween 6 and 7, depending on the cation substituent.11−25 Agreat challenge in the preparation of highly conductive LLZO isoptimizing the synthesis route. On one hand, high temperaturesare necessary to form the garnet phase; on the other, Li loss viavolatile Li compounds is commonly observed at high temper-atures.26 Therefore, an excess of Li and/or a protective coveringto slow Li loss is regularly used during sintering to finally acquirethe desired LLZO composition.27,28 In most of today’s research,optimizing the cation compositions of Li garnets is attempted for

improving the properties of LLZO, while oxygen anionstoichiometry is considered to be fixed at 12 oxygen atoms performula unit and therefore largely ignored.In this work, we show that oxide anion defects indeed exist in

LLZO and that their contribution to the total defect chemistry ofLLZO cannot be neglected. Only a few studies are known to theauthors that speculate about the existence of oxygen vacancies inLLZO or consider them to potentially play a role in the phaseformation and defect equilibria of LLZO.13,29−31 The mainargument for formation of oxygen vacancies given there is thatLi+ loss is connected to simultaneous O2− loss due to chargeneutrality. Here, we give direct proof that indeed oxygenstoichiometry can vary in LLZO. Via isotope exchange depthprofiling using 18O2 as a stable isotope tracer and by subsequenttime of flight secondary-ion mass spectrometry (ToF-SIMS)analysis, we verify that oxygen vacancies are present or evenabundant in all investigated LLZO materials, including singlecrystals, polycrystals, and different cation substituents (Ta5+,Ga3+, and different Al3+ concentrations). The oxide tracerdiffusion coefficient at 350 °C is surprisingly high (up toD* = 8.2× 10−12 cm s−1) and is even close to that of yttria-stabilizedzirconia, a fast oxygen-ion conductor. Consequently, we showthat oxygen vacancies need to be considered to understand theLi-ion conductivity of LLZO. Acting as donors, they directly

Received: March 29, 2017Revised: August 8, 2017Published: August 11, 2017

Article

pubs.acs.org/cm

© 2017 American Chemical Society 7189 DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

affect the defect chemistry of LLZO. Therefore, we expect thetrue Li content of LLZO compositions to be generally lower thanthe values reported in the literature. Also, elastic effects on thecrystal lattice due to the different size of oxide ions and oxygenvacancies require that attention be paid to influencing phaseformation and Li mobility. Oxygen vacancies could turn out to bean important, previously hidden parameter to explain the largescatter in Li-ion conductivities found in the literature for thesame nominal LLZO compositions.

■ MATERIALS AND METHODSA total of five different LLZO compositions were analyzed, either singlecrystals (∼3 mm edge lengths) or polycrystalline pellets (Ø 7 mm × 2mm). The single-crystal Ta−LLZO samples were grown by theCzochralski method directly from the melt. The starting materials,Li2CO3 (99%,Merck), La2O3 (99.99%, Aldrich), ZrO2 (99.0%, Aldrich),and Ta2O5 (99.99%, Aldrich), were dried, mixed in a stoichiometric ratiowith a 10 wt % excess of Li2CO3, and then pressed and sintered at 850 °Cfor 4 h at a heating rate of 5 °C min−1. The pellet was then melted in aniridium crucible by RF induction heating using a 25 kW microwavegenerator. An iridium rod was used as a seed for crystal growth under anitrogen gas atmosphere. The seed pulling and rotation rates were 1.5mm h−1 and 10 rpm, respectively. The transparent crystal was then cutinto ∼3 mm × ∼3 mm × ∼3 mm cuboidal LLZTO samples.32

The synthesis route for polycrystalline pellets is based on theprocedure described by Wagner et al.33 Li2CO3 (99%, Merck), La2O3

(99.99%, Roth), and ZrO2 (99.0%, Roth) were weighed to reach theintended stoichiometry. To compensate for Li loss during heattreatment, an excess of 10 wt % Li2CO3 was added with respect to thestoichiometrically required amount of Li2CO3. The reagents wereground and mixed in an agate mortar after the addition of isopropylalcohol and subsequently pressed into pellets. The pellets were heated to850 °C at a rate of 5 °C min−1 and calcinated for 4 h. The resultingpellets were again ground in an agate mortar and ball-milled for 1 hunder isopropyl alcohol (FRITSCHPulverisette 7, 800 rpm, 2mmZrO2

balls). After being dried, the powder was pressed again into pellets andplaced in an alumina crucible. To avoid undesired incorporation of Al3+

from the crucible and to suppress evaporation of lithium from thesample, the actual pellets were always placed between two pellets ofstoichiometric Li7La3Zr2O12. The final sintering step was performed byheating with a rate of 5 °Cmin−1 and holding at 1230 °C for 6 h. Typicaldensities of these samples are 92−93%, as measured with a He-pycnometer. Polycrystalline pellets were formed with different cationsubstituents, Ta (Li6La3ZrTaO12), Ga (Li6.4Ga0.2La3Zr2O12), and Al, forwhich three compositions (Li7−3xAlxLa3Zr2O12, where x = 0.15, 0.20, or0.30) were investigated. X-ray diffraction was performed using θ−2θscans (X’Pert PRO PW 3050/60, PANalytical) and 2° grazing incidenceX-ray diffraction (GI-XRD). Isotope exchange experiments wereperformed at 350 °C in 200 mbar 18O2 (97.1% isotope-enriched,CAMPRO) in a special quartz setup using samples with a polishedsurface. The elevated temperature is necessary for oxygen ions tobecome sufficiently mobile for exchange and diffusion; 350 °C waschosen as the upper limit as no mass loss of LLZO can be measured bythermogravimetric analysis up to this temperature26 and therefore nochanges in O2− stoichiometry from room temperature are expected.Subsequent three-dimensional (3D) analysis of the oxygen isotopedistribution was performed on a TOF.SIMS 5 instrument (ION-TOF)using 25 kV Bi3

++ primary ions (∼0.03 pA), 2 kV Cs+ for sputtering(∼155 nA), and a low-energy electron gun (20 V) for chargecompensation. Negative secondary ions were measured. Details onthe used measurement mode (“CBA”mode) are given in refs 34 and 35.Areas of 100 μm × 100 μm were analyzed, and sputter crates were 300μm × 300 μm wide and up to 7.9 μm deep to avoid any possibleinfluence of surface roughness on the profiles. Depth information wascalculated from sputter currents and times and referenced by depthmeasurement of deep craters by confocal microscopy (Axio CSM 700,Zeiss).

■ RESULTS AND DISCUSSIONStructural Characterization. Polycrystalline samples were

investigated by XRD and GI-XRD after the 18O2 experiments.Figure 1 displays the smoothed XRD patterns exemplarily for Ta-

and Ga-doped LLZO as well as a reference pattern of cubicLLZO.36 Each sample shows the reflections of the cubic garnetstructure. For Ta-doped LLZO, one minor additional peak wasfound at 2θ = 21°, which could not be explained, despite the factthat no more impurities were found. Formation of the cubicphase could be shown for all LLZO samples. To investigate ifsecondary phase formation in air plays a role in our isotopeexchange experiments, additional grazing incidence XRDmeasurements were performed after polishing a Ga−LLZOsample and keeping it in ambient air for 48 h. The sampling depthfor XRD is ∼10 μm and for GI-XRD ∼0.5 μm. As shown inFigure 1, no secondary phase formation could be detected by GI-XRD on the Ga−LLZO sample.The microstructure of the samples was investigated by

scanning electron microscopy (SEM) as shown for four differentsample types in Figure 2. For Ga−LLZO and Al−LLZO in theoverview images in panels a and b of Figure 2, inhomogeneities ofthe size of a few micrometers are visible. We expect thoseinhomogeneities to be only very close to the surface as nosecondary phases could be detected via either XRD or GI-XRD.The magnified images (Figure 2c−f) show smooth surfaces withonly scratches from polishing. No cracks or grain boundariescould be detected. For Ga−LLZO and Ta−LLZO, smallinhomogeneities are visible.

Characterization of the Oxygen Nonstoichiometry. Inthis work, isotope exchange 3D profiling was applied, using thestable oxygen isotope 18O for an oxygen exchange experiment at

Figure 1. XRD patterns of Ta− and Ga−LLZO (Li6La3ZrTaO12 andLi6.4Ga0.2La3Zr2O12, respectively). The asterisk shows a minor impurityphase found for Ta−LLZO. For Ga−LLZO, a grazing incidencemeasurement is also shown. The sampling depth is ∼10 μm for XRDand ∼0.5 μm for GI-XRD. For comparison, a reference diffractionpattern of cubic LLZO is shown at the bottom.36

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

7190

350 °C. With subsequent TOF-SIMS 3D profiling at roomtemperature, we then image the frozen-in isotope distribution.The underlying principle behind this isotope exchange techniqueis that incorporation and diffusion of oxide ions require defects inthe LLZO lattice. These defects can be either oxygen vacancies oroxygen interstitials. For two reasons, we exclude oxygeninterstitials as being relevant in our experiments. (i) The oxideion is relatively large like most anions, and the involved largelattice distortions upon addition of an additional ion makeoxygen interstitials energetically unfavorable in tightly packedstructures. They are known to exist in layered structures, whichLLZO does not adopt. (ii) Analyzing the system from a defectchemistry point of view, it is known that Li can evaporate at thehigh temperatures involved in the preparation of LLZO singlecrystals or polycrystalline pellets. The thereby missing Li+ ions(Li vacancies) in LLZO act as negatively charged acceptordefects, just the same as oxygen interstitials. The existence ofboth defects in significant numbers would again be energeticallyunfavorable. If the argument is turned around, however, oxygenvacancies would act as a counterbalancing defect to Li vacancies

(positively charged, donor-type) and are therefore much morelikely to exist in LLZO.For our chosen method and the investigated LLZO

compositions, we therefore ascribe incorporation of 18O intoLLZO to a surface process in which oxygen vacancies areinvolved. The kinetics of this surface process can be quantified bythe tracer surface exchange coefficient (k*). Accordingly, weascribe the transport in LLZO to an oxygen vacancy diffusionmechanism that can be quantified by the tracer diffusioncoefficient (D*). Significant incorporation and diffusion of the18O isotope were observed in all investigated LLZOcompositions, proving the existence of oxygen vacancies asdescribed below.

Ta-Substituted LLZO. For the Ta−LLZO compositionLi6La3ZrTaO12, two different sample types were investigated,namely, single crystals and polycrystalline sintered pellets. InFigure 3, the results of isotope exchange experiments and 3D

SIMS profiling on polycrystals are shown. The graph in Figure 3shows the relative 18O isotope concentration with depth, whilethe insets show the lateral 18O distribution that was in goodapproximation homogeneous for the samples. The red fit curvesare non-linear least-squares fits of the data using an analyticalsolution of Fick’s diffusion equation for a semi-infinite solid and aconstant 18O concentration in the gas.37 Via this fitting, it ispossible to quantify the surface exchange and diffusioncoefficients of oxygen. The very first measured points close tothe surface (shown in light colors in Figure 3) show a stronglyreduced 18O concentration, which we attribute to effects after theisotope exchange experiment. As the LLZO samples were in airafter the isotope exchange experiment and before the SIMSmeasurement, we expect surface hydroxides and carbonates toform29,38,39 with oxygen from air that has a natural abundance of18O of only 0.00205 (see Figure 3). Therefore, the 18O content ofsuch surface layers can be expected to be depleted, and thoselayers were consequently ignored in the analysis. The same effectwas observed for Ga−LLZO as shown below. The tracer surfaceexchange coefficient at 350 °C (k*) was calculated to be 4 ×10−11 cm s−1, and the tracer diffusion coefficient (D*) was 1.5 ×10−12 cm2 s−1.

Figure 2. SEM images of the different LLZO samples. (a and b)Overview images for Ga−LLZO and Al−LLZO, respectively. Here,inhomogeneities of the size of a few micrometers are visible. We expectthose to be only very close to the surface as no secondary phases werepresent in either XRD or GI-XRD. The magnified images (c−f) showsmall inhomogeneities for Ga−LLZO and Ta−LLZO as well asscratches from polishing on all samples. No grain boundaries or possiblesecondary phases are visible therein.

Figure 3. Isotope exchange depth profile of a Li6La3ZrTaO12polycrystalline pellet. The isotope exchange was performed at 350 °Cin 200mbar 18O2 for 45min. No significant lateral inhomogeneities wereobserved in the measured area of 100 μm × 100 μm.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

7191

Equivalent measurements were performed on single crystals ofthe same composition (Li6La3ZrTaO12) as shown in Figure 4.

Also here, a homogeneous lateral 18O concentration was found asshown in the inset. Again, the measured data could be describedwell by using the analytical solution of Fick’s diffusion equationfor fitting. The calculated parameters for the measurementshown in Figure 4 are k* = 6.1 × 10−10 cm s−1 for tracer surfaceexchange and D* = 8.2 × 10−12 cm2 s−1 for tracer diffusion. Bothvalues are considerably higher than those for the polycrystal. Onepossible explanation could be grain boundaries. However, grainsizes in the polycrystal were∼10 μm, so pronounced effects fromblocking grain boundaries cannot be expected in the 1.2 μmdeepprofile. A more convincing explanation would be a difference inthe concentration of oxygen vacancies between the two samples.Even though they have nominally the same cation composition,their preparation route and the temperatures and atmospheresinvolved are different. Depending on the actual vacancyconcentration, such different preparation parameters can easilylead to a relative difference of ∼1 order of magnitude in theoxygen vacancy concentration of the polycrystal and singlecrystal, which is necessary to explain the calculated parameters.Al-Substituted LLZO. All three investigated Al-doped LLZO

compositions (Li7−3xAlxLa3Zr2O12, where x = 0.15, 0.20, or 0.30)showed similar results in isotope exchange profiling, which werehowever clearly different from those of the Ta-doped samples.For all Al−LLZO compositions, lateral inhomogeneities wereobserved as exemplar i ly shown in Figure 5 forLi6.4Al0.2La3Zr2O12. The depth profiles shown in Figure 5 werereconstructed from two different areas of the total measured areaof 100 μm × 100 μm as shown in the inset. The “fast” area (blacksquares) showed a significantly larger amount of 18Oincorporated over a depth of several micrometers, while theisotope depth profile of the slow area (blue circles) decayedmuch faster. During the analysis, it became clear that neither areacould be well represented by the used diffusion model (see thered fit curves). This is most probably caused by the diffusion notbeing homogeneous in those areas. Reasons can bemanifold, e.g.,grain boundaries, dislocations, secondary phases, etc. The natureof the inhomogeneous diffusion is detrimental for an analysis ofthe diffusion parameters, which makes an exact analysis

impossible without further knowledge. A detailed investigationis beyond the scope of this study but will be addressed in follow-up investigations.To roughly compare the oxygen exchange parameters to those

of other samples, however, we still use the inappropriate fitsshown in Figure 5 to estimate at least the correct order ofmagnitude. Thus, values were extracted for the fast area: k*∼ 9×10−10 cm s−1, and D* ∼ 3 × 10−12 cm2 s−1. Values for the slowarea were as follows: k* ∼ 9 × 10−11 cm s−1, and D* ∼ 4 × 10−15

cm2 s−1. Interestingly, there is a 1 order of magnitude differencein the surface exchange coefficient and even a 3 order ofmagnitude difference in the diffusion coefficient betweendifferent areas on the same polycrystal. When comparing thesevalues to those of the Ta−LLZO single crystal, we find the fastarea to be on the same order of magnitude in both k* and D*.Assuming the same mobility for oxide ions, this would mean thatthe concentration of oxygen vacancies is similar to that of Ta−LLZO in the fast area and decreased by ∼3 orders of magnitudein the slow area of Al−LLZO. However, possible porosity in thefast areas (compare the SEM image in Figure 2b) could alsocause an apparent increase in the oxygen exchange parameters.

Ga-Substituted LLZO. For Ga−LLZO, 3D isotope profilessimilar to those of Al−LLZO were measured. Again, lateralinhomogeneities were observed as shown in Figure 6. However,here no large “fast” areas were detected, but a few small areas inthe range of ∼2−10 μm were found with more 18O incorporatedthan in the homogeneous matrix with a low 18O content.In Figure 6, two of the faster areas were investigated

individually (1 and 2 in the inset) together with a larger areawithout any “18O hot spots” (3 in the inset). As already observedfor the Ta−LLZO polycrystals, a decrease in the 18Oconcentration was found at the surface and ignored for thepurpose of fitting. These surface effects were even morepronounced for Ga-doped LLZO polycrystals. One reason forthis could be the considerably higher 18O concentration close tothe surface, so that re-equilibration in ambient oxygen leads to astronger effect. In particular, for areas 1 and 2, very high 18O

Figure 4. Isotope exchange depth profile of a Li6La3ZrTaO12 singlecrystal. The isotope exchange was performed at 350 °C in 200mbar 18O2for 160 min. No significant lateral inhomogeneities were observed in themeasured area of 100 μm × 100 μm.

Figure 5. Isotope exchange depth profile of two different grains of aLi6.4Al0.2La3Zr2O12 polycrystal found in a total measured area of 100 μm× 100 μm as shown in the inset. The isotope exchange was performed at350 °C in 200 mbar 18O2 for 335 min. A clear difference in isotopeexchange and diffusion can be observed in different grains. The fit curvesof the isotropic diffusion model cannot reproduce the profiles and giveonly a very rough estimate of the oxygen exchange and diffusionparameters.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

7192

concentrations of up to ∼0.6 were found close to the surface,much higher concentrations than with the other compositions.However, also, the decay of the 18O profiles was fast. Convertedinto values for highest active area 1, the tracer surface exchangerate is very high (k* = 1.8 × 10−8 cm s−1) and the diffusioncoefficient is rather low (D* = 1.4× 10−13 cm2 s−1). This could becaused by possible secondary phases at the hot spot. Such phaseswould, however, have to be located only at the surface as no suchphases could be detected, even by GI-XRD. Interestingly, thediffusion coefficient is similar over the whole sample, which is afurther indication that the bulk LLZO is homogeneous and onlytiny changes at the surface are present. Further investigations arenecessary to understand the connection among the surface ofLLZO, its changes due to interaction with ambient air, and theeffect on isotope experiments. On the basis of existing SIMS data,we could not identify changed surface compositions or cationinhomogeneities (negative ions were measured).Parameters of Oxygen Exchange and Oxygen Diffu-

sion.A summary of the values for oxygen exchange found for thedifferent LLZO is given in Table 1. First, the parameters of theshown fit curves in Figures 3−6 are presented, and underneaththe average values from all measured 100 μm × 100 μm areas(three to five each) of the six different samples are shown in thetable.Focusing first on average surface exchange parameter k*, we

find that the Ta−LLZO single crystal and the Ga−LLZOpolycrystal surfaces are ∼1 order of magnitude more active foroxygen exchange than those of other LLZO compositions do.More specifically, the Ga−LLZO “hot spots” as also shown inFigure 6 exhibited the fastest oxygen exchange with a k* of >10−8

cm s−1. Because of the lack of electronic charge carriers in LLZO,we expect a dominating water-catalyzed oxygen incorporationmechanism as described in ref 40. Therefore, a direct comparisonto oxygen exchange coefficients of mixed electronic ionicconducting oxides is not really meaningful. For this study, therather fast surface exchange parameters (k*) found act mainly asa necessary requirement to incorporate enough 18O to be able tomeasure depth profiles. The thin layers of hydroxide or carbonate

species at the surfaces of different LLZO compositions willstrongly affect the surface exchange kinetics, which furtherhampers an unambiguous discussion of the k* values. Wetherefore put the emphasis on the tracer diffusion coefficient,which indeed characterizes bulk LLZO of the desiredcomposition.The values of the tracer diffusion coefficient D* are

surprisingly high for all measured samples. Values in the rangeof ∼10−14 cm2 s−1 to almost 10−11 cm2 s−1 were found. Thehighest values were found for the Ta−LLZO samples and hereespecially for the single crystal. The diffusion coefficients thereare already close to those of materials optimized for their oxygen-ion conductivity such as 9.5 mol % yttria-stabilized zirconia(YSZ) for which D* = 1.4 × 10−11 cm2 s−1 at 350 °C in air. Thispoints toward a significant or even very large amount of oxygenvacancies present in the investigated LLZO samples.

Oxygen Vacancy Concentration and Its Impact onDefect Chemistry. An exact quantification of the vacancyconcentration based on only the tracer diffusion coefficient is notpossible, but an estimation can be attempted. First, the measuredtracer diffusion coefficient (D*) needs to be converted into theoxygen self-diffusion coefficient (DO). D* differs from DO byHaven ratio H, typically a factor slightly smaller than 1.41 DOdepends (for diffusion via vacancies) on two factors, the diffusioncoefficient of the oxygen vacancies (DV) and the vacancyconcentration (cV). KnowingDO, with the diffusivity of vacancies,one can calculate the vacancy concentration or vice versa.However, for our case of LLZO, neither the diffusion coefficientof oxygen vacancies nor their concentration is known so far.While the concentration of oxygen vacancies can vary over manyorders of magnitude in different oxides, the vacancy diffusioncoefficients are typically more similar. At least a very roughestimation can therefore be attempted by comparison with otheroxides. Data forDV are available for some well-investigated oxidematerials, for example, in slightly doped strontium titanate(SrTiO3). Using De Souza’s equation interpolated fromnumerous studies in ref 42, we obtain a DV value of 2.7 × 10−8

cm2 s−1 at 350 °C. Assuming further a Haven ratio H of 0.7, we

Figure 6. Isotope exchange depth profile of three different areas of aLi6.4Ga0.2La3Zr2O12 polycrystal from a total measured area of 100 μm ×100 μm as shown in the inset. The isotope exchange was performed at350 °C in 200 mbar 18O2 for 45 min. A clear difference in isotopeexchange and diffusion can be observed in the different regions. The fitcurves of the isotropic diffusion model are used to estimate the oxygenexchange and diffusion parameters in the respective regions.

Table 1. Oxygen Exchange Parameters Extracted from IsotopeExchange at 350 °C and Depth Profilinga

k* (cm s−1) D* (cm2 s−1)

Shown Fit CurvesFigure 2, Ta-poly “1” 5.2 × 10−11 1.7 × 10−12

Figure 2, Ta-poly “2” 2.1 × 10−11 1.2 × 10−12

Figure 3, Ta-single 6.1 × 10−10 8.2 × 10−12

Figure 4, Al-poly “fast” ∼9 × 10−10 ∼3 × 10−12

Figure 4, Al-poly “slow” ∼9 × 10−11 ∼4 × 10−15

Figure 5, Ga-poly “1” 1.8 × 10−8 1.4 × 10−13

Figure 5, Ga-poly “2” 3.3 × 10−9 4.3 × 10−14

Figure 5, Ga-poly “3” 5.6 × 10−10 9.4 × 10−14

Average Values of All Measurements on Different LLZO CompositionsLi6La3ZrTaO12 poly 3.0 × 10−11 1.5 × 10−12

Li6La3ZrTaO12 single 5.6 × 10−10 7.3 × 10−12

Li6.55Al0.15La3Zr2O12 poly 3.7 × 10−11 1.2 × 10−14

Li6.4Al0.2La3Zr2O12 poly 2.7 × 10−11 9.7 × 10−15

Li6.1Al0.3La3Zr2O12 poly 6.3 × 10−11 4.3 × 10−14

Li6.4Ga0.2La3Zr2O12 poly 8.1 × 10−10 9.6 × 10−14

YSZ (100) Single Crystal at 350 °C9.5 mol % Y2O3:ZrO2 1.4 × 10−11

aSingle and poly refer to single-crystalline and polycrystalline samples,respectively.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

7193

can estimate the vacancy concentration according to eq 1, with cObeing the oxygen site concentration.

= *cc

DHD

V

O V (1)

For the measured D* values of LLZO in Table 1, we obtainrelative vacancy concentrations (cV/cO) between ∼2 × 10−7 and∼4 × 10−4. For the highest values measured on the single crystal,this would mean an oxygen nonstoichiometry of ∼5 × 10−3 perunit cell as a first rough estimate.Another comparison can be made for the highly doped

fluorite-type oxygen-ion conductor yttria-stabilized zirconia. Theoxygen ionic conductivity of 9.5 mol % single-crystalline YSZ at350 °C in air (σ) was measured as 1.4 × 10−5 S cm−1. The tracerdiffusion coefficient can be calculated to be 1.4 × 10−11 cm2 s−1

via the Nernst−Einstein relation43,44 with H = 0.65 for YSZ.From the known oxygen vacancy concentration of 4.3% ofoxygen sites in YSZ, we can now as a rough estimate assume thesame oxygen-ion mobility for LLZO and YSZ. For the highestmeasured diffusion coefficient (D* = 8.2× 10−12 cm2 s−1) in Ta−LLZO single crystals, this estimation would suggest a vacancyconcentration of even 2.5% of the oxygen sites in LLZO,approximately half of that in YSZ.We can conclude that an exact quantification is not possible at

this point of the investigations, but vacancy concentrations onthe parts per million scale in the areas with the lowest measureddiffusion coefficients up to the parts per thousand range or eventhe low percent range in the fastest diffusion areas can beexpected. At these concentrations, oxygen vacancies become veryimportant to understanding the entire defect chemistry of LLZO.For typical operation conditions of a Li battery close to roomtemperature, direct electrical contributions of oxygen vacanciescan be neglected. However, even though oxygen vacancies arenot mobile then, they still act as a donor and thus affect otherdefect concentrations. In particular, oxygen vacancies willdecrease the Li stoichiometry. Assuming doubly positive oxygenvacancies (with no trapped electrons), every oxygen vacancyneeds two Li vacancies for compensation; e.g., 1% oxygenvacancies per oxide site lead to a decrease of 0.24 in the Listoichiometry per LLZO formula unit Li7La3Zr2O12. From theexistence of oxygen vacancies in all LLZO compositions, weconclude that considering oxygen vacancies, the Li content inLLZO is typically lower than nominally reported values in theliterature.Li-Ion Conductivity. The technically most relevant property

of LLZO garnets is their high Li-ion conductivity. The questionof whether and how oxygen vacancies are relevant here remains.For this study, the Li-ion conductivities of the different LLZOcompositions were measured at room temperature directlybefore the isotope experiment at 350 °C. As all samples weresintered at >1200 °C and thermogravimetric data show nochange in mass, we assume all samples remain unchanged duringthe isotope exchange experiment at 350 °C.Several compositions showed high Li-ion bulk conductivities

at room temperature in air as shown in Table 2. Values of σ of upto 6 × 10−4 S cm−1 (for Ga−LLZO) were obtained. A complexpicture arises when trying to correlate for the different LLZOcompositions, the Li-ion conductivity, and the D* values, whichare in our approximation directly proportional to the oxygenvacancy concentrations as shown in Figure 7. For the Ta-dopedsingle crystal and polycrystal, we find the higher conductivity forthe polycrystal but a higher vacancy concentration for the single

crystal. On the other hand, for the Al− and Ga−LLZO samples,we find the four investigated compositions following theopposite trend: the higher the conductivity, the higher also theoxygen vacancy concentration. Interestingly, the vacancyconcentration is considerably higher for Ta−LLZO than forAl− and Ga−LLZO, even though their Li-ion conductivity issimilar. The different cation site for substitution (the Zr4+ site forTa5+ and the Li+ site for Al3+ and Ga3+) might have a large impacton oxygen vacancy formation.We can conclude that the impact of oxygen vacancies on Li-ion

conductivity is rather complex. This becomes clear whenconsidering the different influences oxygen vacancies can haveon the Li-ion conductivity of LLZO. (i) By acting as a donor,oxygen vacancies reduce the Li stoichiometry. (ii) By elasticallydeforming the LLZO lattice, oxygen vacancies can impact phaseformation and stabilization. (iii) In a closely related manner, anddependent on the exact location of the oxygen vacancies in thelattice, the elastic deformations can also affect the migrationbarriers and conduction paths of Li+ in LLZO. Consideringpoints (ii) and (iii), it is not surprising to find different relationsbetween oxygen vacancy concentrations and Li conductivities forsubstitution either with Ta5+ on the Zr4+ site or with Al3+ or Ga3+

on the Li+ sites of LLZO or also for using differentconcentrations. Further studies of the topics mentioned aboveare necessary, namely, the exact effect of Li stoichiometry on theconductivity and the elastic effects of changes in the oxygen andlithium stoichiometry and their impact on phase formation andon Li migration barriers.

Comparison with Other Garnet Oxides. The high oxygendiffusion coefficients measured in this study induce the questionif this is also observed in other garnet oxides. Experimentalstudies of oxide diffusion and/or conductivity for the garnets

Table 2. Li-Ion Conductivity Data of Different LLZOCompositions Measured at Room Temperature in Aira

σ (S cm−1)

Li6La3ZrTaO12 polycrystal 4.26 × 10−4

Li6La3ZrTaO12 single crystal 1.84 × 10−4

Li6.55Al0.15La3Zr2O12 polycrystal 6.05 × 10−5

Li6.4Al0.2La3Zr2O12 polycrystal 2.47 × 10−5

Li6.1Al0.3La3Zr2O12 polycrystal 1.79 × 10−4

Li6.4Ga0.2La3Zr2O12 polycrystal 6.07 × 10−4

aValues are from the LLZO samples measured before the isotopeexchange experiment.

Figure 7. Double-logarithmic plot of tracer diffusion coefficient D* andLi-ion conductivity of the six investigated sample types. Li+ sitesubstituents (Al3+ and Ga3+) and the Zr4+ site substituent (Ta5+) showdifferent trends.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

7194

yttrium aluminum garnet (Y3Al5O12, YAG),45−48 yttrium iron

garnet (Y3Fe5O12, YIG),49,50 and gadolinium iron garnet(Gd3Fe5O12, GIG)

50 in the temperature range of 800−1300°C have been published. These garnets all have in common arather poor oxygen diffusion coefficient of oxygen ionicconductivity. Because of the high activation energies typicallyin the range of 3−4 eV, D* values of ∼10−30 cm2 s−1 or evenlower result, when extrapolating to 350 °C, at which this studywas conducted.45,46,48,49 This is more than 15 orders ofmagnitude lower than the values found for LLZO in this study.Even though a computational study has found energy barriers ofonly 2.6 eV for YAG,51 the difference after extrapolating to 350°C would still remain extremely large. The reasons for this largedifference between LLZO and other garnets can be caused by ahigher oxygen vacancy concentration or a higher mobility, or aswe suspect by a combination of both.Some indication of the reasons for these enormous differences

can be found in the garnet structure. It is appropriate to use theformula X3Y2Z3 for garnets with the structure element X beingthe larger cation, Y being the smaller cation, and Z being a verylarge anion; e.g., for YAG, X = Y3+, Y = Al3+, and Z = AlO4

5−. ForLi7La3Zr2O12 in this nomenclature, X = La3+ and Y = Zr4+, whichis quite straightforward, whereas Z as the very large anion nowconsists of a structure already defective in the cations, nominallyLi7/3O4

17/3−. Here in contrast to the other garnets discussed, notonly MO4

x− tetraeders exist, but some Li+ ions also have tooccupy some of several additional other sites and their mobility isextremely high. We expect that this flexibility in the cationsstrongly increases the oxygen diffusion coefficient compared toother garnets. This increase might be realized (i) by reducing theformation enthalpy of oxygen vacancies by the possibility ofrearranging Li+ cations around the defect and (ii) by an increasedmobility of the O2− ions caused by a lower energy barrier of anindividual hop into an oxygen vacancy due to the same possibilityof fast rearrangement of Li+ ions. A recent computational studysuggests that Li ions in the vicinity of an oxygen vacancy preferthe octahedral (>2.2 Å distance to a vacancy) over the tetrahedralsites (<2.0 Å distance to a vacancy).31 This would be one possiblerealization of the discussed Li+ rearrangement. Furtherexperimental and computational studies are necessary to testthis assumption and to assess the individual contributions of thesuggested processes.

■ CONCLUSIONSThe existence of oxygen vacancies in fast Li-ion conductinggarnets was proven by using isotope exchange 3D imaging.Oxygen vacancies were found in all investigated garnet samples,which were Al-, Ga-, and Ta-doped LLZO polycrystals or singlecrystals. Isotope profiles were quantified by using twocoefficients, k* characterizing the surface activity and D*characterizing bulk diffusion. Both surface exchange anddiffusion were unexpectedly fast. While Ta-doped LLZO singlecrystals and polycrystals showed a laterally homogeneous 18Oconcentration, for Al-doped and Ga-doped LLZO active “hotspots” or fast and slow grains were observed. To learn moreabout their impact, quantification of the oxygen vacancies basedon diffusion data was attempted and concentrations up to theparts per thousand or even the low parts per hundred range wereestimated. Besides the impact on defect chemistry, which is adecrease in the Li stoichiometry, we also expect oxygen vacanciesto have an influence on both phase stabilization (cubic vstetragonal) and Li mobility due to lattice-elastic effects. Acomplicated relation between oxygen vacancy concentrations

and Li-ion conductivity was observed. Different effects of oxygenvacancies based on the cation substituent, on the substituent site,and on the concentration can be expected because of theirpossible impact on both the defect chemistry and the elasticparameters of LLZO.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: [email protected].

ORCIDMarkus Kubicek: 0000-0001-6623-9805NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

The authors gratefully acknowledge The Austrian ResearchPromotion Agency (FFG), SoLiK project, for financial support.D.R. acknowledges support from the Austrian Science Fund (P30660-N36).

■ REFERENCES(1) Murugan, R.; Thangadurai, V.; Weppner, W. Fast Lithium IonConduction in Garnet-Type Li7La3Zr2O12. Angew. Chem., Int. Ed. 2007,46, 7778−7781.(2) Awaka, J.; Kijima, N.; Hayakawa, H.; Akimoto, J. Synthesis andstructure analysis of tetragonal Li7La3Zr2O12 with the garnet-related typestructure. J. Solid State Chem. 2009, 182, 2046−2052.(3) Awaka, J.; Takashima, A.; Hayakawa, H.; Kijima, N.; Idemoto, Y.;Akimoto, J. Single Crystal Synthesis of Cubic Garnet Related-TypeLi7La3Zr2O12 by a Self-Flux Method. Key Eng. Mater. 2011, 485, 99−102.(4) Murugan, R.; Thangadurai, V.; Weppner, W. Fast Lithium IonConduction in Garnet-Type Li7La3Zr2O12; Wiley-VCH: New York, 2007;Vol. 46, pp 7778−7781.(5) Cussen, E. J. Structure and ionic conductivity in lithium garnets. J.Mater. Chem. 2010, 20, 5167−5173.(6) Thangadurai, V.; Narayanan, S.; Pinzaru, D. Garnet-type solid-statefast Li ion conductors for Li batteries: critical review. Chem. Soc. Rev.2014, 43, 4714−4727.(7) Bachman, J. C.; Muy, S.; Grimaud, A.; Chang, H.-H.; Pour, N.; Lux,S. F.; Paschos, O.; Maglia, F.; Lupart, S.; Lamp, P.; Giordano, L.; Shao-Horn, Y. Inorganic Solid-State Electrolytes for Lithium Batteries:Mechanisms and Properties Governing Ion Conduction. Chem. Rev.2016, 116, 140−162.(8) Buschmann, H.; Dolle, J.; Berendts, S.; Kuhn, A.; Bottke, P.;Wilkening, M.; Heitjans, P.; Senyshyn, A.; Ehrenberg, H.; Lotnyk, A.;Duppel, V.; Kienle, L.; Janek, J. Structure and dynamics of the fastlithium ion conductor ″Li7La3Zr2O12″. Phys. Chem. Chem. Phys. 2011,13, 19378−19392.(9) Ahn, C.-W.; Choi, J.-J.; Ryu, J.; Hahn, B.-D.; Kim, J.-W.; Yoon, W.-H.; Choi, J.-H.; Lee, J.-S.; Park, D.-S. Electrochemical properties ofLi7La3Zr2O12-based solid state battery. J. Power Sources 2014, 272, 554−558.(10) Bernuy-Lopez, C.; Manalastas, W.; Lopez del Amo, J. M.;Aguadero, A.; Aguesse, F.; Kilner, J. A. Atmosphere ControlledProcessing of Ga-Substituted Garnets for High Li-Ion ConductivityCeramics. Chem. Mater. 2014, 26, 3610−3617.(11) Rangasamy, E.; Wolfenstine, J.; Sakamoto, J. The role of Al and Liconcentration on the formation of cubic garnet solid electrolyte ofnominal composition Li7La3Zr2O12. Solid State Ionics 2012, 206, 28−32.(12) Ohta, S.; Kobayashi, T.; Asaoka, T. High lithium ionicconductivity in the garnet-type oxide Li7−X La3(Zr2−X, NbX)O12 (X =0−2). J. Power Sources 2011, 196, 3342−3345.(13) Mukhopadhyay, S.; Thompson, T.; Sakamoto, J.; Huq, A.;Wolfenstine, J.; Allen, J. L.; Bernstein, N.; Stewart, D. A.; Johannes, M.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

7195

D. Structure and Stoichiometry in Supervalent Doped Li7La3Zr2O12.Chem. Mater. 2015, 27, 3658−3665.(14) Li, Y.; Wang, C.-A.; Xie, H.; Cheng, J.; Goodenough, J. B. Highlithium ion conduction in garnet-type Li6La3ZrTaO12. Electrochem.Commun. 2011, 13, 1289−1292.(15) Buschmann, H.; Berendts, S.; Mogwitz, B.; Janek, J. Lithiummetalelectrode kinetics and ionic conductivity of the solid lithium ionconductors “Li7La3Zr2O12” and Li7−xLa3Zr2−xTaxO12 with garnet-typestructure. J. Power Sources 2012, 206, 236−244.(16) Cussen, E. J.; Yip, T. W. A neutron diffraction study of the d 0 andd 10 lithium garnets Li 3 Nd 3 W 2 O 12 and Li 5 La 3 Sb 2 O 12. J. SolidState Chem. 2007, 180, 1832−1839.(17) Deviannapoorani, C.; Dhivya, L.; Ramakumar, S.; Murugan, R.Lithium ion transport properties of high conductive telluriumsubstituted Li7La3Zr2O12 cubic lithium garnets. J. Power Sources 2013,240, 18−25.(18) Thompson, T.; Sharafi, A.; Johannes, M. D.; Huq, A.; Allen, J. L.;Wolfenstine, J.; Sakamoto, J. A tale of two sites: on defining the carrierconcentration in garnet-based ionic conductors for advanced Libatteries. Adv. Energy Mater. 2015, 5, 1500096.(19) Bottke, P.; Rettenwander, D.; Schmidt, W.; Amthauer, G.;Wilkening, M. Ion dynamics in solid electrolytes: NMR reveals theelementary steps of Li+ hopping in the garnet Li6.5La3Zr1. 75Mo0.25O12.Chem. Mater. 2015, 27, 6571−6582.(20) Wolfenstine, J.; Sakamoto, J.; Allen, J. Electron microscopycharacterization of hot-pressed Al substituted Li7La3Zr2O12. J. Mater. Sci.2012, 47, 4428−4431.(21) Allen, J. L.; Wolfenstine, J.; Rangasamy, E.; Sakamoto, J. Effect ofsubstitution (Ta, Al, Ga) on the conductivity of Li 7 La 3 Zr 2 O 12. J.Power Sources 2012, 206, 315−319.(22) Rettenwander, D.; Welzl, A.; Cheng, L.; Fleig, J.; Musso, M.;Suard, E.; Doeff, M. M.; Redhammer, G. J.; Amthauer, G. Synthesis,Crystal Chemistry, and Electrochemical Properties of Li7−2 x La3Zr2−xMo x O12 (x= 0.1−0.4): Stabilization of the Cubic Garnet Polymorph viaSubstitution of Zr4+ by Mo6+. Inorg. Chem. 2015, 54, 10440−10449.(23) Wagner, R.; Redhammer, G. J.; Rettenwander, D.; Tippelt, G.;Welzl, A.; Taibl, S.; Fleig, J.; Franz, A.; Lottermoser, W.; Amthauer, G.Fast Li-ion conducting garnet-related Li7−3xFexLa3Zr2O12 with un-common I4 3d structure. Chem. Mater. 2016, 28, 5943−5951.(24) Wagner, R.; Rettenwander, D.; Redhammer, G. n. J.; Tippelt, G.;Sabathi, G.; Musso, M. E.; Stanje, B.; Wilkening, M.; Suard, E.;Amthauer, G. Synthesis, Crystal Structure, and Stability of Cubic Li7−xLa3Zr2−x Bi x O12. Inorg. Chem. 2016, 55, 12211−12219.(25) Rettenwander, D.; Geiger, C. A.; Amthauer, G. Synthesis andcrystal chemistry of the fast Li-ion conductor Li7La3Zr2O12 doped withFe. Inorg. Chem. 2013, 52, 8005−8009.(26) Larraz, G.; Orera, A.; Sanz, J.; Sobrados, I.; Diez-Gomez, V.;Sanjuan, M. L. NMR study of Li distribution in Li7‑xHxLa3Zr2O12

garnets. J. Mater. Chem. A 2015, 3, 5683−5691.(27) Li, Y.; Cao, Y.; Guo, X. Influence of lithium oxide additives ondens i f i c a t i on and ion i c conduc t i v i t y o f ga rne t - t ypeLi6.75La3Zr1.75Ta0.25O12 solid electrolytes. Solid State Ionics 2013, 253,76−80.(28) Xie, H.; Li, Y.; Goodenough, J. B. Low-temperature synthesis of Li

7 La 3 Zr 2 O 12 with cubic garnet-type structure. Mater. Res. Bull. 2012,47, 1229−1232.(29) Nyman, M.; Alam, T. M.; McIntyre, S. K.; Bleier, G. C.; Ingersoll,D. Alternative Approach to Increasing Li Mobility in Li-La-Nb/TaGarnet Electrolytes. Chem. Mater. 2010, 22, 5401−5410.(30) Wolfenstine, J.; Allen, J. L.; Read, J.; Sakamoto, J. Chemicalstability of cubic Li7La3Zr2O12 with molten lithium at elevatedtemperature. J. Mater. Sci. 2013, 48, 5846−5851.(31) Jalem, R.; Morishita, Y.; Okajima, T.; Takeda, H.; Kondo, Y.;Nakayama, M.; Kasuga, T. Experimental and first-principles DFT studyon the electrochemical reactivity of garnet-type solid electrolytes withcarbon. J. Mater. Chem. A 2016, 4, 14371−14379.(32) Porz, L.; Swamy, T.; Sheldon, B. W.; Rettenwander, D.; Fromling,T.; Thaman, H. L.; Berendts, S.; Uecker, R.; Carter, W. C.; Chiang, Y.-M.

Mechanism of Lithium Metal Penetration through Inorganic SolidElectrolytes. Adv. Energy Mater. 2017, 1701003.(33) Wagner, R.; Redhammer, G. J.; Rettenwander, D.; Senyshyn, A.;Schmidt, W.; Wilkening, M.; Amthauer, G. Crystal Structure of Garnet-Related Li-Ion Conductor Li7−3xGaxLa3Zr2O12: Fast Li-Ion ConductionCaused by a Different Cubic Modification? Chem. Mater. 2016, 28,1861−1871.(34) Holzlechner, G.; Kubicek, M.; Hutter, H.; Fleig, J. A novel ToF-SIMS operation mode for improved accuracy and lateral resolution ofoxygen isotope measurements on oxides. J. Anal. At. Spectrom. 2013, 28,1080−1089.(35) Kubicek, M.; Holzlechner, G.; Opitz, A. K.; Larisegger, S.; Hutter,H.; Fleig, J. A novel ToF-SIMS operation mode for sub 100 nm lateralresolution: Application and performance. Appl. Surf. Sci. 2014, 289,407−416.(36) Li, Y.; Han, J.-T.; Wang, C.-A.; Xie, H.; Goodenough, J. B.Optimizing Li+ conductivity in a garnet framework. J. Mater. Chem.2012, 22, 15357−15361.(37) Crank, J. The mathematics of diffusion, 2nd ed.; Oxford UniversityPress: Oxford, U.K., 1979.(38) Yow, Z. F.; Oh, Y. L.; Gu, W.; Rao, R. P.; Adams, S. Effect of Li+/H+ exchange in water treated Ta-doped Li7La3Zr2O12. Solid State Ionics2016, 292, 122−129.(39) Wang, Y.; Lai, W. Phase transition in lithium garnet oxide ionicconductors Li7La3Zr2O12: The role of Ta substitution and H2O/CO2exposure. J. Power Sources 2015, 275, 612−620.(40) Nenning, A.; Navickas, E.; Hutter, H.; Fleig, J. Water-InducedDecoupling of Tracer and Electrochemical Oxygen Exchange Kineticson Mixed Conducting Electrodes. J. Phys. Chem. Lett. 2016, 7, 2826−2831.(41) Maier, J. Physical Chemistry of Ionic Materials. Ions and Electrons inSolids; John Wiley & Sons, Ltd.: Chichester, U.K., 2004.(42) De Souza, R. A.; Metlenko, V.; Park, D.;Weirich, T. E. Behavior ofoxygen vacancies in single-crystal SrTiO3: Equilibrium distribution anddiffusion kinetics. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85,174109.(43) Baumann, F. S.; Fleig, J.; Habermeier, H. U.; Maier, J. Impedancespectroscopic study on well-defined (La,Sr)(Co,Fe)O3‑delta modelelectrodes. Solid State Ionics 2006, 177, 1071−1081.(44) Kubicek, M.; Huber, T. M.; Welzl, A.; Penn, A.; Rupp, G. M.;Bernardi, J.; Stoger-Pollach, M.; Hutter, H.; Fleig, J. Electrochemicalproperties of La0.6Sr0.4CoO3‑δ thin films investigated by complementaryimpedance spectroscopy and isotope exchange depth profiling. SolidState Ionics 2014, 256, 38−44.(45) Haneda, H.; Sakaguchi, I.; Ohashi, N.; Saito, N.; Matsumoto, K.;Nakagawa, T.; Yanagitani, T.; Yagi, H. Evaluation of oxide ion diffusivityin YAG ceramics. Mater. Sci. Technol. 2009, 25, 1341−1345.(46) Sakaguchi, I.; Haneda, H.; Tanaka, J.; Yanagitani, T. Effect ofComposition on the Oxygen Tracer Diffusion in Transparent YttriumAluminium Garnet (YAG) Ceramics. J. Am. Ceram. Soc. 1996, 79,1627−1632.(47) Rotman, S. R.; Tuller, H. L. Ionic conduction in yttriumaluminum garnet. Solid State Ionics 1990, 40−41, 893−895.(48) Haneda, H.; Miyazawa, Y.; Shirasaki, S. Oxygen diffusion in singlecrystal yttrium aluminum garnet. J. Cryst. Growth 1984, 68, 581−588.(49) Kilner, J. A.; Steele, B. C. H.; Ilkov, L. Oxygen self-diffusionstudies using negative-ion secondary ion mass spectrometry (SIMS).Solid State Ionics 1984, 12, 89−97.(50) Kharton, V. V.; Shaula, A. L.; Naumovich, E. N.; Vyshatko, N. P.;Marozau, I. P.; Viskup, A. P.; Marques, F. M. B. Ionic Transport inGd3Fe5O12- and Y3Fe5O12-Based Garnets. J. Electrochem. Soc. 2003, 150,J33−J42.(51) Li, Z.; Liu, B.; Wang, J.; Sun, L.; Wang, J.; Zhou, Y. Mechanism ofIntrinsic Point Defects and Oxygen Diffusion in Yttrium AluminumGarnet: First-Principles Investigation. J. Am. Ceram. Soc. 2012, 95,3628−3633.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.7b01281Chem. Mater. 2017, 29, 7189−7196

7196