Deflecting dendrites by introducing compressive stress in Li 7 La 3 Zr 2 O 12 using ion implantation

Lithium dendrites belong to the key challenges of solid-state battery research. They are unavoidable due to the imperfect nature of surfaces containing defects of a critical size that can be filled by lithium until fracturing the solid electrolyte. The penetration of Li metal occurs along the propagating crack until a short circuit takes place. We hypothesise that ion implantation can be used to introduce stress states into Li 6.4 La 3 Zr 1.4 Ta 0.6 O 12 which enable an effective deflection and arrest of dendrites. The compositional and microstructural changes are studied via atom probe tomography, FIB-SEM with correlative TOF-SIMS, STEM and nano XRD indicating that Ag-ions can be implanted up to 1 µm deep and amorphization takes place down to 650-700 nm, in good agreement with kinetic Monte Carlo simulations. Based on nano XRD results pronounced stress states up to -700 MPa are generated in the near-surface region. Such a stress zone and the associated microstructural alterations exhibit the ability to not only deflect mechanically introduced cracks but also dendrites, as demonstrated by nano-indentation and galvanostatic cycling experiments with subsequent FIB-SEM observations. These results demonstrate ion implantation as a viable technique to design “dendrite-free” solid-state electrolytes for high-power and energy-dense solid-state batteries.


Introduction
[3][4] Through the usage of non-flammable solid electrolytes which are compatible with lithium metal, such as the oxide garnet Li7La3Zr2O12 (LLZO), safety concerns can be remedied and a resurgence of the lithium metal anode is possible. [5,6]e problem of dendritic growth remains though, still present in solid electrolytes, which are prone to being cracked by lithium dendrites, even when the material is mechanically stronger than lithium metal.2][13] Recently it was found that directed stresses in the material allow for guiding the growing crack in a certain direction.With increasing in plane compressive stress, the path of the growing crack aligns more and more with the stress direction, depending on the initial crack angle, as the path of least resistance from the mechanical perspective diverges from the electrochemical one.It was shown that if the stress is high enough, at least in the 200 MPa range for LLZO, cracks can be diverted to a 90 ° angle from the surface regardless of the initial crack direction.This can stop the dendrite from reaching the opposite electrode and shorting the cell. [14]In the work described, the stress was applied through mechanical bending of the solid electrolyte via a cantilever.Since bending a whole battery pack is rather impractical, the introduction of residual stress is necessary.
There are a few methods to introduce residual stresses, which is also known to increase the fracture resistance, depending on the type of stress introduced into a material.In metallurgy residual stresses are introduced during plastic deformation of the material, which is known as work hardening, if it occurs heterogeneously throughout the material.Work hardening can be used to significantly increase the hardness of the substrate by increasing the defect density in the material. [15]It can in principle be applied to any crystalline material, though it depends on the ductility.While metals can be deformed to introduce stresses, the same approach would shatter brittle ceramics, which LLZO counts as.This disallows the use of large scale plastic deformation like cold rolling and special care has to be taken https://doi.org/10.26434/chemrxiv-2023-311rvORCID: https://orcid.org/0000-0002-2074-941XContent not peer-reviewed by ChemRxiv.[18][19][20] Other possible methods include strengthening via grain boundaries, transformation hardening, second phase hardening and solid solution strengthening. [21][24] Previous research on different ceramics, such as MgO, Al2O3, and SiC has shown that an ion-implanted material can exhibit a higher fracture toughness. [24,25]e implantation of ions has a multitude of effects on the material properties, depending on the used parameters. Both of those parameters also determine the penetration depth than can be reached.This amorphized layer can even be mechanically weaker than the starting material. [24]The conductive properties of the material can also be influenced due to the implantation.Implanted ions can create additional states in the band structure changing the electronic conductivity, which made semiconductor applications possible and is the main reason for its use. [26]If the material is amorphized, pathways for ionic conduction can be diminished, which for battery applications is rather detrimental.Nevertheless, amorphous LLZO in thin film batteries has already been reported to reach reasonably high current densities (3.2 mA cm -²) without dendrite growth.There the thinness of the layer (70 nm) overcomes the diminished ionic conductivity of 4 orders of magnitude compared to bulk crystalline LLZO. [27]In the semiconductor industry annealing processes are typically used to regain the crystallinity and to benefit only from the increased electronic conduction without having to worry about an inhomogeneous distribution in the amorphous structure. [22]Parameters to achieve a similar reconstruction of the lattice, known as recrystallization, while keeping ion migration low, are not known yet in literature for solid electrolytes in general and LLZO specifically.
Contrary to the semiconductor industry, with thin finely polished single crystalline wafers, processing of polycrystalline material would be preferred for battery purposes to keep costs low and production volume high.This means grain boundaries and natural defects need to be kept in mind.A shallow implantation depth can run into orientation effects, leading to an inhomogeneous implantation profile similar to the implantation into nonplanar semiconductors. [28]A defect which extends past the implanted zone can also be filled with lithium and bypass the deflective properties from the stress field generated by the implanted ions.
In a recent publication [29] it was shown that ion implantation is possible for solid electrolytes and given a tentative stress value of -3-20 GPa in LLZO, depending on the dosage of implanted Xenon ions, https://doi.org/10.26434/chemrxiv-2023-311rvORCID: https://orcid.org/0000-0002-2074-941XContent not peer-reviewed by ChemRxiv.License: CC BY 4.0 retrieved from the examination of grazing incidence XRD data and molecular dynamics simulations.It has to be noted that the implantation depth in Yao et al is rather shallow, calculated at up to 60 nm and is below the common defect size in the sub µm size, as also mentioned by Yao et al.Furthermore, in situ plating of lithium in a TEM measurement showed fracture of "defect free" single crystalline LLZO from the local stresses generated by lithium ions, reaching up to GPa levels. [30]is study reveals that stress states in the near surface region up to 1 µm depth and up to -700 MPa can be introduced into Li6.4La3Zr1.4Ta0.6O12(LLZTO) by ion implantation.Such a high stress is introduced by an Ag-ion concentration lower than 0.003 at% in the LLZTO matrix and accompanied by the amorphization of the top 650-700 nm of LLZTO.Since the stressed zone is significantly deeper than the typical defect size mechanically introduced cracks have been shown to be deflected perpendicular to the indentation direction.Finally, we demonstrated that the same is true for dendrites with respect to the penetration direction, which opens a new avenue to designing "dendrite-free" solid-state electrolytes for high power and energy dense solid-state batteries.

SRIM simulation and structural profile
To identify the implantation conditions that allow the introduction of residual stresses in a depth up to 1 µm in LLZTO we calculated the Ag implantation and damage profiles by kinetic Monte Carlo simulations with SRIM software (Figure 1a).It is shown that for implantation within 1 µm a beam energy of 1.93 MeV has to be used.Both, the distribution of incorporated Ag ions as well as the associated damage caused by the irradiation can be estimated with a peak at 630 ± 30 nm and at 430 ± 25 nm for the amount of incorporated Ag ions and damage, respectively.With a dose of 1´10 14 Agions cm -2 an implanted concentration of 0.003 at% can be expected at the peak.There is the question of how accurate the simulation is when dealing with a polycrystalline ceramic, with possible chemical inhomogeneities.When predicting the ranges for single-crystalline silicon, it is rather precise with a slight overestimation of the range by 2-6 nm increasing with implantation energy. [31]For polycrystalline samples, in general and for LLZTO specifically such studies are, to the best of our knowledge, not available.
In order to visualize potential changes in crystallinity Virtual Dark Field imaging was applied to a Scanning Precession Electron Diffraction (SPED) measurement of an implanted polycrystalline LLZTO sample as shown in Figure 1b.It reveals grain size in the range of 3-6 µm and amorphization of LLZTO down to a depth of around 650±10 nm.The region below shows indications of a mixed crystallinity, with amorphous regions dotting crystalline parts, though it is difficult to distinguish between amorphization caused by electron beam damage and the one stemming from ion implantation.This is in accordance with the Monte Carlo simulation and allows the assumption that a significant amount of the incorporated Ag could be detectable in a range from 550 to 750 nm for further measurements.
Since structural degradation of LLZTO takes place even when using low beam current during SPED (see Supplementary Figure .1), atomic resolution imaging and locating the implanted Ag as well as the chemical analysis via EDS or EELS has been hindered (attempted EDS and EELS results are shown in Supplementary Figure 2a-e).
Therefore, atom probe tomography (APT) analyses were performed at the peak implantation depth, around 700 nm, as well as after the peak damage, depth of 1300 nm.The noise level of the mass spectra under the chosen analysis conditions does, however, not allow to detect the low Ag content (around 0.003 at% at the peak of implantation based on SRIM simulation).In Figure 2a the 3D reconstruction of one of the three datasets obtained from a depth of 700 nm is shown which is representative for all of them.The atoms are homogeneously distributed throughout the needle and no grain boundaries are visible, meaning only a single grain was investigated.The chemical composition, shown in Supplementary Table 1 is close to the nominal composition of Li6.4La3Zr1.4Ta0.6O12.Although, LLZTO is intended to be single phase (see XRD, SEM, and TEM analysis) the 3D reconstruction from a depth of 1300 nm on the other hand shows a clearly visible grain boundary, a grain with the same composition as in the previous dataset and a lithium deficient grain on top of it (Figure 2b).The lithium content in the upper grain is around 10.4 ± 2.0 at.%, see Supplementary Table 1, while still having a homogeneous distribution of atoms.A proxigram from the grain boundary, Supplementary Figure 3 shows the transition in lithium content.There is also a dataset only containing a lower lithium content.The Li-poor extra phase could be related to minor extra phases which have already been present in the sample or to a lithium loss in the high vacuum conditions during the APT analysis, which has been recently demonstrated. [32,33] ally, we used FIB-SEM with correlative TOF-SIMS to find evidence for the presence of Ag ions in the implanted zone.Figure 2b depicts the spots where the TOF SIMS measurements were taken and a slight peak, at 106.8 mass to charge ratio is visible in the first 0.9 µm diameter spot, which can be attributed to the 107 Ag implanted into this region.This peak vanishes in the other two spots confirming that the Ag is implanted into the desired depth.Attempts to, at least qualitatively, confirm the implantation profile using larger spots, shown in Supplementary Figure 4 dropped the signal below the threshold where it is distinguishable from noise.

Stress profile
To determine the localized strain of the LLZTO after implantation, scanning precession electron diffraction has been performed (Supplementary Figure 5).The diffraction patterns and the simulated patterns of LLZTO are generally in good agreement (see Supplementary Figure5b-c), though some parts of the sample are only partially crystalline resulting in slightly divergent diffraction spots.Hence, the resolution of the electron diffraction spots has not been high enough to get reliable strain values.Therefore, cross-sectional scanning nano XRD was used on a mechanically thinned sample to quantify the residual stress depth-profile created by the implanted ions.The 80 nm X-ray spot size allows the measurement of residual stresses and crystallinity with a high spatial resolution.The depth-dependent XRD patterns (Figure 1c) confirm a partial amorphization of the top 700 nm as indicated by the broad diffuse scattering peak centered around 18 degrees 2Θ and the inversely proportional decrease of reflections related to the LLZTO phase, as opposed to the underlying region below.From the orientation-dependent peak position a depth-resolved residual stress profile can be calculated.At a depth of 850 nm below the surface, just below the amorphized layer, a compressive stress peak up to approx.-700 MPa was evaluated, whereas throughout the rest of the layer the stress level does not exceed approx.-250 MPa.The slight waviness of the residual stress depth profile is stemming from the fact that the recorded XRD patterns lie somewhere between those corresponding to a singlecrystal and fully powder-like poly-crystal, resulting in residual stresses corresponding to the transition range between 1 st and 2 nd order stress, i.e., including some degree of grain-to-grain stress variation.
Based on this stress value and the methodology established in Fincher et al, it is possible to determine the flaw size of a dendrite that can be deflected with the present stress. [14]Equation 1 is adapted from Fincher et al to determine the flaw size a and assumes a KIC= 1 MPa m 1/2 for LLZO, same as in the reference, that our determined compressive stress at -700 MPa is 10 % higher than the critical stress σcrit to deflect all growing cracks to 90° giving a σcrit =636 MPa.

Proof of concept: deflection of cracks
To test if crack deflection occurs with the implantation of Ag ions, Nanoindentation was used to probe the implanted region on a larger single crystalline LLZTO sample.The resulting crack propagation was observed in a FIB SEM cut (see Figure 3a).Below the indentation imprint a crack runs parallel to the surface at a depth of 850 nm.This is as expected at the depth of the compressive stress zone, determined in the previous section.Notably absent are cracks connecting the indentation imprint and the fracture, which should be present if it is a radial fracture.Above the crack the amorphous region is again visible due to its differing contrast, additionally confirming the previous TEM and XRD results.To prove the hypothesis that the implementation of stressed zones by ion implantation can also deflect dendrites we applied electrical current well above reported critical current densities to a thinned sample with Li metal on both sites until the voltage reached device limitations, as a result of reduced contact area from lithium stripping.The corresponding current-voltage curve is shown in Figure 3b. Figure 3c shows the setup used during this kind of short circuit testing, where the sample is contacted with lithium in two places, carefully applied only on the side, to avoid circumventing the implanted layer and the zone of compressive stress.It is evident from Figure 3d that dendrites deflect after the implanted region, similar to mechanically induced cracks.The deflection zone is somewhat past the stress maximum, determined at 850 nm, and the crack begins to kink at around 700 nm and is fully deflected at 1.5 µm.Where the indented sample fractures at the zone of compressive stress, the growing dendrite has to snake its way through an increasing stress field, which can differ locally from the averaged values determined via XRD.Similarly, the bulk KIC value can differ from KIC value at lower length scales.With this it is entirely plausible for the crack to be fully deflected after the stress maximum, which is in agreement with Fincher et al, where it was shown that once the crack starts to kink and constant stress remains, it will eventually be turned by 90°. [14]

Impact on electrical properties
Since the damage caused by the ion beam and the change in the lattice from the differently sized ions implanted can have significant impact on the interfacial transport properties, electrochemical impedance spectroscopy (EIS) has been performed.In Figure .4 a Nyquist plot of a non-implanted LLZTO reference sample and an Ag-implanted sample can be seen.EIS of the Ag implanted LLZTO samples with non-blocking electrodes show the occurrence of a second semicircle as a result of the implantation procedure.This can be attributed to the now amorphized layer on both sides.The pristine hot-pressed sample was fitted with an equivalent circuit consisting of a resistor and a constant phase element (CPE) in parallel.A constant phase element was chosen instead of the capacitance as the semicircle is slightly depressed.The calculated ionic conductivity is 4.9´10 -4 S cm -1 .Under the assumption that the bulk conductivity does not change and that the amorphous layer is symmetric on both sides the implanted sample was fitted with an equivalent circuit of two parallel R-CPE elements in series.The resulting ionic conductivity for the amorphous layer, calculated with the known thickness of 650 nm, is 8.1´10 -8 S cm -1 .This is in good agreement with already reported values for the ionic conductivity of amorphous LLZTO, which are in a similar order of magnitude. [27]The change in lattice structure hindering the lithium-ion transport.The original crystal structure could be regained by an annealing step, as it is typically used for doped semiconductors.The necessary parameters, how much it will broaden the implantation profile and the impact on the residual stress is not known yet for LLZTO and beyond the scope of this study.

Conclusion
In this study we investigated how Ag ions implanted into LLZTO affect crack growth, specifically induced by growing dendrites.First Ag ions were implanted in LLZTO using conditions determined by kinetic Monte Carlo simulations to introduce ions into a depth of up to 1 µm, with a concentration peak at 750 nm.Determination of the actual concentration distribution and concentration maximum of the implanted ions was then attempted via APT and EELS.We found that the chosen dose of 1´10 14 Ag ions cm -2 resulted in an Ag concentration too low to quantify by these methods, so that a qualitative detection had to be achieved via TOF SIMS.Both spatially resolved cross-sectional nano XRD and SPED measurements revealed an amorphization of the region from the surface to a depth of 650 nm deep, partially also due to electron beam damage, which has been shown to be detrimental for the Li-ion conductivity.Moreover, from the corresponding X-ray diffractograms a compressive stress of up to 700 MPa with its peak at a depth of 850 nm has been determined.Thereafter, the impact of the introduced stressed zone on the crack propagation has been evaluated.Cracks were induced in the material, mechanically via nanoindentation, and they indeed formed preferentially at the zone of compressive stress at 850 nm and parallel to the surface.Finally, it has been shown that electrochemically induced cracks (dendrites) can also be deflected, with the crack beginning to kink at around 700 nm depth and fully deflecting slightly after the stress maximum at 1.5 µm.
In summary, our results suggest that introduction of compressively pre-stressed zones in a depth exceeding the critical crack size, in addition with subsequent annealing steps for re-crystallization of amorphous zones, could be a viable strategy to enable "dendrite-free" solid-state electrolytes to improve the rate-performance of solid-state batteries.

Sample preparation:
The Li6.5La3Zr1.5Ta0.5O12single crystal was pulled from a melt using the Czochralski technique.The thoroughly sintered initial material containing an excess of 20 Mol% Li2O was melted in an inductively heated iridium crucible in pure Ar atmosphere and the crystal was pulled at a rate of 0.4 mm/h on a rotating seed oriented along [100].The obtained crystal ingot was 15 mm in diameter and approximately 30 mm long, mainly transparent of slightly yellow color.The upper, first-grown part was covered by a thick film of opaque white material resulting from Li2O depletion caused by evaporation.The hot pressed polycrystals with the composition Li6.4Ta0.6La3Zr1.4O12were acquired from commercial sources (Toshima).The single and polycrystals were cut into 3x3x0.7 mm cuboids.

Phase purity characterization
For the commercial sample the phase purity was checked via X-ray diffraction.The measurement was carried out on a Bruker Da-Vinci X-ray diffractometer using a Cu Kα source with a scanning window of 10-60° and a step size of 0.013 °min -1 , see Supplementary Figure 6.

Ion implantation Procedure
The samples were stacked in a stainless-steel sample holder placed into a Pelletron implantation system type 3 SDH-2 NEC. 107Ag ions were implanted into both polished sides, flipping over after implantation of one side, with an energy of 1.930 MeV, a peak concentration of 1´10 14 107 Ag ions cm -2 (1´10 18 107 Ag ions cm -3 ) and a current density of 85 nA cm -².The implanted samples were then put in 0.1M HCl for around 20s followed by a rinse with Isopropanol and transferred to an Ar filled glovebox (O2 and H2O < 0.1 ppm) for storage and further experiments.
Before any experiment, this HCl treatment was repeated to remove any Li2CO3 layer formed during storage.

Electrochemical experiments:
The implanted sides of the cleaned samples were then coated with a molten LiSn alloy, containing 20 wt% of Sn.The impedance of the coated samples was measured with a Solartron Modulab XM at 25±1 °C with a perturbation voltage of 10 mV from 1 MHz to 10 Hz, 19 datapoints per decade of frequency after a 1 min OCV period.
For the creation of dendrites, a current of up to 1 mA cm -² was applied for 1 minute at 25±1 °C.All measurements were performed in an Ar filled glovebox with O2 and H2O levels below 0.1 ppm.

TEM/SPED experiments:
The Li6.4La3Zr1.4Ta0.6O12sample was extracted from an Ar glovebox (10 min air exposure) and a 0.5 µm layer of Ti was deposited on the Ag-implanted surface using an AJA electron beam evaporator.This https://doi.org/10.26434/chemrxiv-2023-311rvORCID: https://orcid.org/0000-0002-2074-941XContent not peer-reviewed by ChemRxiv.License: CC BY 4.0 protects against moisture from the air and carbon-containing deposition layers.A lamella was produced and thinned using the standard focused ion beam (FIB) liftout method, with a FEI Helios G4 UX dual-beam instrument.The final thinning was done using 2 kV Ga ions.The lamella was moved to a JEOL 2100F transmission electron microscope (TEM) within 15 min air exposure.
Scanning precession electron diffraction (SPED) was conducted using a precession angle of 0.8 degrees and a frequency of 100 Hz. 20 ms diffraction patterns were acquired with a MerlinEM direct electron detector.Virtual dark-field imaging was used to separate grains and crystalline/amorphous areas.

Atom Probe Tomography experiments:
The tips for APT analysis were prepared from a lamella that was lifted out from the Ti coated Specimen, flipped upside-down and welded to flat Si posts inside the FIB.In this way, the Ti surface coating acts as a buffer between the specimen and the carbon-containing welding material, and the apex of each tip could be formed at a controlled distance from the original surface.The tips were formed using annular Ga ion milling, finishing with 2 kV for the final sharpening.
The APT measurements were carried out on a LEAP 5000 XS from CAMECA equipped with a UV laser.
The specimen temperature was set t to 60 K. Laser pulses with an energy of 20 to 30 pJ were used at a frequency of 200 kHz to trigger field evaporation.The standing voltage was automatically adjusted by the software to maintain a detection rate of 0 0.3 % ions per pulse.
The commercial software IVAS from CAMECA was used for the reconstruction and analysis of the APT data.The reconstructed volumes were built using tip-profile reconstruction using secondary electron images obtained during specimen preparation by FIB.

ESRF experiment
For the Synchrotron measurements two implanted Li6.4La3Zr1.4Ta0.6O12polycrystalline sample crosssections were thinned down to a final thickness of around 80 µm by conventional mechanical polishing as mentioned above.The cross-sectional scanning X-ray nanodiffraction (CSnanoXRD) experiment was carried out at the nanofocus end station of the Microfocus beamline ID13 at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. [39]The sample was scanned by a monochromatic X-ray beam of 13 keV photon energy focused to approx.80 nm spot size by multilayer Laue lenses, applying a scanning step of 50 nm in the direction perpendicular to the implanted surface. [40]40 depth-profile scans spaced 1 µm apart were averaged in order to counteract the poor diffraction statistics that result from the combination of small X-ray beam, thin sample slice and comparatively large grain size of the probed material.Diffraction patterns were acquired in transmission on a Dectris Eiger X 4M two-dimensional (2D) X-ray detector placed 10.98 cm https://doi.org/10.26434/chemrxiv-2023-311rvORCID: https://orcid.org/0000-0002-2074-941XContent not peer-reviewed by ChemRxiv.License: CC BY 4.0 downstream of the sample with an exposure time of 50 ms.The exact detector position was calibrated by measuring a standard corundum reference powder, using the routines provided by the pyFAI software package. [41]2D diffraction patterns in the form of Debye-Scherrer rings were integrated azimuthally using the pyFAI software package and subsequently processed using custom python scripts for peak fitting with Pseudo-Voigt functions and strain/stress calculation.The evaluation of stress was based on the plausible assumption that there is no force acting perpendicularly on the free surface of the sample, which make the application of a sin² -based method possible.In such a case, the only necessary input parameters are the elastic properties of the probed material, i.e. in this case the X-ray elastic constant .In summary, the CSnanoXRD experiment and the applied evaluation technique corresponds to earlier studies of thin films samples at the same beamline, for more details see Ref. [42].

Nanoindentation
An implanted single crystalline Li6.4La3Zr1.4Ta0.6O12sample was placed in a FT-NMT03 nanomechanical testing system utilized within a Cross Beam 340 ZEISS SEM.Indents are controlled by displacement and were made using a Berkovich indenter tip with an indentation depth of 0.1 µm.
After attaching a few µm of Pt protection layer a 20x12x10 µm cross section was cut with FIB using a 20 nA beam current, reduced to 4 nA and 600 pA for final polishing on a cross-beam SEM from Zeiss (AURIGA® -CrossBeam workstation).The Image in Fig. 3a was taken with an acceleration voltage of 2kV and performed in backscattered electron mode.The equivalent circuits used to fit the arcs are shown in the top right corner.The second semicircle is attributed to the 650 nm amorphous layer which is detrimental to ionic transport as shown in previous literature. 26 https://doi.org/10.26434/chemrxiv-2023-311rvORCID: https://orcid.org/0000-0002-2074-941XContent not peer-reviewed by ChemRxiv.License: CC BY 4.0

12
which was calculated for the relevant X-ray reflection {31 2 2} from literature single-crystal elastic values using in a Hill grain interaction model, as MPa 1$

Figures
Figures

Figure 1 .
Figure 1.a) SRIM simulation of the damage and Ag ion distribution in Li6.4Ta0.6La3Zr1.4O12crystal irradiated with 1.93 MeV Ag ions.b) SPED composite dark field image of a hot pressed Li6.4Ta0.6La3Zr1.4O12polycrystal with Ag implanted into the top layer.An amorphous layer 650 nm deep is clearly visible.The material has also been shown to be particularly electron beam sensitive, being easily amorphized by the stationary electron beam between measurements.c) Compilation of cross-sectional nano X-ray diffraction patterns of the implanted polycrystalline LLZTO sample.The implanted region is shaded in orange.An amorphous region can be seen, starting from the surface down to a depth of 700 nm.Compressive in plane stress is shown in d) reaching a peak value of -700 MPa at 850 nm.

Figure 2 .
Figure 2. a) Atom probe tomography reconstruction of a needle prepared from a depth of 700 nm.The atoms are distributed homogeneously throughout the needle b) Reconstruction of a needle from a depth of 1300 nm.While there is still a homogeneous atom distribution within grains the lithium content in the upper part is lower.c) SEM image of the cross section of the implanted LLZTO sample.The 900 nm diameter spots are where TOF SIMS measurements were taken.d) TOF sims data from the three spots in b.There is a barely visible peak for 107 Ag at 106.8 mass to charge ratio for the first spot.This peak vanishes for the other spots which are below the implanted region.

Figure 3
Figure 3 (a) SEM image (Backscattered electrons) of a mechanically induced crack by nano indentation in Ag implanted LLZTO.The crack grows preferably parallel to the surface.(b) Voltage profiles during the short circuiting measurement, spliced together.Current was doubled after 10s and held until the voltage limit was reached, likely due to void formation decreasing the contact area.(c) Optical microsope image of the thinned hot pressed LLZTO during the short-circuit measurement.(d,e) SEM image of an electrochemically induced crack in Ag implanted LLZTO.Top view (d) and 51° angled view (e).The implanted region is shaded in orange, and the crack highlighted with a dashed line.After the implantation zone the crack diverts parallel to the surface.

Figure 4
Figure 4Nyquist plot of an Ag implanted (grey) and a non-implanted (orange) reference hot pressed polycrystal.The equivalent circuits used to fit the arcs are shown in the top right corner.The second semicircle is attributed to the 650 nm amorphous layer which is detrimental to ionic transport as shown in previous literature.26