Insights into the LiMn2O4 Cathode Stability in Aqueous Electrolytes

LiMn2O4 (LMO) cathodes present large stability when cycled in aqueous electrolytes, contrasting with their behavior in conventional organic electrolytes in lithium-ion batteries (LIBs). To elucidate the mechanisms underlying this distinctive behavior, we employ unconventional characterization techniques, including variable energy positron annihilation lifetime spectroscopy (VEPALS), tip-enhanced Raman spectroscopy (TERS), and macro-Raman spectroscopy (with tens of μm-size laser spot). These still rather unexplored techniques in the battery field provide complementary information across different length scales, revealing previously hidden features. VEPALS offers atomic-scale insights, uncovering cationic defects and subnanometer pores that tend to collapse with cycling. TERS, operating in the nanometric range at the surface, captured the presence of Mn3O4 and its dissolution with cycling, elucidating dynamic changes during operation. Additionally, TERS highlights the accumulation of SO42– at grain boundaries. Macro-Raman spectroscopy focuses on the micrometer scale, depicting small changes in the cathode’s long-range order, suggesting a slow but progressive loss of crystalline quality under operation. Integrating these techniques provides a comprehensive assessment of LMO cathode stability in aqueous electrolytes, offering multifaceted insights into phase and defect evolution that can help to rationalize the origin of such stability when compared with conventional organic electrolytes. Our findings advance the understanding of LMO behavior in aqueous environments and provide guidelines for its development for next-generation LIBs.


Introduction
The development of advanced cathode materials for lithium-ion batteries (LIBs) has been at the forefront of research aiming at improving the performance, cost and safety of energy storage systems.Since its first reports by J.C. Hunter and Goodenough's group back in the 80s, [1,2], there has been a continuous and growing interest from both academic and industrial perspectives in LiMn2O4 (LMO), which has led to a deep understanding of its structure, properties, advantages and challenges.[3] LMO belongs to the family of cubic spinel oxides (Fd3m space group), in which Li is tetrahedrally coordinated and Mn is octahedrally coordinated, which allows for 3D Li + diffusion channels that could potentially lead to fast-charging cathodes, as shown, for instance, in other spinel-type materials.[4] Despite its relatively high theoretical capacity (148 and 296 mAh•g -1 for LiMn2O4 in the regular and in the extended cycling range, respectively), there are plenty of reports showing concerns regarding the long-term stability when cycled in conventional organic electrolytes.In particular, Mn dissolution has been claimed to be the main drawback of this material, partially due to Jahn-Teller distortions and Mn disproportionation.[5] However, other effects play a role, since Mn dissolution also occurs in the fully charged state, [6,7] unexpected if disproportionation would be the only responsible of the Mn 2+ dissolution.Recent works have shown with great detail the degradation mechanism in LiMn2O4 in LIBs through advanced characterization techniques, as excellently summarized by Prof. Ju Li's team in their review.[3] Essentially, these recent studies have highlighted significant surface-related phenomena.These observations include Mn reduction during charging and Mn oxidation during discharging at the surface, oxygen loss linked to Mn reduction, and surface reconstruction upon charging.[8][9][10][11] This ultimately results in Mn dissolution and surface phase transformations, contributing to capacity decay and increased impedance.[12] In addition, advanced characterization using in situ techniques has revealed the formation of Mn3O4 at high charge voltages of 4.2V and Li2Mn2O4 at discharge voltages of 3.3V.These electrochemically induced phase transformations are partially irreversible and can lead to particle cracking during cycling, further exacerbating the reactive surface area.[13] Interestingly, despite the intricate challenges posed by surface-related phenomena and structural instability in LiMn2O4 when employed with conventional organic electrolytes, a remarkable shift occurs when this cathode material is cycled in aqueous electrolytes.In this alternative environment, LiMn2O4 exhibits a surprising level of stability, seemingly untouched by the issues that affect its performance in organic electrolytes, besides the well-known Mn dissolution.[14][15][16][17][18][19] This stark contrast in behavior underscores the need for a thorough investigation into the underlying mechanisms responsible for this stability divergence, setting the stage for our exploration of LMO's performance in aqueous environments using unconventional techniques, trying to ask a simple question: why is LMO more stable in aqueous electrolytes?Can we learn anything from the study of highstability operation in aqueous electrolyte that can be extrapolated to organic electrolytes?
In this work, we examine the structural and phase evolution of LMO films in aqueous electrolyte as a function of cycling by three novel characterization techniques that allow us to examine surface and bulk properties at very different length scales: Tip-Enhanced Raman Spectroscopy (TERS), macro-Raman Spectroscopy and Variable Energy Positron Annihilation Lifetime Spectroscopy (VEPALS).
TERS is a relatively new surface-sensitivity technique that overcomes the spatial resolution limitations of conventional Raman (~500 -1000 nm) reaching even single molecule detection.[20][21][22][23] While it is widely used to study organic materials, its application to complex oxides is not yet widespread.The use of TERS in batteries is still in its infancy.[24,25] Essentially, TERS combines the spectroscopic power of Raman Spectroscopy with the spatial resolution of a Scanning Probe Microscope.This is achieved by carefully focusing the laser at the apex of a tip coated with a thin metal able to generate surface plasmon resonances and lightning-rod effect to enhance the Raman signal.[26] With a completely different and complementary approach, macro-Raman Spectroscopy uses a very large laser spot size (in the range of 50 -100 µm in diameter) and ultra-low energy densities (tens of W•cm -2 vs. tens of kW•cm -2 in conventional Raman) to ensure no influence on the sample while being sensitive to phases located in a true representative area, with the averaged signal from 10 3 -10 4 grains under the measurement spot.
In VEPALS, the films are irradiated with positrons that recombine (annihilate) with electrons in the films at defined depths, creating gamma rays, which are subsequently detected.[27] The measurement of positrons lifetime provides a direct information on the type and size of negatively charged and neutral defects through their characteristic time before encountering an electron, whereas pores are characterized based on the lifetime of residing and bouncing in them a bound state of positron and electron, the so-called Positronium (Ps) [28,29].The concentration of positrons and Ps is evaluated through the relative intensities of the different lifetime components.In the context of battery research, VEPALS has found notable applications, particularly in studies involving layered compounds like LiCoO2.In such cases, positron lifetime has been proposed as a reliable indicator of the lithiation state of grain boundaries.[30] Furthermore, in investigations of compounds like LiNi1/3Mn1/3Co1/3O2, operando experiments have revealed a gradual increase in divacancy and vacancy agglomerate formation during the charging process, culminating in the transformation of these agglomerates into one-dimensional vacancy chains as the charge cycle nears completion.[31] These findings underscore the versatility and significance of VEPALS in understanding complex defects and phenomena within battery materials.These three complementary techniques, rarely explored within the battery field, offer a multifaceted perspective on the phase and defect evolution within LMO cathodes and their performance stability in aqueous electrolytes.Operating across a broad spectrum of length scales, they collectively reveal previously unseen features.Notably, VEPALS exposes the existence of cationic defects and sub-nanometer pores, providing insights at the atomic scale of the dynamic changes with cycling.TERS captures the presence of Mn3O4 at the surface and its subsequent dissolution during cycling with a resolution better than 20 nm.Simultaneously, TERS also highlights the accumulation and potential incorporation of SO4 2-ions, primarily at grain boundaries, while macro-Raman Spectroscopy focuses on the material's long-range order, depicting small changes at the micrometer scale.Our work contributes to a more profound understanding of the stability of LMO in aqueous electrolytes with the hope that this knowledge can be extrapolated to implement solutions toward longer stability in conventional organic electrolytes.

Results & discussion
Characterization of the as-deposited LMO films LMO thin films were fabricated by Large Area Pulsed Laser Deposition with our wellestablished method of compensating Li losses with a multilayer deposition of LiMn2O4 parent compound and Li2O acting as Li reservoir.[17,[32][33][34][35][36] When grown on top of Ptcoated Si-chips, this method produces relatively rough polycrystalline layers with high phase purity.For this experiment, a batch of 300 nm thick LMO films was produced (for details, please see the Experimental section).SEM shows a cross-section of the layers that are dense, and rough (as expected) with well-connected grains with sizes of 250-300 nm (Fig. 1a).In line with our previous results, [17,35] the XRD patterns of the layer evidence the polycrystalline nature of our films with small quantities of Mn3O4 impurities.Despite the difficulty of performing more sophisticated approaches (Rietveld refinement) in this data, one can see for instance, that the most intense line that can be unambiguously assigned to Mn3O4 (2θ = 28.91º,I ≈ 40%) is almost not visible in the diffractogram.In contrast, the (400) reflection of LMO (2θ = 43.87º,I ≈ 33%) is identified with large intensity.
We further characterized the films with VEPALS.The average positron lifetime versus positron implantation energy is quite homogeneous along the film thickness (Fig. 1c).The deconvolution of the spectra (see Ref. [35], experimental methods and Supporting Information, section I) showed four main types of defects: small vacancy-like defects and their clusters and two families of sub-nm pores with spherical sizes d3 ≈ 0.47 nm and d4 ≈ 0.75 nm (calculated based on [37]).Overall, both pore families and vacancy-related defects exhibit a homogeneous distribution across the thickness of the film.Comparing the relative intensities, one can see that the vacancy-related defects dominate the average lifetime of the positrons, indicating the larger presence of vacancies than pores.These kinds of vacancies have also been observed by PALS in LiCoO2 cathodes and have been ascribed to lithium vacancies and clusters of lithium vacancies, respectively.[27,30] However, we would rather be cautious over the assignment of the type of vacancy, see Supporting Information, section I for this discussion.Overall, these PALS measurements demonstrate the presence of cationic defects and pores in our films, which potentially play a role in suppressing and/or alleviating commonly observed Jahn−Teller distortions in LMO.[35,38]  Regarding Raman characterization of the films, existing differences between the spectra obtained with macro-Raman Spectroscopy and TERS are considered relevant.On one hand, the low excitation energy, long acquisition times and very large laser spot size of the macro-Raman Spectroscopy assures that the signal arises from tens of thousands of grains, rather than from tens of grains interacting with a conventional Raman laser spot, given the grain sizes of our films, in the range of 250-300 nm.This allows detecting secondary phases and impurities extensively present in the films.The macro-Raman Spectroscopy of the as-deposited state films included in Fig. 2a shows the presence of mainly LiMn2O4 and impurities of Mn3O4, [39,40] in agreement with the XRD results.In addition, the presence of off-stoichiometric LixMn2O4 phase cannot be discarded from the measured spectra, [39] suggesting the presence of vacancy defects in the bulk of the films.On the other hand, TERS measurements with higher surface sensitivity complement this information by assigning a large presence of Mn3O4 at the surface and sub-surface level.This is clearly shown in the average TERS spectra corresponding to a 500 x 500 nm 2 spectroscopy scan, Fig. 2b, where the main peak of Mn3O4 (located at ~660 cm -1 ) is now more intense than the LiMn2O4 peaks, in stark contrast with the macro-Raman Spectroscopy measurements.The topography acquired during the TERS measurement, Fig. 2c, shows several grains, with very distinguishable grain boundaries.The TERS map, Fig. 2d, has been built in such a way that the green color scale represents the LiMn2O4 phase, linked to the intensity of the peak located at 580 cm -1 , and the blue color represents the Mn3O4, linked to the intensity of the peak at 660 cm -1 .While the LiM2O4 phase is more homogeneously distributed, the Mn3O4 phase seems to be located more at the bottom part of the image.The full power of the technique is revealed by superimposing the topography and the spectroscopy maps, Fig. 2e.Here, one can see that the LiMn2O4 is homogeneously distributed across the map, dominating the grains.In contrast, Mn3O4 is clearly present at the grain boundary level (all across the image) and extensively in some particular grains.For more details about how these maps are built and the demonstration of the TERS effect on our setup, the reader is referred to the Supporting Information, Section III.
Combining the different techniques, one can conclude that the LMO films are composed mainly of the electrochemically active spinel phase and impurities of Mn3O4, which are located primarily at the surface of the films and, especially, at the grain boundary level.The films also show the presence of vacancy-like defects and their clusters (probably at grain boundaries) across the film and to a lesser extent, two families of sub-nm pores.

Characterization of the LMO films during cycling
The fabricated polycrystalline LMO films were cycled as previously detailed elsewhere showing the same electrochemical and structural properties as those previously reported by our group.[17,35,41] Upon cycling in a sulphate-based aqueous electrolyte, the films show the expected two voltage plateaus and capacities of ~126 mAh•g -1 after 300 cycles, with an estimated loss of 0.025% per cycle (Fig. 3a and inset).The electrochemical properties also evidence that the films are composed mainly of electrochemically active LiMn2O4 since its capacity approaches the practical capacity of LiMn2O4 that is commonly found in literature.[3] We proceed here with a comprehensive exploration of the Raman signature evolution within our LMO films throughout the cycling process.Specifically, we analyzed films at 100% SoC (i.e., λ-MnO2 phase) at different stages of cycling: After 1, 100 and 300 cycles, macro-Raman Spectroscopy allows us to scrutinize these changes.Fig. 3b illustrates a noteworthy trend: the primary peaks of the electrochemically active phases exhibit a relative intensity decrease concerning Mn3O4 inactive impurities.It is essential to note that our macroscopic Raman Spectroscopy approach, given its true average nature, is expected to detect all electrochemically active phases-namely, LiMn2O4, Li0.5Mn2O4, and λ-MnO2.This detection remains consistent with literature findings, as our capacities stay in the expected range of the practical capacity of the material.[3] This implies that a substantial portion of Li + ions resides within the lattice as LiMn2O4, even when the material is nominally at 100% State of Charge (SoC), as detected by our macro-Raman Spectroscopy measurements and other studies in the field.[42] The observed smearing of the Raman peaks associated with the electrochemically active phases can imply a gradual increase in film disorder and/or decrease of the crystalline quality.This phenomenon correlates with the rise in vacancy-like defects and their clusters, as identified through PALS in the next section.The enhanced disorder and reduced crystalline quality within electrochemically active phases during cycling may ultimately govern the long-term stability of these films in aqueous electrolytes, leading to a minimal capacity loss of only 0.025% per cycle.
In contrast, the Mn3O4 main mode appears to remain largely unaffected at the macroscopic level, emphasizing its inactive electrochemical role.However, a closer examination of the surface using Tip-Enhanced Raman Spectroscopy (TERS) mapping during cycling unveils two critical observations.Fig. 3c compares the average TERS spectra extracted from the corresponding maps.Firstly, there is a significant decrease in the Mn3O4 signal at the surface with cycling.This intriguing observation hints at the partial dissolution of these impurities, considering the presence of Mn 2+ within their structure.[3,43,44] Please note that we do not observe any obvious phase appearance with the dissolution of Mn3O4 seen at the surface.Secondly, the sulfate ions signal exhibits increased intensity and a blue shift, evolving from ~980 cm -1 after 1 cycle to ~1007 cm -1 after 300 cycles.This shift suggests a transition from absorbed SO4 2-ions to a scenario in which these adsorbates form stronger bonds with the film.

Figure 3: Raman Spectroscopy of the cycled films. a) Electrochemical cycling of the LMO films. The inset shows an example of charge-discharge profile (cycle #50) b) Macro-Raman Spectroscopy. c) Average TERS spectrum of the
TERS maps in e, h and k.For the film cycled 1 time, size 500 x 500 nm 2 : d) Topography, e) TERS map and f) overlay of both.For the film cycled 100 times, size 500 x 500 nm 2 : g) Topography, h) TERS map and i) overlay of both.For the film cycled 300 times, size 500 x 440 nm 2 : j) Topography, k) TERS map and l) overlay of both.All the measurements were done at full charge (SoC 100%).
Comparing the spatial distribution of Mn3O4 (blue) and sulfate signals (red) in films cycled 1x (Fig. 3d-f), 100x (Fig. 3g-i) and 300x (Fig. 3j-l) reveals a notable trend.In both cases, the proportion of maps dominated by Mn3O4 (blue) decreases with cycling (see Fig. 2e for as-deposited, Fig. 3f for 1, Fig. 3i for 100, and Fig. 3l for 300 cycles).By 300 cycles, the few visible blue spots are confined to certain grain boundaries.Moreover, both the 100x and 300x cycled maps exhibit the sulfate signal (red) predominantly at grain boundaries.To facilitate the identification, the reader is referred to the Supporting Information, Section III.
To complement our findings, VEPALS measurements were carried out to gain insights into defect evolution during cycling.Detailed descriptions of the deconvolution of the PALS signal into various components can be found in the Supporting Information, Section I.A critical observation emerges from the data analysis: our films' average positron lifetime (τaverage) decreases from the first cycle onward, stabilizing after 100 and 300 cycles.This decline in τaverage, directly linked to the free volume, can be attributed to the dynamic evolution of the electrolyte solution within grain boundaries and through pores within the film.Deconvolution of the different defect signals reveals that the most significant changes occur in τ4, representing a family of larger sub-nm pores.Its relative intensity, proportional to pore density, decreases from approximately 17% after 1 cycle to around 8% after 300 cycles.This decline could imply that a part of these characteristic pores might being filled, likely with sulfate ions, as suggested by our TERS measurements, or the pore size increase by coalescence to some extent.This shift in pore density redistributes the contribution of the other three components within the defects, which tend to become more significant in the signal.Specifically, the smaller vacancies (τ1) and their clusters (τ2) demonstrate an increase with cycling (Fig. 4b).This trend is mirrored in the large family of pores, τ4, which increases in size at a higher rate than vacancy clusters.In contrast, the smaller family of pores, τ3, appears unaffected by cycling.These changes, primarily the increases in τ1, τ2, and τ4 (d4), suggest the dissolution of Mn 2+ (natively present in the Mn3O4 structure) into the solution, leaving behind larger defects.[44] Nevertheless, dissolution rate in pores is expected to be higher than due to vacancies and their clusters, given the higher free volume, as generally probed by PALS.It also aligns with the observed increase in film disorder, consistent with the smearing out of the Raman bands noted by macro-Raman Spectroscopy and TERS.For completeness, it is noteworthy that the substantial reduction in pore density (τ4) is even more pronounced at the film's bottom portion, essentially vanishing progressively with cycling (see Supporting Information, section II).
Our study aligns with recent research that underscores the dominant role of fast grain boundary and surface diffusion, surpassing grain diffusion by a staggering four orders of magnitude.[45] Mürter et al. have convincingly argued that the wide range of reported diffusion coefficients in the literature likely stems from underestimating the contribution of grain boundaries.In our work, we provide experimental evidence of the complex chemistry of grain boundaries during cycling, a critical factor that demands consideration.Notably, Mürter et al. identified residual electrolyte peaks in their X-ray diffraction (XRD) pattern.We propose an intriguing possibility: these residuals may, in fact, become integral parts of the material's structure after cycling, predominantly within the grain boundaries.This notion suggests that grain boundaries can actively evolve and adapt, potentially absorbing ions from the electrolyte as part of their transformation.These insights emphasize the need to account for the active role of grain boundaries when designing protective coatings, especially in high-voltage cathodes.
On the other hand, importantly, grain boundaries within LiMn2O4 are known to carry a positive charge and establish a space charge layer around them.[45,46] This inherent property makes them susceptible to ion adsorption from the electrolyte during cycling, warranting careful consideration in protective coating design.Understanding the behavior of grain boundaries in high-voltage cathodes represents a pivotal step towards the development of more sustainable and stable electrode materials.
The surface dilution of Mn3O4 during cycling, primarily restricted to a few grain boundaries after 300 cycles, stands in stark contrast to its behavior in organic electrolytes.In organic electrolytes, Mn3O4 tends to form at the LiMn2O4 grain surface upon charging to 4.3V vs. Li/Li + , leading to partial irreversibility and cracking.[8,13] A recent work by Zhao et al. has shown that Mn3O4 nanowalls can be in-situ electrochemically oxidized in Li2SO4 electrolyte to form Li4Mn5O12, a transformation with potential applications in supercapacitors.[47] Our TERS and macro-Raman Spectroscopy measurements do not conclusively reveal the formation of Li4Mn5O12 or any other Mn-based polymorph upon cycling that could potentially correlate with the observed Mn3O4 dissolution.However, the stability in charge-discharge profiles suggests that this dissolution at the surface does not compromise the electrochemically active LMO electrode performance.It may even serve to protect the electrode from potential Mn 2+ dissolution.
Our findings regarding the presence of ionic species from the electrolyte anchored at the grain boundaries and surface not only contribute to unraveling the complexities of grain boundary chemistry but also align with recent efforts to incorporate lithium sulfate protective coatings in high-voltage cathodes, as documented in the literature.[48,49] These protective coatings can serve as a proactive shield against the evolving nature of grain boundaries, potentially mitigating the adverse effects of electrolyte interactions.Understanding the intricate interplay between grain boundaries and electrolytes is a pivotal step toward designing cathode materials that can perform well in the long term.Our work also holds broader implications for the stability of cathodes in aqueous electrolytes.For that, we argue that the presence of sub-nm pores and cationic vacancies can potentially help to mitigate the volume expansions in all-solid electrodes.These insights are not only pertinent to aqueous systems but also offer valuable lessons for enhancing the stability of cathodes in organic electrolytes and solid-state composites, where the behavior of grain boundaries and their interactions remain critical yet less explored.

Conclusions
Our research provides valuable structural and chemical insights across a spectrum of length scales for the defect evolution with cycling in LMO cathodes.Beginning at the sub-nanometer scale, we have identified vacancy-like defects and their clusters, with the latter emerging as the most prevalent defect within our films.These defects appear to grow in size with cycling, possibly due to the dissolution of Mn 2+ ions, introducing a degree of disorder into the films.Additionally, we have discerned the presence of two distinct families of sub-nanometer pores, notably, the larger pores (τ4) appear to reach a saturation point with cycling, hinting either at complete fill up of part of the pores in a gradual manner with electrolyte species or pore coalescence with cycling.
Zooming out to the nanoscale, our TERS measurements unveil that, unlike in organic electrolytes, Mn3O4 does not form at the surface during cycling in aqueous electrolytes.Instead, it undergoes dissolution and likely electrochemical transformation into another Mn-based phase, which remarkably maintains overall performance and stability.TERS also highlights the preferential adsorption of SO4 2-species at grain boundaries, with their bond strength appearing to intensify during cycling.In addition to the diminishing Mn3O4 signal, TERS and macro-Raman Spectroscopy suggest an increase in disorder within the electrochemically active Raman modes, aligning seamlessly with the observed enlargement of defects at the sub-nanometer scale.Notably, this marks the first successful utilization of TERS to investigate grain boundary effects in battery materials, revealing the strong bonding of electrolyte-derived species to the grain boundary core and its surroundings.
Expanding our observations to the macroscopic scale, macro-Raman Spectroscopy provides robust confirmation of the increased disorder within electrochemically active phases.Importantly, macro-Raman Spectroscopy maintains the overall intensity of the Mn3O4 impurities peaks.This finding validates our earlier TERS discovery that Mn3O4 impurities primarily dissolve at the surface.Notably, weak points, particularly grain boundaries, appear to be safeguarded by the absorption of sulfate ions.
Our study advances our understanding of this cathode behavior in aqueous electrolytes, holding great promise for the development of more sustainable and stable electrode materials.We hope this knowledge does not only pertain to aqueous systems but also extends towards enhancing the stability of cathodes in organic electrolytes.The dynamic nature of grain boundaries and their interactions with electrolytes must be taken into account in future protective coating designs.
Structural characterization.Scanning electron microscopy (SEM) was performed at IREC with a Zeiss Auriga (30 kV Gemini FESEM column and an in-lens detector).X-ray diffraction was performed thanks to an XRD Bruker-D8 Advance diffractometer in θ−2θ configuration between 10 and 60° (step size of 0.01°).The region around the main reflection of Pt was avoided during the measurement to protect the detector.
Raman characterization.Tip-Enhanced Raman Spectroscopy (TERS) measurements were performed on an Xplora Nano (HORIBA), with a 632 nm laser with p-polarization (parallel to the tip axis) and Au-coated OMNI TERS tips and 100x objective and a laser power of ~3 mW.Maps were acquired with a resolution of 10 nm per pixel to build 500 × 500 nm maps, plotted using a 2 px weight average.Acquisition times of 5 s per spectra and a 1200 g•mm −1 grating were used.The protocol to ensure TERS signal consists of checking the TERS signal on a carbon nanotubes sample test before and after the experiment.In addition, TERS is demonstrated in our films, the reader is referred to the Supporting Information, section III, for more details about our TERS configuration.For the macro-Raman Spectroscopy measurements, Raman scattering spectra were measured using iHR320 Horiba Jobin-Yvon monochromator coupled with i-DUS CCD from Andor.NIR solid-state laser (785 nm, ~80 W/cm 2 ) was used as excitation source.The measurements were performed in backscattering configuration through the specific probes developed in IREC.The spectral position was corrected using monocrystalline Si as a reference by imposing the main Raman peak to 520 cm -1 .
Variable energy positron annihilation lifetime spectroscopy.VEPALS was carried out at the Mono-energetic Positron Source (MePS) beamline at HZDR (Germany) [50] using a digital lifetime CrBr3 scintillator detector.This was coupled to a Hamamatsu R13089-100 PMT µ-metal shielded and housed inside an Au solid casing.We used homemade software employing a SPDevices ADQ14DC-2X digitizer with 14-bit vertical and 2GS/s horizontal resolutions [51].The time resolution function was down to 0.250 ns.The spectra analysis consisted of a resolution function of two Gaussian with distinct intensities and shifts depending on the positron implantation energy, Ep.All spectra contained at least 1•10 7 counts.Typical lifetime spectrum N(t) is described by N(t)=i (1/i) Ii exp(-t/i), where i and Ii are the positron lifetime and intensity of the i-th component, respectively (Ii=1).All the spectra were deconvoluted using a non-linear least-squares fitting method employed within the fitting software package PALSfit .[52] It consisted of 5 discrete lifetime components, which directly show localized annihilation at 2 different defect types (sizes; τ1 and τ2), i.e., small vacancy-like defect and their agglomerations or clusters.3 rd and 4 th lifetime components (τ3 and τ4) are two pore populations of diameters d3 and d4.The 5 th component (not shown) is residual and originates from ortho-positronium annihilation in vacuum and pore networks.The pore size was derived using the Wada and Hyodo shape-free model.[37] The positron lifetime and its intensity have been probed as a function of positron implantation energy Ep or equivalently, implantation depth or film thickness.Positrons have been accelerated and monoenergetically implanted into samples in the range between Ep = 1-12 keV for depth profiling.A mean positron implantation depth was approximated using a simple material density (ρ = 4.02 g•cm -3 ) dependent formula: <z>=36/ρ•Ep 1.62 .[53] The average positron lifetime τaverage is defined as τaverage = ∑i τi•Ii.
Electrochemical measurements.For electrochemical measurements, we established electrical contacts with the LMO films by accessing the exposed Pt regions at the sample edges, securely embedding them within a durable dual-component resin.This resin was applied thinly enough to facilitate mechanically stable TERS and PALS characterizations.The electrochemical measurements were conducted employing an open cell assembly, wherein a substantial portion of the film's surface remained exposed, facilitated by the resin's coverage.The exposed surface area typically ranged from 30 to 40 mm 2 .This configuration allowed for the rapid quenching of the electrochemical cell, subsequent immersion of the film in DI water, and fast drying using a nitrogen gun for storage within a protective atmosphere.The electrochemical experiments encompassed a voltage window spanning from 0.5 to 1.05 V versus Ag/AgCl (3M KCl), employing a Pt mesh as the counter electrode.

Figure 1 -
Figure 1 -Characterization of the as-deposited films.a) SEM cross-section of the as-deposited LMO films.b) XRD pattern of the LMO films and the bare Pt-coated Si substrate.c) Positron lifetime vs. implantation energy.

Figure 2 :
Figure 2: Raman Spectroscopy and electrochemistry.a) Macro-Raman Spectroscopy.b) Average TERS spectrum of the TERS map in d).c) Topography obtained during the TERS mapping.Size 500 x 500 nm 2 d) TERS map, 500 x 500 nm 2 .Green represents the LiMn2O4 phase and blue the Mn3O4 phase.e) Overlay of topography and TERS, 500 x 500 nm 2 .

Figure 4 :
Figure 4: Positron annihilation lifetime spectroscopy of cycled films.a) Relative intensity of each type of defect as a function of cycling.Inset shows the positron average lifetime.b) Lifetime for components τ1 (vacancy-like defects) and τ2 (clusters of vacancies) as a function of size.d) Pore size derived from τ3 and τ4, resulting in d3 and d4, respectively, as a function of cycling.