Deciphering the Origin of Interface‐Induced High Li and Na Ion Conductivity in Nanocomposite Solid Electrolytes Using X‐Ray Raman Spectroscopy

Solid‐state electrolytes (SSEs) with high ionic conductivities are crucial for safer and high‐capacity batteries. Interface effects in nanocomposites of SSEs and insulators can lead to profound increases in conductivity. Understanding the composition of the interface is crucial for tuning the conductivity of composite solid electrolytes. Herein, X‐ray Raman Scattering (XRS) spectroscopy is used for the first time to unravel the nature of the interface effects responsible for conductivity enhancements in nanocomposites of complex hydride‐based electrolytes (LiBH4, NaBH4, and NaNH2) and oxides. XRS probe of the Li, Na, and B local environments reveals that the interface consists of highly distorted/defected and structurally distinct phase(s) compared to the original compounds. Interestingly, nanocomposites with higher concentrations of the interface compounds exhibit higher conductivities. Clear differences are observed in the interface composition of SiO2‐ and Al2O3‐based nanocomposites, attributed to differences in the reactivity of their surface groups. These results demonstrate that interfacial reactions play a dominant role in conductivity enhancement in composite solid electrolytes. This work showcases the potential of XRS in investigating interface interactions, providing valuable insights into the often complex ion conductor/insulator interfaces, especially for systems containing light elements such as Li, B, and Na present in most SSEs and batteries.

The addition of the inert oxide nanoparticles can increase the ionic conductivity by several orders of magnitude, and in most cases, without compromising, and sometimes even improving, other important properties of the electrolytes such as the electrochemical, mechanical, and interfacial stability.[26] This layer formation is caused by the electrochemical potential difference of the two compounds, which leads to a local ion redistribution.The discontinuity at the interface results in a deviation from local electroneutrality and consequently, the formation of a space-charge zone.Within this zone, the concentration of the charge-carrying defects is enriched, leading to enhancement in ionic conductivity.
Although the space charge model has often been used to explain the increased ionic conductivity in nanocomposite electrolytes, recent observations highlight a different phenomenon.Specifically, interface reactions can occur in several systems, particularly within metal hydride/oxide-based nanocomposites.This can lead to the formation of a defect-rich tertiary compound/phase at the interface between the two compounds. [18]ue to the highly defective nature of the interphase, the ionic conductivity can be several orders of magnitude higher than the starting compounds, thereby profoundly increasing the overall conductivity of the composite.The presence of such interphases will also undoubtedly modify the space charge layer.This could be the reason why the space-charge model has failed to quantitatively predict increased ion conductivity in several heterogeneous solid electrolytes. [13,15]nraveling the nature and composition of such interphases is crucial for the rational design of highly conducting nanocomposite electrolytes.Regrettably, only a few studies have been reported on this topic to date.This is likely due to the complex nature of interphases.They are often amorphous with nanometric thickness and low concentrations, making their characterization challenging using most conventional techniques.In addition, it is non-trivial to use techniques that require an in vacuo set-up, such as soft X-ray methods, for example, X-ray photoelectron spectroscopy (XPS) and electron microscopy techniques.These challenges are particularly pronounced when working with metal hydride-based nanocomposite electrolytes, as they tend to decompose under high vacuum conditions.
[29][30][31][32][33] Although each of these studies confirmed interface reaction between the metal hydride and the oxide, the nature of the interaction is still being debated.Taking LiBH 4 /oxide nanocomposite as an example, some NMR studies suggest that the [BH 4 − ] anion near the interface remains intact, while other NMR studies indicate the formation of B─O, SiO─BH 3 , Si─H, and Li─O bonds. [29,30,32,33]or example, Dou et al. [28] suggested the formation of SiO─BH 3 structure, while Lambregts et al. [33] proposed a structure in which BH 4 − interacts with a Si site, as depicted in Figure 1a.The presence of B─O bonds has also been observed with XPS, FT-IR, and NEXAFS.However, these results are obscured due to possible air exposure during the measurements, the small penetration depth of the techniques, and the decomposition of LiBH 4 in high vacuum. [28,30,32]Thus, there is still no conclusive and wellresolved information available regarding the local interface structure.
The limitations outlined above might be circumvented by utilizing a hard X-ray spectroscopic technique called X-ray Raman Scattering (XRS) spectroscopy, sometimes referred to as nonresonant inelastic X-ray scattering. [34]In this technique, a hard X-ray beam excites electronic transitions at soft X-ray absorption edges. [35,36]By measuring the intensity of the scattered photons (into solid angle and energy window) as a function of the energy loss between the incident and the scattered X-ray energy, information similar to soft X-ray absorption spectroscopy is obtained.This is schematically depicted in Figure 1b.Since XRS is an inelastic scattering technique, the scattering angle can be used to enhance quadrupole excitations, like electron inelastic scattering, that is, electron energy loss spectroscopy or EELS.[39][40][41][42] In addition, it is possible to perform in situ measurements, [37][38][39][40][41][42][43] as well as 3D tomography studies [37,44,45] by utilizing the imaging properties of XRS.Thus, XRS is a promising technique that could be applied to determine local structural environments in metal hydride/oxide nanocomposites for battery applications.
In this study, we explored XRS as a technique to study interface effects in nanocomposite solid electrolytes.Employing LiBH 4 -, NaBH 4 -, and NaNH 2 /metal oxide nanocomposites as archetypes, we investigated the chemical and structural transformations that occur in the local environments of Li, Na, and B upon nanocomposite formation.To this end, we prepared LiBH 4 /oxide, NaBH 4 /oxide, and NaBH 4 /oxide nanocomposites with mesoporous SiO 2 and -Al 2 O 3 , the most widely used oxides for composite electrolytes.The use of the two different mesoporous oxide scaffolds enabled us to unravel the effects of the nature of the oxide on the interface reaction/interaction, that is, the composition of the interface.Furthermore, we could uniquely probe the interface layer responsible for the high conductivities in nanocomposites with different metal hydride-metal oxide weight ratios.The B, Li, and Na local environments in this layer are immensely different compared to the pristine compounds.For example, near the interface between the metal hydride and SiO 2 , boron from [BH 4 − ] changes from tetrahedral coordination to a trigonal configuration.In addition, changes in the Li and Na environment indicate that Li + and Na + (from the metal hydrides) are greatly affected by interaction with the oxide, while an indication of a Na─O─N like bond is observed in the NaNH 2 -based nanocomposites.This work provides new insight into the interface interaction in metal hydride/oxide  [28] and Lambregts et al. [33] b) A schematic representation of the X-ray Absorption and X-ray Raman scattering process during an incident photon (with E 0 ) interacts with an atom in the ground state.This results in the excitation of a core electron and the case of XRS, an inelastically scattered photon (with E f ).
nanocomposites.Furthermore, it demonstrates that XRS is a promising technique to study light elements and amorphous materials.

Samples and Sample Preparation
Several LiBH 4 -, NaBH 4 -, and NaNH 2 /oxide nanocomposites were prepared by melt infiltration following the procedures described earlier. [18,46]LiBH 4 (purity > 98%, Sigma-Aldrich) was mixed with mesoporous alumina (-Al 2 O 3 ), silica (SBA-15), and grafted silica (M-SBA-15, with M = Al, Zr) and subsequently infiltrated by heating to 285 °C under hydrogen pressure.NaBH 4 (99.99%,Sigma-Aldrich) was mixed with mesoporous alumina (-Al 2 O 3 ) and silica (MCM-41), and subsequently infiltrated by heating to 525 °C under hydrogen pressure.The amount of metal hydride confined in the oxide pores, that is, the pore filling fraction (PF), was varied from 15% to 130% to probe the interface interaction specifically.In other words, the volume of LiBH 4 corresponds to 0.15 to 1.3 times the pore volume of the oxide.By excluding the contribution of bulk metal hydride to the XRS signal (via this approach), the interface interaction becomes more predominant at low concentrations of metal hydride in the nanocomposites.Reference compounds (Li Grafted silica (M-SBA-15, with M = Al, Zr) was prepared by drying 1.2 g SBA-15 in static air for 2 h at 250 °C.The grafting reaction was performed under an N 2 atmosphere using a Schlenk line.The reaction mixture was prepared by dissolving the needed amount of precursor salt, either Al(OC 3 H 7 ) 3 or Zr(OC 3 H 7 ) 4 , to obtain a Si/M ratio 10 in dry isopropanol.Subsequently, the dried silica scaffold was added, and the mixture was left to stir overnight.The resulting suspension was filtrated and washed with isopropanol.After a final drying (2 h at 120 °C) and calcina- tion procedure (4 h at 500 °C, GHSV = 20 mL min −1 g −1 N 2 /O 2 flow), the grafted silica was placed in an argon-filled glovebox.

General Characterization
Diffuse reflectance infrared Fourier transform spectroscopy measurements were performed in a Perkin-Elmer 2000 spectrometer equipped with a liquid nitrogen-cooled MCT detector.Spectra were recorded between 4500 and 500 cm −1 with 4 cm −1 resolution, averaging over 16 scans and using anhydrous KBr as a background.The porosity of the mesoporous oxides was probed with nitrogen physisorption measurements performed on a Micromeritics Tristar 3000.Using the Brunauer, Emmett, and Teller (BET) and Barrett, Joyner, and Halenda adsorption model theories, surface area and pore size distribution could be obtained. [47,48]The specific surface area (A BET ) and total pore volume, as determined from the adsorbed quantity close to nitrogen saturation pressure (p = p 0 ) are summarized in Table 1.
The conductivities of the LiBH 4 /(grafted) SiO 2 nanocomposites were determined by measuring electrochemical impedance spectroscopy (EIS) using a Princeton applied research Parstat 2273 potentiostat placed in a custom-made Büchi B-585 glass oven.Pellets (t = 0.5-1.0mm, ø = 13 mm) were prepared by pressing (P = 150 MPa) about 80-150 mg between stainless-steel electrodes covered with lithium foil.The EIS measurements were performed by heating the samples from RT to 130 °C with increments of 10 °C.An EIS measurement was acquired at each increment with a 20 mV RMS modulated alternating current po- tential in a frequency range from 1 MHz to 1 Hz.The complex impedance spectra were fitted to a least squares minimum with a circuit consisting of a resistor (R) in parallel with a constant phase element.Based on the obtained resistance value, the electrode area (A = 1.33 cm 2 ), and the thickness (t) of the pellet, the conductivity, , was calculated via  = t/AR.

X-Ray Raman Scattering Experiments
Lithium (Li) and boron (B) 1s (K-edge) XRS spectra were collected at the ID20 beamline [49] of the European Synchrotron Radiation Facility (ESRF), Grenoble, France.An air-tight, ex situ XRS cell developed by the authors' group (Figure 2) was placed in a pre-mounted cell holder in a transmission configuration, so the cell was in the same position in every measurement.During the measurement, the cell was kept under a vacuum environment.Sodium (Na) and boron (B) K-edge XRS spectra were collected at the P01 beamline of the Deutsches elektronen-synchrotron (DESY) Petra III, in Hamburg, Germany.A slightly modified version of the air-tight, ex situ XRS cell was mounted on a preconstructed cell holder in a configuration with an incidence angle of 10°.Both at ESRF and at DESY, the samples were prepared in an argon-filled glovebox by compressing 10-40 mg into a pellet (ø = 10 mm).Subsequently, the compressed pellet was placed in the XRS cell between two Kapton foils (ESRF) or between a Kapton foil and a Kapton/aluminium foil (DESY).The aluminum provides rigidity to the Kapton foil and prevents beam damage, thereby preventing contamination of the samples by air and moisture during measurement.
The XRS scans were performed using the inverse energy scan technique in which the scattered photons were analyzed at a fixed energy, and the energy transfer was controlled by tuning the incident photon energy.About 3-10 scans were taken for a sin-gle measurement, depending on the quality of the signal from the measured edge.The incident photon energy was selected with a Si(311) monochromator.The XRS spectra were collected by scanning the incident beam energy relative to the fixed analyzer energy of 9690 eV with a resolution of 7 eV.At ESRF, the XRS spectra were collected using Medipix detectors (2D photoncounting X-ray detectors with a 55 μm spatial resolution) with an average q-vector of 4.1 to 4.9 Å −1 (2 = 50-60°).At DESY, the XRS spectra were collected using Medipix detectors with an average q-vector of 4.5 Å −1 (2 = 55°).The identification of the detector pixels that record scattering from the sample, or the regions of interest (ROIs), were defined manually.The scattering signals obtained from the selected ROIs were normalized to the f-sum rule [50] by background subtraction of parameterized Pearson VII functions guided by Hartree-Fock calculated core atomic profiles as described by Sahle et al. using the XRStools software package. [51]The final spectra were plotted as normalized scattered intensity versus energy loss.The spectra of the nanocomposites were smoothened by adjacent averaging over five points.Examples of the data analysis procedure are shown in Figures S1-S4 (Supporting Information).
Boron (B) K-edge XRS spectra of high purity LiBH 4 (99% and 95% pure) were acquired to verify that the measurement cells were air-tight and that the samples were not exposed to air during sample preparation, transfer, and measurements.The spectrum of the high-purity sample did not show any peaks related to impurities or oxidized compounds (Figure S5, Supporting Information), after 90 min scan time (ten scans) which confirmed that these measurement cells were air-tight and protect the samples from air exposure during sample preparation, transfer, and measurement.Oxidation or beam-induced sample damage during the XRS measurement was monitored by comparing the initial scans of the measurement to subsequent scans.In this case, a slight decrease in the intensities of the peaks was observed (Figure S5, Supporting Information).This indicated that while the samples did not oxidize during the measurement, the sample was affected by prolonged beam exposure.These changes were attributed to the low stability of borohydrides in the beam.To minimize beam damage, measurement durations were reduced, and for extended measurements, the beam was repositioned across various areas of the sample at regular intervals during the measurement.

X-Ray Raman Scattering Analysis of LiBH 4 /SiO 2 Nanocomposites
Our analysis starts with LiBH 4 /SiO 2 nanocomposites, one of the most widely investigated hydride/oxide nanocomposite electrolytes.Li and B K-edges XRS spectra of pristine LiBH 4 and LiBH 4 /SiO 2 with 130%, 50%, and 15% pore filling were measured (Figures 3b-e, S6, Supporting Information).In the LiBH 4 /SiO 2 nanocomposite with 130% pore filling, the LiBH 4 volume is 1.3 times the total pore volume of the scaffold, ensuring complete filling of the scaffold pores.Moreover, additional LiBH 4 covers the outer surface of the oxide particles, establishing a conductive path for long-range Li-ion transport over the non-conducting oxide particles.The lower pore-filling fractions roughly correspond to a 1 nm (PF = 50%) and 0.3 nm layer (PF = 15%) of LiBH 4 covering the SiO 2 pore walls, assuming that LiBH 4 completely wets the silica surface.The composition of the nanocomposites is schematically illustrated in Figure 3a.Since the interface layer in LiBH 4 /oxide nanocomposites is 1-2 nm, [16,52,53] the nanocomposites with ≤50% PF fractions will likely provide more specific information on the LiBH 4 -SiO 2 interface.
The changes in the chemical structure of LiBH 4 upon nanocomposite formation in mesoporous SiO 2 are explored by comparing pristine LiBH 4 to the LiBH 4 /SiO 2 nanocomposites.In the Li K-edge spectra of pristine LiBH 4 (Figure 3b, black line), a clear absorption peak is visible at 59.9 eV.56][57] The B K-edge spectra of pristine LiBH 4 show three characteristic features (Figure 3c, black line) as follows: I.An edge peak is observed at 191.5 eV, which is associated with the transition of a B 1s electron from the tetrahedral [BH 4 ] − anion to an unoccupied boron antibonding 2a 1 orbital (Figure 4a). [58]I.A small peak can be observed at 193.8 eV, which is attributed to the transition of B 1s electrons to unoccupied a″ orbitals in planar system [58][59][60][61][62] (Figure 4b), such as BF 3 , BH 3 , or trigonal B─O from LiBO 2 or B 2 O 3 impurities commonly found in commercially available LiBH 4 .III.[60][61] These results are in line with previous XRS results reported by our group. [38,39]pectedly, clear differences exist between the spectra of pristine LiBH 4 and those of the nanocomposites.First, in the Li K-edge spectrum (Figure 3b) of the LiBH 4 /SiO 2 nanocomposites with 130% PF, the peak at 59.9 eV has become less intense compared to pristine LiBH 4 .Second, a broad feature is observed around 64 eV.The intensity of the peak at 59.9 eV decreases with lower pore filling, and a shoulder appears at 55.6 eV.Third, the region between 62 and 75 eV starts to show several features.The reduced intensity of the peak at 59.9 eV, combined with the formation of a shoulder at 55.6 eV, demonstrates that the unoccupied orbitals of Li + become progressively filled and the bond between the BH 4 anion and Li becomes less ionic or weaker, for instance, due to the presence of a less electronegative anion (O, BH 3 ), as would be expected for Li─OBH x . [39,55,56]Likewise, the features in the fingerprint region between 62 and 75 eV is similar to multiple scattering resonances in lithium compounds such as Li 2 O, LiOH, Li 2 O 2 , and LiBO 2 [56,63,64] Overall, these results suggest that near the SiO 2 surface, two different lithium compounds are present, one in which Li + is weakly bonded to the complex anion (e.g., BH 3 or highly distorted BH 4 ), and one in which Li is closely bonded to O, such as in Li─O, Li─O─BH x , or Si─O─Li.
Similarly, in the B K-edge spectra (Figure 3c), clear changes are observed when comparing pristine LiBH 4 to the LiBH 4 /SiO 2 nanocomposites.In the spectrum of the nanocomposite with 130% pore filling, the peak at 191.5 eV associated with the tetrahedral 2a 1 transition is less intense compared to the spectrum of pristine LiBH 4 , while the feature attributed to trigonal boron is more intense and appears at a slightly more positive energy (194.2eV).In the B K-edge spectra of the nanocomposites with 50% (Figure 3c) and 15% PF (Figure S6, Supporting Information), the characteristic [BH 4 ] a 1 peak observed at 191.5 eV is not present anymore.Instead, the feature at 194.2 eV, which corresponds to trigonal B, has transformed into a prominent and well-defined peak.The change in the ratio between the tetrahedral and the trigonal boron peaks shows that the nanocomposites contain significantly more trigonal boron than pristine LiBH 4 , especially in the nanocomposites with a lower pore-filling fraction.It strongly suggests that upon nanocomposite formation with the mesoporous silica, an interface reaction occurs between LiBH 4 and the oxide, forming trigonal boron compounds from BH 4 − .While these analyses provide general information about the chemical environment of lithium and boron in the nanocomposites, the precise interface composition remains unclear.Therefore, to obtain more detailed information, the spectra were fitted using a linear combination of available relevant reference compounds (Figure S7, Supporting Information).The experimental data of the LiBH 4 /SiO 2 nanocomposites were fitted with the B K-edge spectra of pristine LiBH 4 , Li 2 B 12 H 12 , LiBO 2 , and H 3 BO 3 (Figure S8, Supporting Information) to distinguish between the tetrahedral and trigonal boron species present.It should be noted that the trigonal boron species in the nanocomposites could also be based on hydrogen (e.g., BH 3 ), especially as both LiBO 2 and H 3 BO 3 are not ionically conductive.However, there are no suitable references for this configuration.The resulting linear combination fits are plotted in Figure S8 (Supporting Information), and the corresponding fitting parameters are given in Table S1 (Supporting Information).Although not perfect, a comparison of the linear combination fits to the measured data indicates that the model explains the data to a reasonable/acceptable extent (adjusted R 2 > 0.83).The origin of the limitation will be explained later.
The linear combination fit of the LiBH 4 /SiO 2 nanocomposite with 130% pore filling confirms that the nanocomposite consists largely of LiBH 4 (Figure S8a, Supporting Information).A small amount of Li 2 B 12 H 12 , a well-known decomposition product of LiBH 4 , is observed as well.Furthermore, the trigonal feature at 194.2 eV can be fitted with a combination of the B─O species, essentially resembling LiBO 2 and partially resembling H 3 BO 3 .Note that the peak around 191.5 eV in the fit has a higher intensity compared to the experimental data.Hence, either the amount of LiBH 4 or Li 2 B 12 H 12 present in the nanocomposite is overestimated, or the anions contain more negative charge (possibly due to the formation of Li + vacancies).In the nanocomposites with lower pore filling fractions, that is, 50%, the contributions of LiBH 4 and Li 2 B 12 H 12 are not considered since the characteristic peak around 191.5 eV is not present in these samples (Figure S8b,c, Supporting Information).Instead, the experimental data are fitted solely with the trigonal reference compounds.Based on the resulting linear combination fits of both nanocomposites, it seems that a mixture of trigonal boron phases is present at the LiBH 4 -SiO 2 interface.Interestingly, the ratio between LiBO 2 and H 3 BO 3 characters in the LiBH 4 /SiO 2 nanocomposites changes upon going from 50% PF to 15% PF (Table S1, Supporting Information).The fitting parameter of the LiBO 2 contribution becomes smaller, while the fitting parameter of the H 3 BO 3 contribution becomes larger.It appears that closer to the SiO 2 surface, the interface layer between metal hydride and oxide contains compounds where the energy associated with the trigonal a″ transition shifts to higher values.This phenomenon is observed in compounds like H 3 BO 3 when compared to LiBO 2 .This suggests that the structure within the interface layer differs depending on the proximity of the oxide or metal hydride.Additionally, it might be influenced by interactions with the different silanol groups on the SiO 2 surface.These groups include surface siloxanes, isolated, geminal, and vicinal silanols, which vary slightly in their chemical bonds. [65]t is important to note that the interface layer is a highly defective and distorted phase.These defects and distortions can greatly affect the local chemical environment of the boron species.The B K-edge spectra (Figure 3c) illustrate that the LiBH 4 -SiO 2 interface layer contains trigonal boron (such as LiBO 2 or BH 3 ).However, the precise chemical environment will be different from the purely crystalline phases that have been measured as reference compounds.The same is true for the chemical environment of Li + .While the Li K-edge spectra exhibit features that can be attributed to the presence of Li 2 O, it is more likely that these features are related to the formation of Li─O bonds.In summary, our observations are in accordance with the formation of a Si─H─BH 3 like structure and Li─O (or more precisely Si─O─Li) bond via the reduction of a siloxane bond as proposed by Lambregts et al., though the formation of the Si─O─Li─BH 3 structure discussed by Dou, and Wang, et al. cannot be excluded, as both cases result in trigonal boron. [28,32,33]We also observed the formation of a Li + -compound that has a weak interaction with the surrounding anions.This is in accordance with the reduced activation energies for ionic transport in LiBH 4 /SiO 2 nanocomposites (0.4-0.5 eV) compared to pristine LiBH 4 (0.7-0.9 eV), [16,52,66,67] and could explain the enhanced ionic conductivity upon nanocomposite formation.For a more definite understanding of the precise chemical environment at the interface, the utilization of model systems in combination with theoretical simulations is required.

NaBH 4 -and NaNH 2 /Oxide Nanocomposites
Following a methodology similar to the investigation conducted on the Li-ion conductor LiBH 4 , the effect of nanocomposite formation on the chemical structure of the Na-ion conductors NaBH 4 and NaNH 2 was examined.This was achieved by studying NaBH 4 /SiO 2 and NaNH 2 /SiO 2 nanocomposites with different (PF) fractions.In Figure 5a,c, the B and Na K-edge XRS spectra of pristine NaBH 4 and NaBH 4 /SiO 2 nanocomposites with 30% and 100% PF are depicted, while the spectra of some relevant reference compounds are shown in Figure S9a-c (Supporting Information).
In the B K-edge spectra of the NaBH 4 /SiO 2 nanocomposites (Figure 5a), evident distinctions from the pristine NaBH 4 spectrum are observed.The peak at 191.7 eV, associated with the 2a 1 transition of tetrahedral B (in BH 4 − ), becomes less intense upon nanocomposite formation.At the same time, the spectra of the nanocomposites contain a feature at 194.1 eV, which is not observed in pristine NaBH 4 .Based on Figure 4, the peak around 194 eV is attributed to the transition of B 1s electrons to unoccupied a″ orbitals of trigonal boron, such as BO 2 − and H 3 BO 3. In the NaBH 4 /SiO 2 nanocomposite with 30% pore filling, the peak related to NaBH 4 is smaller than the peak corresponding to trigonal boron.Thus, at the interface between NaBH 4 and SiO 2, a layer consisting of trigonal boron, Na─O, and weakly coordinated Na + seems to form, similar to the interface layer between LiBH 4 and SiO 2 .The Na K-edge spectra of the NaBH 4 /SiO 2 nanocomposites (Figure 5c) are compared to that of the pristine NaBH 4 .In the spectrum of pristine NaBH 4 , two peaks are observed at 1076.2 and 1078.9 eV.These peaks are attributed to the transition of a 1s electron to an unoccupied 3p state of NaBH 4 . [68,69]The Na K-edge spectrum of NaBH 4 /SiO 2 with 30% PF contains an additional pre-edge peak at 1073.9 eV and a broad feature around 1082 eV.As similar changes were also observed in the Li K-edge spectra of LiBH 4 /SiO 2 , (Figure 3b), this points to the formation of Na + bonded to, or in closer proximity to O, as well as Na + that weakly interacts with the anion (highly distorted BH 4 or BH 3 ).
Analogous to the NaBH 4 /SiO 2 , the interfacial interaction in NaNH 2 /SiO 2 nanocomposites was investigated.In Figure 5f, it is quite clear that when confined in mesoporous SiO 2 , NaNH 2 exhibits more profound changes in the Na K-edge compared to NaBH 4 as the two characteristics peaks are both shifted to much higher energy loss values than in NaBH 4 /SiO 2 (Figure 5c).The shift to the higher energy loss is due to the formation of a Na─N─O x ─H y -like compound (as in NaNO 2 or NaNO 3 ) at the NaNH 2 /SiO 2 interface (See Figure S9c, Supporting Information), in which N has a higher binding energy than in NaNH 2 .This is a clear evidence that NaNH 2 reacts with SiO 2 surface groups to form an interface compound that is distinct and more conductive than the original NaNH 2 .The presence of Na─NOH x like bond is also a major indication that the interface compounds in LiBH 4 and NaBH 4 -based nanocomposites indeed contain "B─O like" bonds.

Effects of the Nanoscaffolds: SiO 2 versus Al 2 O 3
The nature of the oxide scaffolds has been reported to have a profound impact on the conductivity of enhancements in composite electrolytes. [13,16,17,18,19]Therefore, it is interesting to study the effect of the chemical nature of the oxide scaffolds on the interface chemistry/composition.In line with this objective, we also measured the XRS spectra of nanocomposites prepared with mesoporous -Al 2 O 3 , another widely employed oxide.Figures 3b-e and 5a-f compare the K-edge spectra of the LiBH 4 -, NaBH 4 -, and NaNH 2 -based nanocomposites prepared using SiO 2 and Al 2 O 3 .A comparison of Figure 3b,d shows major similarities, but also minor differences, between the Li K-edge spectra of LiBH 4 /SiO 2 and LiBH 4 /Al 2 O 3 .At low LiBH 4 concentration (50% PF), a minor peak is observed around 62 eV in LiBH 4 /Al 2 O 3 which is not obvious in LiBH 4 /SiO 2 .This reveals that the chemical environment (nature/composition) of Li is mostly, but not completely, similar to the SiO 2 and Al 2 O 3-based nanocomposites.This is remarkably in agreement with previous reports of similar ionic conductivities and activation energies of 0.43 and 0.44 eV, respectively for LiBH 4 /SiO 2 and LiBH 4 /Al 2 O 3 prepared in the same manner. [16,27]However, note that the conductivity might vary if the specific surface area, pore volume, and/or pore size of the oxide is changed.
For the B K-edge spectra, although similarities exist, there are more notable dissimilarities between the two samples.Analogous to LiBH 4 /SiO 2 , LiBH 4 /Al 2 O 3 contains the a″ peak around 194.1 eV indicative of interfacial trigonal boron compounds, and the peak at 191.5 eV which corresponds to BH 4 − becomes smaller with lower LiBH 4 concentration (pore filling).In contrast, in LiBH 4 /Al 2 O 3 , the a″ peak is broader and at a slightly lower energy loss value (≈193.7 eV) than for SiO 2 .Intriguingly, the peak shifts to energy loss values similar to LiBH 4 /SiO 2 (194.1 eV) when the concentration of LiBH 4 is reduced to 15% PF (see Figure S6, Supporting Information).This is a clear indication of a significant difference in the local environments of the interfacial boron compounds in SiO 2 and Al 2 O 3 .The origin of these differences will be discussed later, although it appears the differences have negligible impact on the ionic conductivity.
For the B K-edges (Figure 5a These differences in the nature of the interface compounds for SiO 2 and Al 2 O 3 can be attributed to the differences in the chemical nature or surface chemistry and acidity of the two oxides, which dictates the nature of the surface reaction.For example, the surface of silica generally contains about 4 to 5.5 hydroxyl groups per nm 2 , which are weak Brønsted acidic sites, while the surface of -Al 2 O 3 contains between 10 and 15 hydroxyl groups per nm 2 , both Brønsted bases and Brønsted acids.[72][73][74] Thus, it is expected that the cation (Li + or Na + ) and the complex anion (BH 4 − or NH 2 − ) in the molten metal hydride will interact differently with the -Al 2 O 3 and SiO 2 surfaces.For example, if BH 4 − or NH 2 − interacts with an uncoordinated Al-site instead of an acidic hydroxyl group, the tetrahedral boron configuration might remain intact rather than forming a trigonal boron compound.As a result, the energy of the 1s to 2a 1 transition would shift to slightly different values compared to the latter case.This means that at least two different interfacial B species would be expected, arising from the BH 4 − interacting with an uncoordinated Al site, and those bonded to the hydroxyl groups.
Likewise, the presence of a more stable interfacial compound in NaNH 2 /Al 2 O 3 than in NaNH 2 /SiO 2 can be related to the strong Lewis acid/basic sites of the Al 2 O 3, which will expectedly lead to a stronger interaction with the complex hydrides.To assign these subtle differences in the chemical environment of the elements to specific compounds in the metal hydride-oxide interfacial layer, the experimental data need to be supported by ab initio simulations.While this is a topic for a future project, we have successfully showcased that the interface layer in metal hydride/alumina nanocomposites exhibits a markedly different composition compared to metal hydride/silica nanocomposites.It is remarkable that both scenarios lead to the creation of highly conductive nanocomposites.

Correlation of Interface Composition to Ionic Conductivity
Based on the XRS results discussed in the previous sections, it is clear that the interphases in nanocomposite electrolytes depend strongly on the type of oxide scaffold used.This agrees with the conductivity results (Figure S10, Supporting Information), and previous studies that highlight the strong impacts of the metal oxide type on the conductivity of the nanocomposites.Notably, the impact on conductivity differs among the three hydride-based ion conductors studied here.For emphasis, higher conductivity is obtained for NaNH 2 /SiO 2 than NaNH 2 /Al 2 O 3 while the reverse is the case for NaBH 4 /SiO 2 and NaBH 4 /Al 2 O 3 .As explained in detail in our recent work, this behavior is a result of the complex interplay between the nature or reactivity of the oxide surface groups and the stability or reactivity of the ionically conductive metal hydrides.These properties dictate the accurate hydrideoxide reaction/interaction, hence the nature, stability, and conductivity of the interface compounds.For example, the lower conductivity of NaNH 2 /Al 2 O 3 compared to NaNH 2 /SiO 2 can be attributed to the formation of a more chemically stable (less defected) interface compound with Al 2 O 3 than with SiO 2 , as revealed by the XRS results.
Due to differences in the physical properties of the oxides (surface area, pore size, pore volume, and morphology) which are known to also influence ionic conductivity, it is not straightforward to compare these oxides on the same scale.To overcome this limitation and thereby establish a link between the interface layer and conductivity, we explored the concept of surface functionalization by grafting.For this, we prepared aluminium-and zirconium-grafted SiO 2 and studied the conductivity and interface composition of LiBH 4 /oxide nanocomposites based on these grafted SiO 2 nanoscaffolds.The surface chemistry of the Al-and Zr-grafted SiO 2 scaffolds differs from pristine SiO 2 , [19,[75][76][77] while their physical properties (e.g., surface area, pore volume, and morphology) are the same.In this way, the strength of the interface interaction can be tuned, making it possible to uniquely determine how the ionic conductivity is influenced by the LiBH 4 -SiO 2 interface composition.
In Figure 6a, the B K-edge XRS spectra of LiBH 4 /M─SiO 2 nanocomposites (M = Al and Zr) with 130% pore filling are reported.Surprisingly, the peak at 191.5 eV decreases in intensity when going from SiO 2 to Al─SiO 2 to Zr─SiO 2 , while the peak at 194.0 eV increases.This suggests a reduction in the number of tetrahedral boron species, namely BH 4 -, alongside an increase in the prevalence of trigonal boron compounds.The interface interaction between LiBH 4 and Zr─SiO 2 is stronger (Zr is a stronger Lewis acid than Al and SiO 2 ), and as a result, more LiBH 4 is converted to trigonal B interface species.In other words, the interphase extends further from the interface, or more interface compound is present.Notably, the conductivity data (Figure 6b) manifest a strong correlation between the interaction strength or composition of the LiBH 4 -oxide interface and the ion conductivity.At 30 °C, LiBH 4 /SiO 2 exhibits a conductivity of 0.2 × 10 −5 S cm −1 , whereas the LiBH 4 /Al─SiO 2 and LiBH 4 /Zr─SiO 2 nanocomposites exhibit higher conductivities of 1.4 × 10 −5 and 5.1 × 10 −5 S cm −1 , respectively.These results clearly underscore the fact that LiBH 4 /oxide nanocomposites with more trigonal boron species exhibit higher conductivity.Therefore, our study reveals that the formation of more trigonal boron also corresponds to the formation of a more highly defected interphase and/or weakly coordinated Li-ions.As a result, this facilitates fast Li-ion transport in nanocomposites based on LiBH 4 and NaBH 4 .Accordingly, the ionic conductivity of nanocomposite solid electrolytes can be significantly increased by maximizing the interface reaction via the use of a more reactive surface groups or scaffold materials with very high surface areas.

Conclusions
XRS has been utilized to decipher the nature of ion conductor/insulator interfaces leading to high conductivities in composite solid electrolytes, specifically, nanocomposites containing complex hydrides (LiBH 4 , NaBH 4 , and NaNH 2 ) and metal oxides.Investigation of the local environment of Li, Na, and B shows that these ionic conductors react with the oxide's (SiO 2 and Al 2 O 3 ) surface groups upon nanocomposite formation.This results in the formation of an interphase at the metal hydride/oxide interface.For the boron-based hydrides (LiBH 4 and NaBH 4 ), the original tetragonal (BH 4 ) structure transforms into mostly trigonal boron such as ─BH 3 with a "B─O" like character.Likewise, for NaNH 2 , a clear indication of the N─O bond is observed at the interface, a major validation of the reaction of the hydrides with the surface OH groups.Moreover, in this interphase, part of the Li and Na are connected to oxygen, suggesting an M─O bond at the interface.While unveiling the intricate structure of the metal hydride-oxide interface poses challenges, the XRS results show that the formed interphase is strongly influenced by the physico-chemical properties of the oxide employed.Remarkably, nanocomposites with higher concentrations of the interface compound also exhibit higher ionic conductivities.Therefore, our work reveals that surface reactions play a dominant role in interfacial ion dynamics and establishes for the first time that a direct correlation exists between the nature of the interphase and ion mobility in nanocomposite solid electrolytes.This fundamental insight is of utmost importance for the rational design of novel superionic conductors via interface engineering.Furthermore, these findings underscore the potential of XRS as a promising technique for studying battery materials and interfaces with often low atomic weight.These aspects are not easily explored by conventional techniques.

Figure 1 .
Figure 1.a) 2D representation of the LiBH 4 -SiO 2 interface structures proposed by Dou et al.[28] and Lambregts et al.[33] b) A schematic representation of the X-ray Absorption and X-ray Raman scattering process during an incident photon (with E 0 ) interacts with an atom in the ground state.This results in the excitation of a core electron and the case of XRS, an inelastically scattered photon (with E f ).

Figure 2 .
Figure 2. a) Schematic representation of the air-tight cell used for ex situ XRS measurements used in the b) transmission and c) grazing incidence configurations.

Figure 3 .
Figure 3. a) Schematic representations of LiBH 4 /SiO 2 nanocomposites with 130% and 50% of the oxide pores filled with LiBH 4 .b,d) Li K-edge and c,e) B K-edge XRS spectra of pristine LiBH 4 powder (95% purity) and LiBH 4 /oxide nanocomposites based on SiO 2 and Al 2 O 3 with 130% and 50% PF.The spectra of nanocomposites are smoothened using adjacent averaging over five points.

Figure 4 .
Figure 4. Schematic representation of the molecular orbital diagrams of a) BH 4 − and b) BF 3 based on symmetry-adapted linear combinations (SALCs).The corresponding molecular geometries, that is, tetrahedral for BH 4 − and trigonal planar for BF 3 , are provided as well.

Figure 5 .
Figure 5. a,b) B K-edges and c,d) Na K-edges of NaBH 4 /SiO 2 and NaBH 4 /Al 2 O 3 nanocomposites, respectively.e,f) Na K-edges of NaNH 2 /SiO 2 and NaNH 2 /Al 2 O 3 nanocomposites, respectively.The spectra of the pristine NaBH 4 and NaNH 2 are added for reference.The spectra are smoothened using adjacent averaging over five points.
,b), the position of the peak attributed to the interface trigonal boron is shifted from ≈194.1 eV in NaBH 4 /SiO 2 to ≈193.7 eV in NaBH 4 /SiO 2 .This implies a slightly different chemical environment for B in SiO 2 and Al 2 O 3based samples, as observed in the LiBH 4 /oxide nanocomposites.From Figure 5c,d, it can be seen that although the Na K-edges of NaBH 4 looks similar in the Al 2 O 3 and SiO 2 -based nanocomposites, the pre-peak feature at 1073.9 eV in NaBH 4 /SiO 2 is absent in NaBH 4 /Al 2 O 3 .Extending the comparison to NaNH 2 /SiO 2 versus NaNH 2 /Al 2 O 3 (Figure 5e,f), it is clear that the two main peaks of the Na K-edge spectra are located at a slightly higher energy loss values in Al 2 O 3 than in SiO 2 .Specifically, from 1076.4 and 1083.4 eV in NaNH 2 /SiO 2 to 1076.7 and 1083.8 eV in NaNH 2 /Al 2 O 3 .This suggests that NaNH 2 forms a slightly more stable (higher binding energy) interface compound with Al 2 O 3 than with SiO 2 , which could explain why the NaNH 2 /SiO 2 is far more conductive than the NaNH 2 /Al 2 O 3 .