Conductor–Insulator Interfaces in Solid Electrolytes: A Design Strategy to Enhance Li-Ion Dynamics in Nanoconfined LiBH4/Al2O3

Synthesizing Li-ion-conducting solid electrolytes with application-relevant properties for new energy storage devices is a challenging task that relies on a few design principles to tune ionic conductivity. When starting with originally poor ionic compounds, in many cases, a combination of several strategies, such as doping or substitution, is needed to achieve sufficiently high ionic conductivities. For nanostructured materials, the introduction of conductor–insulator interfacial regions represents another important design strategy. Unfortunately, for most of the two-phase nanostructured ceramics studied so far, the lower limiting conductivity values needed for applications could not be reached. Here, we show that in nanoconfined LiBH4/Al2O3 prepared by melt infiltration, a percolating network of fast conductor–insulator Li+ diffusion pathways could be realized. These heterocontacts provide regions with extremely rapid 7Li NMR spin fluctuations giving direct evidence for very fast Li+ jump processes in both nanoconfined LiBH4/Al2O3 and LiBH4-LiI/Al2O3. Compared to the nanocrystalline, Al2O3-free reference system LiBH4-LiI, nanoconfinement leads to a strongly enhanced recovery of the 7Li NMR longitudinal magnetization. The fact that almost no difference is seen between LiBH4-LiI/Al2O3 and LiBH4/Al2O3 unequivocally reveals that the overall 7Li NMR spin-lattice relaxation rates are solely controlled by the spin fluctuations near or in the conductor–insulator interfacial regions. Thus, the conductor–insulator nanoeffect, which in the ideal case relies on a percolation network of space charge regions, is independent of the choice of the bulk crystal structure of LiBH4, either being orthorhombic (LiBH4/Al2O3) or hexagonal (LiBH4-LiI/Al2O3). 7Li (and 1H) NMR shows that rapid local interfacial Li-ion dynamics is corroborated by rather small activation energies on the order of only 0.1 eV. In addition, the LiI-stabilized layer-structured form of LiBH4 guarantees fast two-dimensional (2D) bulk ion dynamics and contributes to facilitating fast long-range ion transport.


■ INTRODUCTION
Hydride-based solids attracted great attention as promising electrolytes for lithium-ion batteries 1 due to their compatibility with Li metal and their mechanical robustness. 2,3 While Li + -ion transport in polycrystalline oxide-type electrolytes 4 may suffer from large grain-boundary resistances, such regions do not hinder long-range ion transport in the mechanically softer hydrides. 5 The most prominent model hydride is LiBH 4 whose hexagonal modification, which is stable above T pt = 110°C, shows high conductivities in the mS cm −1 range. 6,7 The corresponding orthorhombic form, being the favorable crystal structure below T pt , is, however, a rather poor ion conductor, 7 likely because of much higher defect formation energies. 8 While ultraslow Li + ion exchange in orthorhombic LiBH 4 is assumed to take place in three dimensions (Figure 1), for layer-structured LiBH 4 , a two-dimensional (2D) conduction mechanism prevails (Figure 1), as has been shown by both frequency-dependent 7 Li NMR spin-lattice relaxation (SLR) measurements 9−11 and calculations. 12 This 2D diffusion behavior is illustrated in Figure 1 using bond valence site energy estimations.
Two approaches have been established and presented in the literature that successfully enhance the room-temperature ionic conductivity of LiBH 4 by several orders of magnitude, viz., (i) nanoconfinement of LiBH 4 in insulating oxides 13,14 and (ii) and partial cationic and especially anionic substitution 6,7 of the BH 4 − units with halogen ions like I − , Br − , or Cl − . It is strongly anticipated that these two approaches lead to fundamentally different diffusion mechanisms. While anion substitution in LiBH 4 -LiX (X = I, Br, Cl) stabilizes the highly conductive hexagonal phase at much lower temperatures than T pt , 15 through nanoconfinement, a large fraction of Li + -ion conductor−(ionic)insulator interfacial regions are introduced, which are suggested to be responsible for increased long-range ion transport. A definite proof of the latter concept or effect is, however, still missing for the LiBH 4 /Al 2 O 3 composites. For LiBH 4 /SiO 2 nanocomposites, the important role of surface groups has been discussed recently. 16,17 In general, heterocontacts between two different phases, viz., an ion conductor and an insulating phase, or even between two (mixed) conductors, may generate a percolation network of space charge regions with enhanced charge carrier mobility. The most prominent two-phase system is composed of alternating layers of F − -ion-conducting BaF 2 and CaF 2 with thicknesses of 9 nm, which were grown by molecular beam epitaxy. 18 The foundations of space charge zones in such nanostructured artificial ion conductors were laid by Maier, 19−22 explaining such nontrivial size effects that rely on overlapping space charge zones. CuBr/Al 2 O 3 (TiO 2 ) composites, as studied by Knauth and co-workers, 23 belong to another group of such composites that show enhanced electrical conductivity. 24,25 In the case of Li-ion conductors, Liang observed increased ionic conductivities in samples of LiI/Al 2 O 3 . 26 Later, for the nanocrystalline system Li 2 O/X 2 O 3 (X = B, Al), similar effects 27,28 were observed by 7 Li NMR SLR rate measurements. 29,30 Although enhanced Li + diffusivity was probed, the resulting conductivities 28 could not reach any practical benchmark needed to realize all-solid-state batteries equipped with ceramic electrolytes.
This situation is, however, different for nanoconfined LiBH 4 -LiI/Al 2 O 3 and LiBH 4 /Al 2 O 3 , both showing ionic conductivities in the order of 10 −4 S cm −1 , 31 which is by more than a factor of 100 higher than in orthorhombic LiBH 4 . 13 To understand the synthetic approaches and their impact on overall ionic transport, the role of the conductor−insulator interface in achieving such high conductivities needs to be studied in detail. Preferably, such studies should include spectroscopic methods 32 being sensitive to local Li + hopping processes in or near these interfacial areas. Here, we used 7 Li NMR spectroscopy to quantify the effect of the conductor−insulator interfacial regions in nanoconfined LiBH 4 -LiI/Al 2 O 3 and LiBH 4 /Al 2 O 3 prepared by melt infiltration. Although a recent 27 Al NMR study in our labs suggested that (unsaturated) penta-coordinated Al centers near the Al 2 O 3 surface regions are involved in creating a defect-rich LiBH 4 /Al 2 O 3 interface, 33 the ultimate proof via 7 Li NMR SLR measurements is still missing. In the present study, we directly compared the 7 Li (and 1 H) NMR response of longitudinal SLR of LiBH 4 (-LiI)/ Al 2 O 3 with those of bulk LiBH 4 and LiI-stabilized LiBH 4 . We observed a tremendous effect of the insulator Al 2 O 3 on 7 Li NMR SLR, which is directly proportional to the diffusive motions of the Li + ions, clearly showing the superior role of conductor−insulator regions in solid electrolytes with nanometer-sized dimensions. Only in such samples, the volume fraction of these regions is large enough to have a dominant effect on overall ion transport properties.
Methods and Characterization. The composite electrolytes investigated here, i.e., LiBH 4 /Al 2 O 3 and LiBH 4 -LiI/Al 2 O 3 , were prepared via melt infiltration; LiBH 4 -LiI served as a reference compound. A detailed description of the corresponding procedure 34 as well as of the preparation and characterization of the composites 31 can be found elsewhere as the same samples were used for earlier studies. Ionic substitution was realized in a molar ratio of 80:20 (LiBH 4 /LiI). The samples were kept at 295°C for 30 min under 50 bar H 2 pressure in a stainless steel high-pressure autoclave (Parr). The average diameter of the pores in Al 2 O 3 is in the order of 10 nm. 31 As mentioned above, at room temperature, bulk LiBH 4 crystallizes with orthorhombic structure and transforms into its hexagonal phase at temperatures higher than 110°C; the corresponding X-ray diffraction patterns are shown in Figure S1. Here, results from differential scanning calorimetry (DSC), see Figure S2, reveal that the corresponding signal of LiBH 4 /Al 2 O 3 splits into two peaks at 103°C and 114°C; the signals are significantly decreased compared to the expected one of bulk LiBH 4 , which was found at 117°C. We do not observe any diagnostic DSC signals pointing to a phase change in LiBH 4 -LiI/Al 2 O 3 as for LiBH 4 -LiI, the hexagonal modification is stabilized by the introduction of LiI already at lower temperatures. For the sake of completeness, LiBH 4 -LiI shows a slight endothermic signal at −19°C. The thermal behavior of LiBH 4 /Al 2 O 3 is useful when interpreting the diffusion-induced 7 Li NMR data, which were collected as follows.
Variable-temperature 7 Li (and 1 H) NMR 1/T 1 SLR rates were measured with a Bruker Avance III 300 spectrometer that is connected to a 7-Tesla cryomagnet. The corresponding Larmor frequencies were 116 MHz for 7 Li and 300 MHz for 1 H. All samples were smoothly hand-pressed in Duran tubes under protective atmosphere and sealed. Relaxation rates were determined at temperatures ranging from −100 to 200°C with an increment of usually 20°C. In the region of the diffusioninduced rate peaks, additional 7 Li NMR 1/T 1 rates were recorded every 10 or 5°C. The laboratory-frame 7 Li (and 1 H) NMR 1/T 1 rates were acquired with the well-known saturation recovery pulse sequence; depending on temperature, the 90°p ulse lengths (200 W) varied from 2.5 to 2.9 μs ( 7 Li) and from 1.1 to 3.3 μs ( 1 H). Usually four to eight scans were accumulated to obtain a single free induction decay. For a detailed description of the pulse programs used and for a discussion of the procedure employed to parameterize the longitudinal NMR transients (partly displayed in Figure S3), we refer to our previous study. 33 We also performed bond valence site energy estimations combined with a bond valence pathway analyzer using the softBV software tool developed by Adams and co-workers. 35,36 We took the structural information from published synchrotron X-ray powder diffraction data. 37 The softBV software executes a structure plausibility check, calculates surface energies, and gives information about the positions of interstitial sites and saddle points, as well as the topology Figure 2. (a) Arrhenius representation of the 7 Li NMR spin-lattice relaxation rates 1/T 1 (116 MHz) of microcrystalline LiBH 4 and the LiIstabilized form LiBH 4 -LiI; the former rates were taken from an earlier study by some of us. 9 Below the transition temperature of ca. 110°C (LiBH 4 , see gray area), the rates are governed by fast rotational BH 4 − dynamics in the orthorhombic form of LiBH 4 . In hexagonal LiBH 4 , 1/T 1 is determined by Li + translational dynamics. For LiBH 4 -LiI, the transformation temperature is reduced (see also (b)); furthermore, rotational dynamics gets enhanced as the 7 Li NMR rate peak is shifted toward lower temperatures, it appears at ca. 190 K. Solid and dashed lines are to guide the eye; see text for further details. (b) The same representation as in (a) but with the 7 Li NMR spin-lattice relaxation rates 1/T 1 of the two nanoconfined samples included, viz., LiBH 4 -LiI/Al 2 O 3 and LiBH 4 /Al 2 O 3 . Importantly, even for the sample free of any LiI, rather rapid Li + exchange processes are probed. This observation reveals the importance of the conductor−insulator interfacial regions determining overall 7 Li spin fluctuations in nanoconfined LiBH 4 /Al 2 O 3 . Activation energies refer to the almost linear regions of the 1/T 1 (1/T) dependence. Values as low as 0.1 eV point to an extremely flat potential landscape characterizing the conductor−insulator heterocontacts. See text for further explanation. and dimensionality of ion-migration paths and the respective migration barriers. 35,36 For the calculations, Li + was chosen as the mobile ion and the grid resolution was set to 0.1 Å. Pros and cons of the approach via softBV are discussed elsewhere. 36 For the visualization of the data, we used the VESTA software package. 38 ■ RESULTS AND DISCUSSION NMR Spin-Lattice Relaxation: Li-Ion Translational Dynamics and Rotational Jumps of the Polyanions. As mentioned above, LiBH 4 crystallizes either with orthorhombic or with hexagonal symmetry. At temperatures lower than T pt = 110°C, the poorly conducting orthorhombic modification is present (see Figure 1). In ortho-LiBH 4 , the 7 Li NMR spinlattice relaxation rates indirectly sense the rapid rotational BH 4 − dynamics rather than Li + translational diffusion (see Figure 2a). Thus, at temperatures below T pt , the 7 Li NMR rates pass through two rate peaks that mirror the two distinct rotational jump processes of the BH 4 − polyanions (see Figure  2a), which shows the 7 Li NMR SLR rates of coarse-grained, that is, microcrystalline LiBH 4 . The broad rate maximum located at 220 K is composed of two individual rate peaks, which are represented by dotted lines and labeled P1 and P2 in Figure 2a. These rate peaks were analyzed in detail by both NMR 9,39,40 and quasi-elastic and inelastic neutron scattering earlier; 41 additionally, the mobility of boron atoms is discussed elsewhere. 42 Above T pt , the overall 7 Li NMR response in bulk LiBH 4 is governed by rapid Li + (translational) jump processes in the layer-structured form of LiBH 4 . This dynamic process, which is 2D in nature as is illustrated in Figure 1, produces a single rate peak that points to an activation energy of 0.5 eV. In general, diffusion-induced 7 Li NMR rate peaks appear if the motional correlation rate 1/τ c , which is expected within a factor of ca. 2 to be identical with the jump rate 1/τ, reaches the order of the (angular) Larmor frequency ω 0 . 11 At the temperature where the peak appears, the condition ω 0 τ ≈ 1 is fulfilled. 43 A symmetric rate peak is only obtained for uncorrelated and isotropic (three-dimensional, 3D) diffusion. In many cases, structural disorder combined with Coulomb interactions results in asymmetric NMR peaks whose low-T flank shows a lower slope than that characterizing the flank on the high-T side. 43 Moreover, while the slope in the high-T regime is characteristic for long-range ion diffusion, the low-T flank of the peak is sensitive to short range, that is, local diffusion processes. 11 Stabilizing the hexagonal phase of LiBH 4 by the incorporation of LiI leads to several changes of the overall 7 Li NMR response (see Figure 2a). First, it shifts the phase transition toward lower temperatures. Consequently, the 7 Li NMR 1/ T 1 (1/T) rates pass into the low-T flank of the rate peak, which characterizes translational Li + , dynamics already at temperatures equal to or larger than 340 K. In agreement with faster Li + diffusion in LiBH 4 -LiI, the slope of the low-temperature flank of the rate peak yields an activation energy E a of 0.36 eV (Figure 2a) instead of 0.5 eV for LiBH 4 . 9 This comparison shows that LiI does not only stabilize the hexagonal form at lower T but also reduces the mean activation barrier for Li + translational diffusion as it is seen by NMR (Figure 2a).
Apart from the change of the rates above T pt , we recognize that also the rotational jump processes change when going from microcrystalline LiBH 4 to nanocrystalline LiBH 4 -LiI. The original, overall maximum located at 220 K shifted toward a much lower temperature (Figure 2a), indicating an increase in the corresponding motional correlation rate sensed by the 7 Li spins. We assume that this increase is a direct consequence of the expanded lattice through the incorporation of I − having a larger radius than BH 4 − . A deconvolution of this response into two rate peaks turned out to be no longer possible; for the LiIcontaining sample LiBH 4 -LiI, the former two rate peaks P1 and P2 (see above) merge into a much broader peak located at T ≈ 190 K. Most likely, this change originates from a broader distribution of rotational jump rates in LiBH 4 -LiI. The shift toward lower temperatures agrees with a reduction of the activation energies E a associated with the BH 4 − rotational jumps. Here, E a decreases from 0.26 to 0.12 eV if we consider the high-T flank of the 7 Li NMR rate peaks just below T pt (see Figure 2a). The introduction of LiI does also reduce the overall NMR coupling constant determining the maximum rates at T = 190 K. Figure 2b shows the NMR responses of the two nanoconfined samples, LiBH 4 -LiI/Al 2 O 3 and LiBH 4 /Al 2 O 3 . Starting from low temperatures, we recognize that the rates evolve in a similar manner to those of LiBH 4 and LiBH 4 -LiI, respectively. However, especially for LiBH 4 /Al 2 O 3 , without any LiI incorporated, we notice enhanced BH 4 − rotational dynamics compared to the microcrystalline reference sample LiBH 4 having no contact to any insulator phase. For LiBH 4 -LiI/ Al 2 O 3 and LiBH 4 -LiI, the 7 Li response turned out to be rather similar, while variable-temperature 1 H NMR SLR measurements ( Figure S4) showed some subtle differences in this temperature regime (see the Supporting Information). As suggested by Figures S4 and 2b, rotational ion dynamics in or near the conductor−insulator interfacial regions are enhanced for LiBH 4 (-LiI)/Al 2 O 3 compared to those in the bulk regions of LiBH 4 .
Most importantly, the largest effect of the insulating phase on 7 Li NMR spin-lattice relaxation is seen at higher temperatures when Li + translational ion dynamics start to govern the spin fluctuations (Figure 2b). In contrast to the sample without any Al 2 O 3 , we clearly recognize that the 7 Li NMR rates start to increase at temperatures as low as 240 and 270 K, respectively. These temperatures are clearly lower than T pt = 340 K for LiBH 4 -LiI (see Figure 2a). At temperatures slightly above 270 K, we recognize that the LiI-free sample does almost show the same NMR SLR response as seen for LiBH 4 -LiI/Al 2 O 3 , which unequivocally reveals that the interface effect is the main reason for the longitudinal recovery of the magnetization mirroring Li + diffusivity.
Here, this effect turned out to be much larger than that seen for LiBH 4 /Al 2 O 3 composites that were earlier prepared by (high-energy) ball milling. 13 Melt infiltration leaves behind a defective LiBH 4 phase, and we assume tight conductor− insulator contacts. The nanoconfined samples provide a large fraction of these heterocontacts. Our comparative NMR results clearly show that the interfacial regions play a dominant role in explaining enhanced ion dynamics in the nanoconfined samples regardless of whether LiI is present or not. The latter finding is supported by recent calculations revealing that the poor ion transport in orthorhombic LiBH 4 originates from very high defect formation energies. 8 The LiBH 4 /Al 2 O 3 zones are, however, expected to be rich in defects, thus facilitating ion transport. 33 A similar effect has been described very recently by first-principles calculations for the interface in LiBH 4 /MoS 2 composites. 44 Importantly, in LiBH 4 /SiO 2 composites, the role of surface groups should not be underestimated. 16,17 Surface The Journal of Physical Chemistry C pubs.acs.org/JPCC Article effects are also important for LiBH 4 /Al 2 O 3 : as has been shown quite recently by 27 Al NMR, 33 penta-coordinated Al centers Al IV get saturated while generating Al IV BH 4 − −Li + , forming a defect-rich zone with vacant or interstitial Li + sites. The same mechanism has also been proposed by some of us for the recently studied LiF/Al 2 O 3 nanocrystalline composites. 45 Up to 330 K, the 7 Li NMR rates of the two nanoconfined Al 2 O 3 -containing samples (see Figure 2b) follow linear behavior. The associated activation energies turn out to be rather low and take values in the order of only 0.1 eV. This average value mirrors a flat potential landscape and is even lower than that of nanocrystalline LiBH 4 (0.18 eV) prepared by ball milling. 46 Values in the order of 0.07 eV were also observed indirectly by 1 H NMR SLR measurements above 380 K (see Figure S4, Supporting Information). Again, we ascribe the reduction in activation energy when going from nanostructured LiBH 4 to nanoconfined LiBH 4 /Al 2 O 3 to the interfacial "insulator effect" generating a percolation network of fast diffusion pathways for the Li + ions. Most likely, as detailed above, such a network benefits from defect-rich space charge regions that influence the Li + hopping processes.
The 1/T 1 rates of nanoconfined LiBH 4 /Al 2 O 3 level off near 373 K and pass into a region that is characterized by a very low activation energy (see arrow in Figure 2b). In this region, the phase transition from orthorhombic to hexagonal LiBH 4 is marked by anisothermic peaks in our DSC measurements (see Figure S2, Supporting Information). As the addition of LiI stabilizes the hexagonal phase at temperatures well above room temperature, the anisothermic peaks are virtually absent. Again, this comparison shows that it is not the crystal structure but rather the interaction with the Al 2 O 3 interfaces that governs ion dynamics in the interfacial regions of the nanoconfined composite samples below the phase transition temperature. Our results show how this effect can be used to turn a poor ion conductor, such as orthorhombic LiBH 4 , into a superior material with fast Li + exchange processes assisting in facile macroscopic ionic transport (see below). 7 Li NMR Line Shapes of the Nanoconfined Composites. While NMR spin-lattice relaxation rates, especially when probed in the temperature regime of the low-T flank of the given 1/T 1 (1/T) rate peak, are sensitive to the elementary hopping processes, NMR line shapes, which are governed by spin-spin-relaxation rates, can be used to probe Li + transport on longer length scales. To see whether and to which extent the conductor−insulator effect does also affect the corresponding 7 Li NMR lines, we recorded variable-temperature spectra of the reference sample LiBH 4 -LiI (see Figure 3a) and the two nanoconfined samples (see Figure 3b,c). Figure 3a shows the 7 Li NMR lines of nanocrystalline LiBH 4 -LiI. Starting from a broad signal at 173 K, which reveals sluggish Li + translational ion dynamics in the bulk regions, the line undergoes heterogeneous motional narrowing upon heating. At temperatures above 294 K, a narrow line superimposes the broader Gaussian-shaped main signal. We attribute the narrowed line to Li + ions in the interfacial regions of this nanocrystalline sample. These regions offer fast Li + diffusion pathways as has recently been shown for nanocrystalline, orthorhombic LiBH 4 . 46 The line recorded at 433 K reflects Li + -ion dynamics in hexagonal LiBH 4 -LiI. At this temperature, all Li + ions take part in rapid exchange processes. The quadrupolar satellite signals seen at ±8 kHz represent the 90°singularities of the powder pattern that is diagnostic for For LiBH 4 -LiI heterogeneous motional narrowing sets in at 294 K, stepwise narrowing is ascribed to two spin reservoirs with one of them representing the much less mobile Li + ions in the bulk regions crystallizing with orthorhombic structure at very low T. As LiI stabilizes the hexagonal form well above room temperature, the narrow line at elevated T reflects Li + ions in the hexagonal phase. For LiBH 4 -LiI/Al 2 O 3 and even for LiBH 4 /Al 2 O 3 , the narrowing process is clearly shifted toward much lower temperatures of T < 253 K revealing that the conductor−Al 2 O 3 interfacial regions govern overall Li + translational motions sensed by the 7 Li NMR spectra. Compared to the spectra shown in (a), a large fraction of Li ions benefits from this insulator or interface effect. The magnified spectrum in (c) shows the quadrupole powder pattern of LiBH 4 /Al 2 O 3 . Dashed (vertical) lines refer to the position of the quadrupole singularities on the kHz scale; see text for further details.
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article this sample. In general, quadrupole intensities mirror the interaction between the electric quadrupolar moment of the 7 Li nucleus (spin-quantum number I = 3/2) with a nonvanishing electric field gradient (EFG) at the nuclear site. This interaction alters the Zeeman levels such that for I = 3/2, four inequivalent levels are generated that depend on the crystallite orientation in the external magnetic field. Assuming an (averaged) axially symmetric EFG at the nuclear sites, the corresponding quadrupolar coupling constant C q of the powder sample is given by C q = 32 kHz. Nanoconfinement, i.e., the introduction of conductor− insulator interfacial regions, ensures that (heterogeneous) motional narrowing does already set in at temperatures lower than 250 K. This temperature agrees with the temperature at which the 7 Li NMR rates start to increase. Satellite singularities come into the picture at 313 K. Already at 294 K, approximately 50% of the Li + ions in LiBH 4 -LiI/Al 2 O 3 (see Figure 3b) have access to fast diffusion pathways as is indicated by the ratio of the area fractions of the broad and the narrow NMR lines. For LiBH 4 /Al 2 O 3 , the area fraction of the narrow line amounts to ca. 30% at 294 K (see Figure 3c).
As discussed earlier, 31,33 the overall coupling constant C q (≈ 20 kHz) turned out to be clearly reduced compared to that of LiBH 4 -LiI. Importantly, for LiBH 4 /Al 2 O 3 , almost the same line shapes are detected as for the LiBH 4 -LiI/Al 2 O 3 . Again, this result demonstrates the leading role of Al 2 O 3 in governing the 7 Li NMR signals. Motional narrowing is slightly shifted toward higher T, which is in excellent agreement with the temperature behavior of the 7 Li NMR rates. The corresponding coupling constant C q = 23 kHz of LiBH 4 -Al 2 O 3 resembles that of LiBH 4 -LiI/Al 2 O 3 ; see also the magnified spectrum recorded at 433 K ( Figure 3c). It shows that nanoconfinement is also responsible for the electric quadrupole interactions and the majority of Li spins are subjected to in or near the conductor−insulator interfacial regions. As the pore size of Al 2 O 3 is less than 10 nm, as has been reported earlier, 31 bulk regions, if confined to such small cages, are obviously also affected by the insulating surface regions. 31,33 The magnification of the spectrum recorded at 433 K (see Figure 3c) shows an additional pair of satellite regions, which we earlier ascribed to the Li ions farther away from the interface regions. The corresponding coupling constant of ca. 37 kHz agrees well with that which is obtained for pure LiBH 4 at this temperature. Hence, NMR is able to reveal the different electrical interactions the spins are sensing in nanoconfined LiBH 4 /Al 2 O 3 , with most of them being subjected to the insulator surface interactions and some residing in the smaller bulk areas. 31 As mentioned above, this view is also corroborated by the NMR central lines shown in Figure 3.
Importantly, fast spin diffusion connecting the two spin reservoirs in the nanoconfined samples causes single exponential 7 Li NMR T 1 magnetization transients; thus, a separation of the two spin ensembles, as it was possible earlier for high-energy ball-milled LiBH 4 , is almost impossible if we use the longitudinal transients for this purpose (see Figure S2). Moreover, at a given temperature, the associated 1/T 1 7 Li NMR rates of the fast and slowly decaying part of the underlying free indication decays do almost coincide. Hence, we conclude that for the nanometer-sized architecture in the conductor−insulator composites, rather efficient spin diffusion is present. Hence, from the point of view of SLR NMR, LiBH 4 in LiBH 4 /Al 2 O 3 appears as an almost homogeneous phase. Noteworthy, while NMR is able to monitor the fast Li + exchange processes in the interfacial regions of the nanocomposites, it is, in the present case, less sensitive to longrange ion transport in LiBH 4 -(LiI)/Al 2 O 3 . While the two samples LiBH 4 -LiI/Al 2 O 3 and LiBH 4 /Al 2 O 3 show almost the same 7 Li NMR response, through-going ionic transport in LiBH 4 -LiI/Al 2 O 3 is easier (ca. 1.3 × 10 −4 S cm −1 (298 K)) 31 than in LiBH 4 /Al 2 O 3 (0.3 × 10 −4 S cm −1 (298 K)). 31 Most likely, this difference originates from the orthorhombic bulk regions in the latter compound that hinder Li + ion dynamics; see the schematic illustration in Figure 4 that summarizes the findings. A similar picture has been proposed for other dispersed ion conductors, such as nanocrystalline Li 2 O/Al 2 O 3 composites, whose heterogeneous transport properties were explained by the percolation concept. 47,48 Here, for both compounds, the LiBH 4 /Al 2 O 3 interface (heterocontacts) provides fast Li + diffusion pathways. While at low temperatures Li + diffusion in the orthorhombic bulk regions of LiBH 4 in LiBH 4 /Al 2 O 3 is slow, and only enhanced at the LiBH 4 /LiBH 4 homocontacts, anion substitution in LiBH 4 -LiI/Al 2 O 3 additionally ensures fast Li + self-diffusivity in the hexagonal bulk regions that do not benefit from interactions with the surface regions. Therefore, the combination of nanoconfinement and anion substitution enables facile, overall Li + long-range ion transport as it is necessary for, e.g., battery applications. 31,33 ■ CONCLUSIONS Using LiBH 4 and nanoconfined LiBH 4 -LiI as model systems, we investigated the influence of conductor−insulator inter-  7 Li NMR is able to trace rapid fast Li + self-diffusivity along (or near) the interfacial pathways generated by the conductor−insulator heterocontacts. However, if no percolation pathways are formed, the poorly conducting orthorhombic phase hinders long-range (through-going) Li + ion transport (see (a)), whereas for LiBH 4 -LiI/Al 2 O 3 , fast 2D Li + diffusion in the LiBH 4 -LiI bulk regions 33 ensures facile long-range cation motions. As ionic conductivity depends on both the mobility μ and the charge carrier density N, the combination of anion substitution and nanoconfinement is a perfect tool to tune overall Li + transport. 31,33 The Journal of Physical Chemistry C pubs.acs.org/JPCC Article facial regions on the overall Li + translational ion dynamics, which we sensed by 7 Li NMR spin fluctuations. Al 2 O 3 served as an insulating phase, that is, the phase usually blocking Li +ion transport. As irregular diffusive motions trigger (longitudinal) 7 Li NMR spin-lattice relaxation, with the help of variable-temperature measurements, activation energies and motional correlation rates can be probed. While in nanoconfined LiBH 4 -LiI rapid Li + translational motions (0.36 eV) influence the 7 Li NMR rates 1/T 1 at temperatures above 340 K, in the Al 2 O 3 -bearing nanoconfined samples, a drastic change in overall NMR response is seen. For LiBH 4 -LiI/Al 2 O 3 , the low-T flank of the corresponding diffusion-induced rate peak 1/T 1 (1/T) is already seen at a temperature as low as 240 K, thus shifted by 100 K toward lower temperatures. Surprisingly, above 270 K, the same flank is also seen for the LiI-free sample unequivocally showing that the conductor−insulator (Al 2 O 3 ) effect has the authoritative role to explain the enhanced Li + -ion dynamics in these samples. Clearly, reduced activation energies in the order of only 0.1 eV agree with the low onset temperatures and underpin the idea of a percolation network of fast diffusion pathways generated by the conductor−insulator interfacial regions. 7 Li NMR line shape measurements corroborate the results from 7 Li NMR relaxometry and reveal, for both samples LiBH 4 -LiI/Al 2 O 3 and LiBH 4 /Al 2 O 3 , an ensemble of mobile Li + ions being subjected to rapid diffusive motions already at temperatures well below ambient. 7 Li NMR quadrupole interactions seen in the spectra of nanoconfined LiBH 4 / Al 2 O 3 do reflect both bulk and interfacial regions, with the spins in the latter areas being highly mobile and benefiting from the interaction with the insulator surface. Our work highlights the importance of conductor−insulator interfacial regions in advanced solid electrolyte research. The clever introduction of such artificial interfaces, influencing ion dynamics by both structural disorder and space charge regions, represents an adjustable tool to manipulate overall (Li + )ion dynamics in solids with nanometer-sized dimensions.