Combined Effects of Anion Substitution and Nanoconfinement on the Ionic Conductivity of Li-Based Complex Hydrides.

Solid-state electrolytes are crucial for the realization of safe and high capacity all-solid-state batteries. Lithium-containing complex hydrides represent a promising class of solid-state electrolytes, but they exhibit low ionic conductivities at room temperature. Ion substitution and nanoconfinement are the main strategies to overcome this challenge. Here, we report on the synthesis of nanoconfined anion-substituted complex hydrides in which the two strategies are effectively combined to achieve a profound increase in the ionic conductivities at ambient temperature. We show that the nanoconfinement of anion substituted LiBH4 (LiBH4-LiI and LiBH4-LiNH2) leads to an enhancement of the room temperature conductivity by a factor of 4 to 10 compared to nanoconfined LiBH4 and nonconfined LiBH4-LiI and LiBH4-LiNH2, concomitant with a lowered activation energy of 0.44 eV for Li-ion transport. Our work demonstrates that a combination of partial ion substitution and nanoconfinement is an effective strategy to boost the ionic conductivity of complex hydrides. The strategy could be applicable to other classes of solid-state electrolytes.

Introduction

Solid-state electrolytes are indispensable for the realization of safe batteries offering high energy densities[1,2] crucial for the development of both mobile applications and large-scale stationary systems that can effectively store electricity from renewable but intermittent energy sources such as solar, wind, or tidal. Current battery systems, especially those designed for electric vehicles, may suffer from flammable and volatile organic based liquid electrolytes. In many cases, this narrow electrochemical stability window of conventional aprotic electrolytes prevents the use of anode materials providing high energy densities like metallic lithium. These disadvantages have led to a renewed interest in inorganic solid-state electrolytes, because they are potentially safer than liquid electrolytes and chemically compatible with Li metal. In addition, if sulfur-based cathode materials are considered, they prevent the dissolution and shuttling of polysulfides, which is one of the most serious hurdles that needs to be overcome in these type of batteries that promise high energy densities.[3]

Various classes of materials have been investigated as solid-state ion conductors for all-solid-state batteries.[4−8] These materials include garnets,[9] perovskites,[10] and polymers[11] as well as glass type electrolytes.[12] The complex metal hydrides, particularly those containing Li and Na, such as LiBH4, Li2B12H10, NaB10H10, and NaCB11H12 constitute a relatively new class of solid electrolytes.[13−17] Due to their lightweight and high hydrogen content, these materials have been intensively investigated over the last 20 years for reversible hydrogen storage. The hydrogen could be used for fuel cells with polymer electrolyte membranes.[18−21] Over a decade ago, it was shown that they also exhibit fast ionic conductivity and some also show good electrochemical stability windows up to 3 V versus Li/Li+.[22−24]

The main challenge for solid-state electrolytes in general has been the inherently low room temperature ionic conductivity, compared to liquid electrolytes. Therefore, most research effort has been focused on enhancing the ionic conductivities by structural modifications. For instance, the high ionic conductivity in complex hydrides is often a result of structural phase transition at high temperatures. A typical example is LiBH4, which exhibits an ionic conductivity of 1 × 10–3 S cm–1 above 110 °C,[14,23] due to the formation of the hexagonal phase, whereas at room temperature, the compound crystallizes into the orthorhombic structure, which shows poor ion conductivity.[25−27]

Two main approaches, which have separately been introduced in literature, have proved to be successful in boosting the room temperature ionic conductivity of solid electrolytes, especially complex hydrides. The first approach takes advantage of the partial substitution of the complex anion (e.g., BH4– in LiBH4) by halides such as Cl– and I– or by amides.[28−30] Iso- or aliovalent replacement of the Li+ cations by Na+, Ca2+ or Ce3+ has been reported as well.[31−33] Partial ion substitution is generally achieved by high-energy ball milling[34] or by heating a physical mixture of the compounds at temperatures of 200 to 300 °C; note that the melting point of LiBH4 is 278 °C.[35] For instance, solid solutions of LiBH4–LiX (X = Cl, Br, I) and LiBH4–Li3N or new compounds like in LiBH4–LiNH2 with room temperature ionic conductivities that are much higher than the individual compounds have been reported. Several studies have been conducted to investigate the effect of heat treatment,[36] the influence of LiX contents,[35] and the influence of the kind of the substituting anion.[37]

It is generally believed that anion substitution leads to an increase in distance between neighboring BH4– units which is associated with weaker Coulomb interactions in LiBH4 and hence a decrease of the transition temperature at which the compound changes from orthorhombic to hexagonal symmetry.[38] Indeed, substitution with larger halide anions leads to stabilization of the hexagonal phase of LiBH4 at near ambient temperatures as clearly seen in the most investigated anion substituted complex hydride LiBH4–LiI.[14] Alternatively, treatment at elevated temperatures (150 °C) of LiBH4 together with LiNH2 leads to the formation of a new phase, Li2(BH4)(NH2). The formation of a new phase causes a conductivity enhancement and a relatively low phase transition temperature of approximately 50 °C.[30,39] A main disadvantage of this method is, however, that the resulting compounds suffer from poor electrochemical and thermal cycling stability because of phase segregation.

The second approach, which has led to a large increase in the ionic conductivity in several classes of solid electrolytes, is interface engineering by forming nanocomposites with oxides such as SiO2 and Al2O3.[40−43] This approach, especially when using carbon scaffolds, was originally proposed to enhance the hydrogen sorption properties of complex hydrides,[44−48] but it was additionally shown to influence the ion mobility in the materials as well. The increase in ionic conductivity is currently believed to be caused by interactions of the hydride with the scaffold surface leading to either interfacial space charge zones[49] or to the formation of highly conducting compounds at this interface[50] due to changes in structure or defect density. The exact nature of the hydride/oxide interface is still a subject of intensive investigation. Alternatively, it has been shown that nanocrystalline LiBH4 has an enhanced ionic conductivity compared to the microcrystalline form.[51] Interface engineering of LiBH4 is normally achieved by nanoconfinement, e.g., via melt infiltration[52] of LiBH4 in the nanopores of the oxides[40,42] or by ball milling[41−43] a mixture of LiBH4 and the metal oxide. The two preparation methods have been reported to lead to comparable effects. Results from testing all-solid-state Li–S batteries using LiBH4/SiO2 nanocomposites as electrolytes showed that this approach enhances the ionic conductivity of LiBH4 and leads to better electrochemical and cycling stability.[53−55]

It has been shown that in these so-called dispersed ionic conductors, only the conductor or electrolyte (e.g., LiBH4) near the interface with the oxide (within 1–2 nm) exhibits very high ion mobility at room temperature.[48,55] Nevertheless, the volume fraction far from the interface is crucial to achieve interconnected LiBH4 particles as the silica or alumina particles do not contribute to the ionic conductivity. Only a percolating network of fast Li+ diffusion pathways will guarantee facile Li ion transport over long distances. We hypothesize that the overall long-range ionic conductivity of the nanocomposite can be further improved if the conductive regions are interconnected via highly conductive Li+ diffusion pathways rather than just bulk LiBH4.

For this purpose, we prepared nanoconfined anion substituted LiBH4. In these nanocomposites, the combined effects of partial anion substitution, either with I– or NH2–, and nanoconfinement in metal oxides (SiO2 and Al2O3) indeed leads to high room temperature Li-ion conductivities. In agreement with this observed enhancement, the activation energy for Li ion transport is lower than those probed for nanoconfined LiBH4 and the unconfined anion-substituted (LiBH4–LiI, LiBH4–LiNH2) systems. By using different preparation methods, we show that the enhancement seen for nanoconfined LiBH4–LiI and LiBH4–LiNH2 is indeed due to the combined effects of interfacial interactions with the metal oxide surface groups and the presence of highly conducting anion-substituted LiBH4 located further away from the SiO2 or Al2O3 surfaces. LiBH4 was used as an excellent model system to demonstrate the effect of combining anion substitution and nanoconfinement, and we believe that this approach and outcome are applicable to a wide variety of solid-state electrolytes.

Experimental Section

Synthesis of Silica Supports

MCM-41 was synthesized using the procedure described by Cheng et al.[56] In brief, hexadecyltrimethylammonium bromide (Sigma-Aldrich, ≥96.0%) and tetramethylammonium hydroxide solution (Sigma-Aldrich, 25 wt % in H2O) were mixed with deionized water. After the addition of the silica source (Aerosil 380), the white suspension was stirred for 2 h at 30 °C and kept at this temperature for another 24 h unstirred in a closed polypropylene bottle. The composition of the mixture was 1.00 SiO2: 0.19 (TMA)OH: 0.27 (CTA)Br: 40 H2O. The jelly like product was heated to 140 °C in stainless steel autoclaves and kept there for 48 h. After being naturally cooled to room temperature, the mixture was thoroughly washed, filtered, and dried at 120 °C for approximately 12 h. The final calcination step (550 °C, 12 h) was carried out after heating the sample first to 100 °C for 1 h as an additional drying step.

SBA-15 was prepared according to Zhao et al.[57] Poly(ethylene glycol)-block-poly(propylene glycol)-block-poly(ethylene glycol) (Sigma-Aldrich, PEG–PPG-PEG, Pluronic, P-123), hydrochloric acid fuming 37% (Merck, for analysis), and deionized water were stirred at 35 °C. Tetraethyl orthosilicate (Sigma-Aldrich, ≥99.0% GC, TEOS) was added dropwise to the solution, and the solution was then stirred for 24 h at 40 °C resulting in a composition of 0.015 P123:5.2 HCl: 129 H2O: 1 TEOS. This mixture was kept at 100 °C in a closed polypropylene bottle for 48 h, followed by extensive washing and filtration. Subsequently, the product was predried (60 °C, 24 h, air), dried (120 °C, 8 h, air), and calcined (1.2 °C min–1, 550 °C, 6 h, air).

Electrolyte Preparation

Alumina (Al2O3) was purchased from Sasol (product brand Puralox SCCa-5/200), while lithium amide (95% pure), lithium iodide (98% pure), and lithium borohydride (95% pure) were purchased from Sigma-Aldrich. The metal oxide supports (MCM-41, SBA-15 and Al2O3) were first dried under a vacuum at 220 °C overnight; then we stored them in an Ar purified glovebox (MBraunLabmaster, typically H2O and O2 < 1 ppm). All subsequent sample handling and transfer were carried out in the glovebox to avoid contamination with air or traces of moisture. The LiBH4–LiI/oxide nanocomposites were prepared using two different methods. In the first method, LiBH4 and LiI were physically mixed in molar ratios of 10, 20, 30, and 40 mol % LiI with respect to LiBH4. Subsequently, the materials were mixed with the desired amount of the oxide and placed in a quartz reactor which was then inserted inside a stainless-steel high-pressure autoclave (Parr). The amounts were calculated in order to fill the oxide pores by 130%, meaning that all pores were filled and voids between particles and space between grains were filled additionally. Melt infiltration was carried out at 50 bar H2 pressure and a temperature of 295 °C for 30 min; the heating rate was approximately 3 °C min–1.[48] During this process, LiI-LiBH4 forms a solid solution ((1–x)LiBH4-xLiI with x = 0.1, 0.2, 0.3, 0.4) which melts and infiltrates the pores of the oxide. Upon cooling, the molten solid solution solidifies in the pores of the support material, and the excess amount remains at the external surface of the support.

In the second approach, the samples were prepared by combing solution impregnation and melt infiltration. A solution of LiI and water or ethanol was prepared. The desired amount of the solution was added dropwise, using a syringe and a septum, to the metal oxide support contained in a round-bottom flask. This was done outside the glovebox but by using a Schlenk line to avoid contamination. The impregnated oxide was kept at room temperature for 3 h, after which the solvent was removed. Subsequently, the mixture was dried at 250 °C overnight under a dynamic vacuum. In order to reach the desired amount of LiI in the pores, the procedure was repeated twice. The LiI/metal oxide nanocomposite was mixed with LiBH4 to reach a molar ratio LiI/LiBH4 of 20:80 with the volume of LiBH4–LiI corresponding to 130% of the total pore volume of the silica (or alumina). The mixture was then inserted into a sample holder placed inside a stainless-steel high-pressure autoclave, pressurized to 50 bar H2 and heated at 3 °C min–1 to 295 °C. The dwell time was 30 min. The molten LiBH4 infiltrates the oxide pores and reacts with the nanoconfined LiI to form LiBH4–LiI.

Reference samples of LiBH4–LiI solid solutions and nanoconfined LiBH4 were prepared under the same autoclave conditions as outlined above. Solid solutions were synthesized by heating mixtures of LiBH4 and LiI without adding the metal oxide support; nanoconfined LiBH4 was obtained without adding LiI to the mixture. A third reference sample was bulk LiBH4, which was ground and melted under the same autoclave conditions and recrystallized.

LiBH4–LiNH2/oxide nanocomposites were prepared using a two-step preparation method. First, LiBH4 and LiNH2 were physically mixed in a molar ratio of 50% LiNH2 with respect to LiBH4. Afterward, the physical mixture was placed in a stainless-steel reactor which was then inserted into a stainless-steel high-pressure autoclave. The solid-state reaction was carried out at 50 bar H2 pressure and at 150 °C (heating rate 2.5 °C min–1) for 30 min to form a solid solution with the composition 0.5LiBH4-0.5LiNH2. Subsequently, the solid solution was mixed with the desired amount of oxide in order to fill the pores by 130%. Melt infiltration was carried out at 50 bar H2 pressure at 120 °C (2.5 °C min–1) for 30 min. Upon cooling, the molten solid solution recrystallized in the pores of the support material to form nanoconfined LiBH4–LiNH2.

Characterization of Pristine Materials and Composites

X-ray diffraction was performed with a Bruker-AXS D-8 Advance X-ray diffractometer with Co Kα1,2 radiation (λ = 1.79026 Å). The samples were placed in an airtight sample holder, and diffractograms were recorded at room temperature covering a 2θ range of 10° to 100° for alumina-based samples and of 20 to 80° 2θ for the silica containing samples as well as the crystalline samples. The increment and scan duration per point was 0.12° 2θ and 4 s, respectively, for the alumina samples; 0.06° 2θ and 2 s, respectively, for the silica samples; and 0.03° 2θ and 1 s, respectively, for the crystalline samples. Rietveld refinement was carried out using the software X’PertHighScore Plus. A Le-Bail fit was applied to analyze the pattern; literature patterns of hexagonal LiBH4 taken from the Inorganic Crystal Structure Database served as reference. To refine the patterns, we used the lattice parameters of hexagonal LiBH4 as starting values (a = 4.28 Å, b = 4.28 Å, c = 6.98 Å).

Diffuse reflectance infrared Fourier transform spectra (DRIFTS) were obtained by a PerkinElmer 2000 spectrometer and a MCT detector. Sixteen scans were accumulated with a resolution of 4 cm–1 in the range of 500 to 4500 cm–1. An airtight sample holder (KBr background) guaranteed no air contamination during the measurements. Data acquisition was realized by recording absorbance versus wavenumber. The absorbance is directly converted to K-M units, introduced by Kubleka and Munk, which includes a scattering component and is, therefore, typically used for the analysis of powder samples.

Conductivity Measurement

Alternating current (AC) impedance spectroscopy measurements were performed using a Princeton Applied Research Parstat 2273. Lithium foil (Sigma-Aldrich, 99.9%, 0.38 mm thick and 12 mm in diameter leading to a surface of 1.33 cm2) was firmly placed on top of two 13 mm stainless steel dies. A 100–300 mg portion of the electrolyte was placed between the two lithium foils in a standard pellet die set. The sample was pressed using a pressure of 2 tons, resulting in a final electrolyte thickness of 1 to 2 mm, excluding the Li foil. With the weight of the samples, we calculated that the void fraction of the pellets is below 20%. The pressed sample pellet, which is tightly connected to the Li foils and stainless-steel dies, was placed in a custom-made impedance cell housed in a Büchi B-585 glass oven that was placed in an Ar-filled glovebox. The voltage amplitude of the AC signal was 1 V; we measured complex impedances over a frequency range from 1 MHz to 1 Hz. Generally, the pellets were heated from room temperature to 50 or 130 °C (depending on the sample), and then the samples were cooled down to room temperature. During this temperature cycle, impedance scans were acquired in steps of 5 or 10 °C. Before the acquisition of each scan, the measurement cell was allowed to equilibrate at the desired temperature for 45 min (while heating), 90 min (while cooling), and 150 min (for measurements at room temperature). The entire sequence was repeated for 2 to 3 cycles to investigate any hysteresis behavior and to detect any changes of the samples. Matlab and ZView software were used to fit the raw data by using Nyquist plots. A constant phase element (CPE) and a resistor connected in parallel were used as appropriate equivalent circuit to parametrize the data. Capacitances, C, were calculated according to C = R(1– × Q1/. R is the resistance in Ω, i.e., it denotes the real part of the complex impedance; Q has the numerical value of the admittance at ω = 1 rad s–1. n is a dimensionless variable characterizing the deviation of the CPE from the behavior of an ideal RC unit, which would yield n = 1.

NMR Line Shape Measurement

To underpin the findings by conductivity spectroscopy, we recorded 7Li (spin-3/2) nuclear magnetic resonance (NMR) spectra at a magnetic field of 7 T, corresponding to a Larmor frequency of 116 MHz, by employing a Bruker Advance III solid-state spectrometer. We used a standard broadband probe to acquire variable-temperature NMR spectra with a one pulse sequence under static, i.e., nonrotating conditions. The π/2 pulse length slightly depended on temperature and ranged from 2.1 to 2.3 μs. Such short pulse lengths ensured nonselective excitation of the whole spectra. Up to 16 scans were accumulated to form an average free induction decay, which, after Fourier transformation, yield the 7Li NMR spectra. The temperature in the sample chamber was monitored by a Eurotherm controller. Temperature adjustment was achieved, with an accuracy of ±2K, with a heater that was constantly flushed with a stream of dry nitrogen gas.

Results and Discussion

Structure of ((1–x)LiBH4-xLiI and Its Nanoconfined Counterpart As Seen by XRD and DRIFTS

First, we discuss the structural properties of (i) the LiBH4–LiI solid solutions ((1–x)LiBH4-xLiI) containing 10 to 40 mol % I (x = 0.1, 0.2, 0.3 and 0.4) and (ii) the nanocomposites with different oxides viz. γ-Al2O3, SBA-15, and MCM-41. The compositions of the samples in wt % are given in Table S1. Structural details of the hydrides, oxides, and nanocomposite materials are shown in Figure as well as in Table S2 and Figures S1 to S6. Figure shows the XRD powder pattern of a 20 mol % LiI-LiBH4 solid solution and the patterns of the nanocomposites prepared using two different routes, i.e., comelt infiltration and impregnation with LiI followed by melt infiltration with LiBH4. For comparison, the XRD pattern of LiI and the patterns of orthorhombic and hexagonal LiBH4 are also included. The influence of LiI on the on XRD patterns of the solid solutions and the composites is illustrated in Figures S7 and S8. The patterns shown here are normalized to the highest intensities; hkl values are added to distinct reflections of LiBH4[58,59] and LiI,[60−62] respectively. The XRD patterns of the LiBH4–LiI samples with 10 and 20 mol % LiI (Figure S7) clearly differ from those of orthorhombic LiBH4 and LiI. Instead they resemble the pattern of hexagonal LiBH4 being the stable phase at elevated temperatures. The reflections in the range from 27° to 32° 2θ are shifted toward lower 2θ values by approximately 1° 2θ. This shift reveals successful incorporation of LiI and is caused by lattice expansion because I– is larger than BH4–.[35] A similar shift has been reported in literature.[14] Rietveld refinement of the diffraction data for the 0.8LiBH4-0.2LiI solid solution yielded an hcp unit cell with the following lattice parameters a = 4.44 Å, b = 4.44 Å, and c = 7.19 Å. Simultaneously with lattice expansion, the density increased from 0.67 g/cm3 for bulk LiBH4 to 1.20 g/cm3 for 0.8LiBH4-0.2LiI. Samples with more than 20 mol % LiI revealed reflections of pure LiI indicating a solubility limit for the LiBH4–LiI system (Figure S7).

Figure 1

XRD powder patterns of the various LiBH4–LiI/oxide nanocomposites investigated. For comparison, the positions of the reflection of LiBH4 in its hexagonal form are included as well. In addition, the pattern of LiBH4–LiI (20 mol % of LiI) and LiI are also shown. Values in brackets refer to hkl indices. The shift of the reflections toward lower diffraction angles indicates successful incorporation of LiI that stabilizes the hexagonal form of LiBH4.

Confinement of the LiBH4–LiI solid solutions in the oxide nanopores led to both broadening and a decrease in intensity of the diffraction peaks (see Figures and S8). Peak broadening is expected because of size effects and lattice strain, whereas the decrease in intensity suggests a decrease in the long-range order. Note, however, that the composites shown here contain 30 vol % more LiBH4–LiI than is required to fill all the pores of the scaffold. This is essential for interconnectivity between the LiBH4 particles. For the nanocomposites with LiBH4–LiI ≤ 100% of the total pore volume of the scaffold, no crystalline phase was observed. Thus, we conclude that nanoconfinement led to a significant decrease in crystallinity of the samples. The similarity in the diffraction patterns of the nanocomposites prepared using the different methods (Figure ) suggests that both methods are useful for the preparation of LiBH4–LiI/metal oxide solutions. Moreover, the use of different oxide supports did not lead to major differences in the XRD patterns of the nanocomposites.

Further evidence for successful incorporation of the solid solutions into the oxide pores is provided by nitrogen physisorption measurements. The measurements showed that only a negligible amount of nitrogen was adsorbed by the nanocomposites. This finding proved that the pores were occupied by the electrolyte LiBH4–LiI.

DRIFTS was used to investigate the nature of chemical bonding in the different samples prepared. In Figure the spectrum of the LiBH4–LiI/Al2O3 nanocomposite is compared to spectra of bulk LiBH4, LiBH4–LiI, and pristine Al2O3. The spectra are presented in arbitrary K-M units (for further explanation see the Experimental Section). Macrocrystalline, that is, bulk LiBH4, shows characteristic bands between 1000 and 1500 cm–1 which correspond to [BH4]− bending vibrations; bands appearing in the range from 2000 to 2800 cm–1 can be associated with stretching vibrations in [BH4]−.[63,64] The most preeminent bands of LiBH4 are marked with dashed lines in gray color; the corresponding wavenumbers are indicated. The spectrum of the LiBH4–LiI solution resembles that of LiBH4, the band characterizing the stretching vibrations (2379 cm–1) is, however, slightly shifted toward a lower wavenumber. Most likely, this shift is due to the effect of negative chemical pressure resulting from an increased unit cell by addition of I, as previously observed for halide-substituted BH4.[65] Also, a stronger electronic interaction between iodine and Li, because of the higher electronegativity of the halides compared to BH4–, can lead to such a change in vibration frequencies.[66] For LiBH4–LiI/Al2O3, the bands are clearly broadened due to nanoconfinement. A similar broadening effect is also observed for LiBH4/Al2O3. It suggests that the structure of the confined materials is different from the bulk compound. This observation is in line with previous studies on nanoconfined LiBH4.[67−69]

Figure 2

DRIFT spectra of Al2O3, nanoconfined LiBH4–LiI/Al2O3, nanoconfined LiBH4/Al2O3, and LiBH4–LiI (20 mol % of LiI). For comparison, the spectrum of LiBH4 is also shown. Main peaks are marked by vertically drawn dashed lines with the wavenumbers indicated. K-M intensities (see the ordinate axis) are in arbitrary units. See text for further explanation.

Interestingly, the bands of Al2O3 in the region from 3400 to 3800 cm–1, representing OH surface groups,[70] almost disappear after the pores are filled with electrolyte; see the vertical arrows in Figure . The same behavior is found for the characteristic vibrations of the surface OH-silanol groups of silica[71,72] in the samples LiBH4/SiO2, LiI/SiO2, and LiBH4–LiI/SiO2 (see Figures S9 and S10). Interaction of LiBH4 with the surface is the origin of the high ionic conductivity of LiBH4/oxide nanocomposites.

Ionic Conductivity of Nanoconfined LiBH4–LiI

To evaluate the effects of different LiI concentrations on ionic conductivity, we recorded complex impedance data at different temperatures and analyzed the results in the Nyquist representation; see Figure . An overview of results from impedance spectroscopy of the LiBH4–LiI solid solutions and the nanocomposites is given in Table S3 and Figure S11. The overall ionic conductivity of the confined and pure solid solutions increased with increasing amounts of added LiI. At LiI contents higher than 20 mol %, the conductivity started to decrease (Figure S12). This is in line with results from XRD pointing to crystalline (unreacted) LiI above this compositional limit. Hence, the sample LiI-LiBH4 with 20 mol % LiI was chosen for a more detailed study.

Figure 3

(a) Nyquist plots, that is, the imaginary part, – Z″, of the complex impedance plotted versus the real part Z′, of nanoconfined LiBH4–LiI/Al2O3 and LiBH4/Al2O3. The LiBH4–LiI sample (20 mol % LiI) is also shown. Values in pF show the capacitances obtained after parametrizing the main (nondepressed) semicircles with the equivalent circuit shown; see also Experimental section. The line approximating the second semicircle of the curve belonging to LiBH4/Al2O3, which shows up at higher frequencies, is drawn to guide the eye. (b) Arrhenius plot (half-logarithmic plot of σ′ vs 1000/T) to illustrate the change of conductivity with increasing temperature. Dashed and solid lines represent linear fits to determine activation energies EA, which range from 0.44(1) eV to 0.59(1) eV. Nanoconfined LiBH4–LiI/Al2O3 shows the highest conductivities. At room temperature (25 °C), its ion conductivity is slightly larger than 10–4 S cm–1; a conductivity of 10–3 S cm–1, needed to realize Li-ion batteries, is reached at 66 °C.

Figure a shows the corresponding Nyquist plot recorded at 25 °C; in Figure b, the temperature dependence of the ionic conductivity is displayed using an Arrhenius plot. For comparison, the Nyquist plots and conductivity data referring to LiBH4–LiI and nanoconfined LiBH4 are also shown. Capacitances C ranged from 147 to 210 pF; values larger than 100 pF typically indicate electrical relaxation processes influenced by interfacial regions.[73] For nanoconfined composites, it is widely believed that ion transport mainly occurs along the heterogeneous solid–solid interphase, that is, at the interface between the insulating oxide and the electrolyte.[41−43] The exponents n turned out to take values close to 1 meaning that the corresponding CPEs (constant phase elements) of all three samples behaved almost like an ideal RC unit. This observation is in line with the interpretation that the semicircle seen in the complex plane plot is governed by a response strongly influenced by grain boundary effects.

For LiBH4/Al2O3, a second semicircle is seen at lower frequencies, which is either too small to be detected or is absent in LiBH4–LiI and LiBH4–LiI/Al2O3. The presence of the second semicircle suggests two different conducting phases. We attribute the main semicircle with the higher electrical relaxation rate to LiBH4 interacting with the oxide surface and the semicircle appearing at lower frequencies to LiBH4, which is farther away from the interface. This feature is also seen in 7Li NMR spectroscopy; see Figure . We suppose that the addition of LiI led to both an increase of the high conducting regions and an enhancement of interfacial conductivity. Thus, for LiBH4–LiI/Al2O3, the two contributions could not be resolved any longer when data recorded at 25 °C were analyzed.

Figure 4

7Li NMR spectra of (a) LiBH4–LiI, (b) nanoconfined LiBH4/Al2O3 without LiI, and (c) nanoconfined LiBH4–LiI/Al2O3. Spectra were recorded at a Larmor frequency of 116 MHz at the temperatures indicated. Dashed lines in parts a and b show the deconvolution of the entire line with appropriate Gaussian and Lorentzian functions to estimate the number fraction of mobile Li ions in these compounds. For LiBH4–LiI/Al2O3, the spectrum has almost adopted its final form at temperatures as low as 30 °C. While the sharp line represents fast Li ions, the broader foot comprises both the central line of a fraction of slower Li ions and quadrupole intensities. The latter become visible as a sharp powder pattern at elevated temperature where dipole–dipole interactions are effectively averaged out due to rapid Li+ exchange. See text for further information.

The Arrhenius plot shown in Figure b shows that ion transport at temperatures lower than 100 °C is clearly faster in the nanocomposites LiBH4–LiI/Al2O3 and LiBH4–LiI/SiO2 than in LiBH4–LiI and the nanoconfined samples LiBH4/Al2O3 and LiBH4/SiO2. For instance, at room temperature (25 °C), the ionic conductivity of LiBH4–LiI/Al2O3 (0.1 mS cm–1) was four times higher than that of LiBH4/Al2O3 and eight times higher than that of LiBH4–LiI. In agreement with the trend for the increase in ionic conductivity, the activation energy for long-range ion transport decreased from 0.52(1) eV for LiBH4/oxide to 0.44(1) eV for LiBH4–LiI/oxide. LiBH4–LiI showed a rather high activation energy of 0.59 eV. These values are similar to those reported in literature for LiBH4/Al2O3 and LiI-LiBH4 systems.[35,41]

7Li NMR line shapes of these samples, which have been recorded at room temperature and above, clearly revealed that Li+ acts as mobile charge carrier.[74] Selected lines are shown in Figure . For the LiBH4–LiI solid solution (see Figure a), the line at room temperature is composed of two contributions. The narrow line on top of the broader signal reflects the mobile Li spins whose jump rates exceed the line widths of this line in the rigid lattice, which turns out to be approximately 13 kHz. Narrow NMR lines are caused by sufficiently fast Li+ exchange processes able to average local dipole–dipole interactions that lead to line broadening at low temperatures. In the case of LiBH4–LiI, the line shape did not change much when going to 30 °C; however, a significant change was seen at 90 °C where a fully narrowed central line appeared, that is, on top of a quadrupole powder pattern. This distinct pattern, showing sharp 90° singularities separated by Δ = 15.6 kHz, is characteristic for hexagonal LiBH4(-LiI). A similar situation is seen for nanoconfined LiBH4/Al2O3 (see Figure b). However, the number fraction of rapid Li+ ions was higher at 22 and 30 °C (24%) compared to that seen for nonconfined LiBH4–LiI. This difference is in line with the slightly higher conductivity seen for LiBH4/Al2O3. It is worth noting that the motionally narrowed spectra recorded at 90 °C and at 120 °C were governed by electric quadrupole intensities being different than those of bulk LiBH4 and bulk LiBH4–LiI. The spectra of nonconfined LiBH4 and nonconfined LiBH4–LiI reveal patterns produced by a symmetric electric field gradient (EFG) the ions were subjected to. They agree with those of similar systems studied earlier.[40]

In contrast to the nonconfined samples, the NMR line of nanoconfined LiBH4/Al2O3 recorded at 90 °C shows a nonsymmetric EFG. Its shape points to structural disorder and strain which the Li spins sense. Δ reduces from 15.6 to 11.5 kHz. Careful inspection of the powder pattern shows that another set of singularities is present (see inset of Figure b), which is characterized by Δ = 18.3 kHz. Assuming axial symmetry for this pattern, we obtained a quadrupole coupling constant δq of ca. 36.6 kHz which was identical to that of bulk LiBH4 (δq = 37 kHz).[75] The two quadrupole patterns represent the Li ions near the insulator surface (Δ = 11.5 kHz) and the ions farther away, that is, located in the bulk regions (Δ = 18.3 kHz). NMR revealed that these two species are exposed to different electric interactions. Two sources of electrical relaxation have also been seen in the corresponding Nyquist plot, vide supra.

For nanoconfined LiBH4–LiI/Al2O3 (Figure c), we also observed a quadrupole powder pattern that is characterized by a lower Δ (= 9.5 kHz) than that expected for bulk LiBH4(-LiI). However, a pronounced pattern attributable to Li ions in bulk LiBH4–LiI, as seen for LiBH4/Al2O3 ,was missing. Instead, already at temperatures as low as 30 °C, an almost fully narrowed 7Li NMR line was observed which clearly points to very fast ion dynamics in this nanocomposite.[74] We conclude that the majority of ions in this nanocomposite take part in rapid Li+ exchange, which perfectly agrees with the conductivity trend seen in Figure b. From a structural point of view, the single EFG pattern observed points to a homogeneous sample as compared to nanoconfined LiBH4/Al2O3. Presumably, if the ions reside in areas farther away from the surface of the oxide, they are subjected to a structurally stressed LiBH4–LiI phase with high ionic conductivity. This modified region, e.g., influenced by space charge zones, regions with higher defect density or increased structural disorder, may extend over almost the whole LiBH4–LiI phase leading to the enhancement in conductivity observed.

Note that all our samples were prepared under the same conditions and, therefore, the remarkable increase in ionic conduction for LiBH4–LiI/Al2O3 seen by impedance spectroscopy and 7Li NMR is mainly attributed to the combined effects of anion substitution and interface engineering by nanoconfinement. Table compares conductivities, activation energies, and Arrhenius prefactors of the samples investigated. The slight differences in ionic conductivity of the nanocomposites prepared with different oxides (SBA-15, MCM-41, or Al2O3) are most likely due to differences in properties of these materials (see Figures S1 to S6 and Table S2). For example, the oxides differ in morphology, pore size, and pore size distribution, surface area, surface/interface energy, density of the surface groups, and the nature of the pores (e.g., pore corrugations). Detailed elucidation of the exact influence of these properties on ionic conductivity is, however, beyond the scope of the present work.

Table 1

Room Temperature Conductivities (σ) of the Samples Studied by Impedance Spectroscopyd

sample	σ(25 °C) (S cm–1)	EA(eV)	log10(A) (S cm−1K)
LiBH4/MCM-41	2.29 × 10–5	0.49(2)	6.0(3)
LiBH4–LiI/MCM-41 comelt infiltration	3.86 × 10–5	0.43(1)	5.3(1)
LiBH4–LiI/MCM-41 impregnation (H2O)	1.63 × 10–5	0.52(2)	6.5(3)
LiBH4–LiI/MCM-41 impregnation (EtOH)	4.57 × 10–6	0.47(0)	5.2(1)
LiBH4–LiI/SBA-15 comelt infiltration	1.29 × 10–4	0.44(1)	6.0(2)
LiBH4–LiI/Al2O3 comelt infiltration	1.27 × 10–4	0.44(1)	6.1(1)
LiBH4–LiI	1.54 × 10–5	0.59(2)	7.8(3)
LiBH4–LiNH2	2.92 × 10–6	1.03(1)a 0.19(1)b	1.8(1)a 13(1)b
LiBH4–LiNH2/ MCM-41	1.16 × 10–4	0.43(1)c	5.5(2)

EA determined in the temperature range from 30 to 50 °C.

EA determined in the temperature range from 60 to 85 °C.

EA determined in the temperature range from 30 to 85 °C.

The table also includes activation energies (EA) and pre-factors (log10(A)) of the Arrhenius laws used to approximate the temperature dependence of the ionic conductivity. If not stated otherwise, EA has been determined in the temperature range from 25 to 130°C.

Importance of LiBH4(-LiI)/Oxide Interface

To further demonstrate that both the interaction of LiBH4 with the oxide interface and partial anion substitution are important for the enhancement in ionic conductivity, we employed a preparation technique that is supposed to hinder the interaction of LiBH4 with the oxide interface but still form LiBH4–LiI in the pores. This comparison showed that the nanocomposites prepared by coinfiltration of a physical mixture of LiBH4 and LiI exhibited much higher conductivities than those which were prepared via solution impregnation (Table ). If we first add LiI to fill the pores via impregnation with LiI/H2O or LiI/C2H5OH solution and then add LiBH4 by melt infiltration as a second step, we see that the resulting ionic conductivity is significantly lower. At first glance, this difference is surprising as results from XRD and IR (DRIFTS) suggest that both samples have similar structures (cf. Figure and Figure S9). We attribute the marked change seen in conductivity to the fact that if LiI is added first, we do not have the original SiOH groups present at the interface anymore (see Figure S9 for the loss of silanol groups in LiI/SiO2). This changed the properties of the interface with the LiBH4, and as a result, the conductivity is not as high as that with the other preparation technique.

LiBH4–LiNH2 System

To demonstrate the general applicability of the strategy outlined above, we measured also the conductivity of another nanoconfined electrolyte containing two complex anions, i.e., nanoconfined LiBH4–LiNH2. XRD revealed the formation of two new phases,namel, Li2(BH4)(NH2) and Li4(BH4)(NH2)3 (see Figure a). This is unlike the LiBH4–LiX systems where the high temperature phase of LiBH4 was stabilized through the replacement of BH4– by halides causing lattice strain but no change in crystal structure. For nanoconfined LiBH4–LiNH2/MCM-41, XRD points to a loss of crystallinity, that is, long-range order. As mentioned above, the same feature was observed for the LiBH4–LiI solid solutions. In addition, results from DRIFTS measurements (see Figure b) revealed that the characteristic vibrations related to LiBH4–LiNH2, (1000 to 1500 cm–1 and 2000 to 2800 cm–1 (BH4–), 1500 to 1600 cm–1 and 3200 to 3300 cm–1 (NH2–), shifted toward lower wavenumbers and became significantly broader upon nanoconfinement. The bands related to the surface silanol groups (3700 cm–1) were absent, as seen for nanoconfined LiBH4–LiI. Hence, we conclude that the LiBH4–LiNH2 composite was successfully infiltrated into the nanopores of MCM-41 leading to profound changes of its structure.[66]

Figure 5

(a) X-ray powder diffraction patterns of nanoconfined and nonconfined LiBH4–LiNH2. For comparison, the expected patterns of orthorhombic LiBH4 and LiNH2 are also shown. The pattern at the top represents that of the oxide substrate, SiO2. (b) DRIFT spectra of the samples shown in part a; the spectra reveal broadening of the signals, which shift toward lower wavenumbers upon nanoconfinement. Those bands which results from silanol OH groups are absent for LiBH4–LiNH2/SiO2 indicating surface reactions between the electrolyte and the surface of the oxide. See text for further explanation.

In Figure , the ionic conductivities of selected LiBH4–LiNH2 samples are shown. First, when compared to LiBH4, it is clear that the addition of LiNH2 to LiBH4 increases the room temperature ionic conductivity by approximately 2 orders of magnitude. This increase is ascribed to the formation of Li2(BH4)(NH2).[39] The sudden increase in conductivity of Li2(BH4)(NH2) at approximately 35 °C originates from a structural phase change leading to a highly conducting phase at temperature higher than 40 °C. Nanoconfined LiBH4–LiNH2/MCM-41 showed an even better ionic conductivity at this temperature; remarkably, this high ionic conductivity was also preserved at lower temperatures. When compared to LiBH4 and LiBH4–LiNH2, the room temperature ionic conductivity of nanoconfined LiBH4–LiNH2/MCM-41 was higher by 4 and 2 orders of magnitude, respectively. It also exceeded that of nanoconfined LiBH4/MCM-41 by a factor of 2 if conductivities at T = 30 °C were considered (cf. Figure ). At approximately 50 °C, LiBH4–LiNH2/MCM-41 reached a conductivity of 1 mS cm–1. Below 45 °C, the overall activation energy governing ion transport in LiBH4–LiNH2/MCM-41 (0.43 eV) is comparable to that of bulk LiBH4 and significantly lower than that of LiBH4–LiNH2 at room temperature. For LiBH4/MCM-41 and LiBH4–LiNH2 at higher temperatures, we see that EA is somewhat lower, 0.26 and 0.19 eV, respectively; see Figure and Table . On the basis of the results from the DRIFTS measurements, the remarkable increase in ionic conductivity is again attributed to the combined effect of anion substitution and interface effects, as observed for nanoconfined LiBH4–LiI. These results illustrate that the synergistic effects of nanoconfinement, that is, interface engineering, and partial ion substitution is applicable to different Li-based electrolytes in various nonconducting nanoporous scaffolds.

Figure 6

Ionic conductivity of nanoconfined LiBH4–LiNH2/SiO2 as a function of the inverse temperature. For comparison, data on LiBH4/SiO2, nonconfined LiBH4–LiNH2 and bulk LiBH4 are also included. The lines are to guide the eye.

Conclusion

We have shown how two routes, namely, ion substitution and interface engineering, can be effectively combined to enhance the ionic conductivity of solid-state electrolytes. Using complex hydrides as model systems, we developed an approach where anion substituted LiBH4 (Li2(BH4)I1– and Li2(BH4)(NH2)1–) are confined in nanoporous SiO2 or Al2O3 in order to exploit both the effect of ion substitution and nanoconfinement or interface engineering to boost the Li-ion conductivities of LiBH4 at ambient conditions. Indeed, the ionic conductivity of the nanocomposites of LiBH4–LiI/Al2O3 reached 0.1 mS cm–1 at room temperature. The room temperature conductivities of nonsubstituted LiBH4/Al2O3 and LiBH4–LiI without nanoconfinement were 1 order of magnitude lower. Activation energies are in line with this trend, with 0.44, 0.52, and 0.59 eV for the LiBH4–LiI/Al2O3, LiBH4/Al2O3, and LiBH4–LiI, respectively. Detailed structural investigations and 7Li NMR line shape measurements show that the combined effects of interaction with the interface of the oxides and phase stabilization due to partial anion substitution (by the iodide anion) produces faster Li+ diffusion pathways in LiBH4–LiI/oxide than those in the case of LiBH4/oxide and LiBH4–LiI. Results on LiBH4–LiNH2 confined in mesoporous silica (MCM-41) show that this concept is also applicable to other Li-bearing hydrides. The enhancement effect depends also on the type and property of the scaffold. Our study clearly shows that combining partial anion substitution and nanoconfinement is a very promising approach to achieve high room temperature ionic conductivities in solid-state ion conductors.