Structural investigationSpecific surface area, pore-size distributions, and pore volume of the SBA-15 scaffold were determined by the N2 adsorption-desorption technique to be 825 m2 g−1, 5.8 nm, and 0.77 cm3 g−1, respectively (see Supplementary Fig. 1). Rietveld refinement of PXD data reveal that the as-prepared composite contained 47.7 wt% (53.1 mol%) of Mg(BH4)2·NH3 and 52.3 wt% (46.9 mol%) of Mg(BH4)2·2NH3, resulting in the average composition Mg(BH4)2·1.47NH3 (see Supplementary Fig. 2). Melt infiltrated (MI) samples are denoted as MI100, MI200 and MI300, as a reference to their calculated degree of pore filling, i.e. 100, 200 and 300% (see Table 1). However, a specific surface area and total pore volume of 24.1 m2/g and 0.09 cm3/g are observed for the MI100, suggesting that the degree of pore-filling is only 88%. For MI200 and MI300, the specific surface area and total pore volume are negligible, implying that the pores are either filled or blocked (see Supplementary Fig. 3 and Supplementary Table 1). Powder X-ray diffraction (PXD) data of the confined samples are shown in Fig. 1a. The PXD data of the Mg(BH4)2·1.47NH3 and SBA-15 are also presented in Fig. 1a for comparison. The lack of Bragg peaks in the diffraction pattern of MI100 suggests that the Mg(BH4)2·1.47NH3 composite is confined in the pores of SBA-15. This is also supported by the change in the background scattering of sample MI100 and SBA-15, which suggests the presence of additional amorphous material in the former. In the case of MI200 and MI300, diffraction peaks corresponding to Mg(BH4)2·NH3 are observed, suggesting a recrystallization of excess Mg(BH4)2·NH3 on the surface of SBA-15 upon melt infiltration, while the eutectic composition Mg(BH4)2·1.5NH3 appear to be stabilized also on the outer surface of the SBA-15 nanoparticles. After 5 months of storage, small amounts of Mg(BH4)2·NH3 appear to recrystallize from the MI200 and MI300 samples, while small amounts of both Mg(BH4)2·NH3 and Mg(BH4)2·2NH3 recrystallize from MI100, see Fig. 1b.Table 1 Pore filling degreeFig. 1: Structural characterization.Powder X-ray diffraction diagrams of the composite, the scaffold and the nanoconfined samples, a after synthesis and b after 5 months of storage at RT under argon atmosphere.Solid-state 11B magic-angle spinning (MAS) NMR spectra of Mg(BH4)2 ∙ 1.47NH3 as well as the nanoconfined samples were collected after 5 months of storage, see Fig. 2. The composite Mg(BH4)2 ∙ 1.47NH3, reveal three different boron resonances at −40.5, −41.5, and −42.1 ppm. This observation indicates a physical mixture of Mg(BH4)2 ∙ NH3 and Mg(BH4)2 ∙ 2NH3, which contain one (−40.5 ppm) and two (−41.5 and −42.1 ppm) distinct boron sites, respectively, in agreement with earlier studies13,26. The resonances can be satisfactorily simulated with Lorentz-shaped peaks, from which a full width at half maximum (FWHM) linewidth of 1.08 ± 0.05 ppm is estimated for the tallest resonance at −40.5 ppm from Mg(BH4)2 ∙ NH3. For all confined samples, the sharp peak at −40.3 ppm with a Lorentzian line shape indicates the presence of a high degree of boron dynamics in the eutectic molten state, suggesting the stabilization of the molten state as seen in previous studies13.Fig. 2: Boron chemical environment.11B MAS NMR spectra, illustrating the central-transition region, for the composite and the nanoconfined samples.The MI100 sample (FWHM = 0.66 ± 0.05 ppm for the −40.3 ppm resonance) is estimated to consists of 37 mol% Mg(BH4)2 ∙ 2NH3 and 63 mol% of a mixture of nanoconfined material and Mg(BH4)2 ∙ NH3. It should be noted that it was not possible to separate contributions from the nanoconfined material and Mg(BH4)2 ∙ NH3 in MI100, due to the significant overlap of the resonances from nanoconfined material, Mg(BH4)2 ∙ NH3 and Mg(BH4)2 ∙ 2NH3. However, the presence of Mg(BH4)2 ∙ NH3 in MI100-300 can be confirmed by the PXD data (see Fig. 1b). In contrast, the sharp peak (at −40.3 ppm) for MI200 and MI300 can be simulated using two Lorentzian peaks, one with FWHM = 1.08 ppm (Mg(BH4)2 ∙ NH3) and the other with FWHM = 0.62 ± 0.05 ppm (MI200) or FWHM = 0.67 ± 0.05 ppm (MI300). The intensities from these simulations indicate that MI200 consists of 72 mol% nanoconfined material in the non-crystalline “eutectic molten state” and 28 mol% Mg(BH4)2 ∙ NH3, whereas MI300 contains 74 mol% nanoconfined material and 26 mol% Mg(BH4)2 ∙ NH3. This is consistent with the PXD, where diffraction from Mg(BH4)2 ∙ NH3 is observed.The presence of crystalline Mg(BH4)2 ∙ NH3 and Mg(BH4)2 ∙ 2NH3 in MI100 observed in the PXD data after 5 months of storage (see Fig. 1b) may be attributed to an incomplete infiltration, allowing for the recrystallization of the eutectic composition. As the degree of pore filling increases to 200% and 300%, there is no evidence of Mg(BH4)2 ∙ 2NH3 in the 11B NMR spectra, in accordance with the PXD results in Fig. 1b. There is no discernible difference in linewidth between the infiltrated samples, indicating that degree of pore filling does not affect boron dynamics in the range from 100 to 300% (see Supplementary Table 2). Thus, the highly dynamic eutectic molten state is stabilized both inside the pores and on the surface of the SBA-15 nanoparticles.Ionic conductivityThe Mg2+ ionic conductivity as a function of temperature of the first heating and cooling cycles of the samples are shown in Fig. 3a. The associated Nyquist plot and equivalent circuit of MI100 is presented in Supplementary Fig. 4. The Mg2+ ionic conductivity of MI100 is 3.1 × 10−8 to 6.3 × 10−6 S cm−1 in the temperature range of 32 to 80 °C. However, the Mg2+ conductivity gradually decreases in the 2nd and 3rd cycles of MI100, see Supplementary Fig. 5. This may be attributed to the electrolyte further intercalating into SBA-15, which leads to reduced particle-particle contact, resulting in a lower ionic conductivity. Both MI200 and MI300 achieved higher Mg2+ ionic conductivities of 9.1 × 10−6 to 2.7 × 10−4 S cm−1 and 5.5 × 10−6 to 7.4 × 10−4 S cm−1, respectively, in the temperature range of 32–80 °C. This reflects the existence of electrolyte on the surface of the SBA-15 particles, which allows for Mg2+ migration across the SBA-15 particles and provides additional interface for Mg2+ ionic migration, as compared to MI100. Overfilling of the mesoporous scaffolds, i.e., sample MI200 and MI300, reveal that both the inner and the outer surface stabilize the eutectic molten state of the composite. This suggests that interface conductivity in three dimensions is faster and more efficient as compared to one-dimensional intra-pore conductivity in the mesoporous silicate scaffold. The highest conductivity of MI300 at high temperature could be due to the higher amount of the active conducting material (i.e., eutectic molten material). However, the hysteresis in ionic conductivity of MI300 upon heating and cooling may be caused by the remelting/recrystallization of Mg(BH4)2·1.47NH3 owing to an insufficient amount of SBA-15 (only 35 wt%) to suppress these phenomena.Fig. 3: Ionic conductivity.a Mg2+ ionic conductivity as a function of temperature for the samples MI100, MI200, and MI300 during the 1st heating (full lines) and cooling (dashed lines) cycles. b Mg2+ ionic conductivity as a function of temperature compared to other ammine magnesium borohydrides doped with oxide nanoparticles (Mg2+ ionic conductivities of MI100, MI200 and MI300 are from the 2nd heating cycle)13,26.A significant increase in ionic conductivity of MI200 is observed upon the 1st cooling cycle and is stable for at least 3 cycles of heating and cooling between the temperatures of 32–80 °C (see Supplementary Fig. 5). The increase in conductivity after the initial heating may be related to the elimination of grain boundaries that form as a result of pressing the pellet. For the MI300, the apparent hysteresis during each heating and cooling may be due to melting or recrystallization of the Mg(BH4)2 ∙ 1.47NH3 composite. This suggests that the amount of SBA-15 is too low to suppress the recrystallization, and thus it behaves like the bulk sample (see Supplementary Fig. 5). Thus, the ionic conductivity measurement suggests that MI200 is the ideal composition, where the pores of SBA-15 is filled and a sufficient amount of the Mg(BH4)2 ∙ 1.47NH3 composite is present on the surface of SBA-15 to ensure contact between the particles, but in a sufficiently small surface layer to preserve the highly dynamical state of Mg(BH4)2 ∙ 1.5NH3.Compared to the Mg(BH4)2 ∙ 1.47NH3 composite, the MI200 and MI300 samples show higher Mg2+ ionic conductivity at T < 50 °C, see Fig. 3b. Compared to the other oxide composites, the MI200 shows a slightly higher Mg2+ ion conductivity relative to Mg(BH4)2·1.5NH3 mixed with 75 wt% of MgO in the low temperature range of 32–45 °C13, while that with 67 wt% Al2O3 display a higher ionic conductivity26. In all cases the activation energy is lower than for Mg(BH4)2·1.5NH3 (3.4 eV at T < 55 °C) in the temperature range of 32 to 80 °C (see Table 2), demonstrating efficient confinement using SBA-15. Activation energy for all experiments is provided in Supplementary Table 3.Table 2 Ionic conductivityThe electrochemical stability of MI200 was evaluated by cyclic voltammetry (CV) (Fig. 4a). A potential range between –0.5 and 0.5 V vs. Mg/Mg2+ was applied to an asymmetric cell, Mo|MI200|Mg at 70 °C for initially 15 cycles, and later 85 additional cycles. The cathodic and anodic current peaks corresponding to Mg2+ plating and stripping, respectively, were observed during the 15 initial cycles. The increasing plating/stripping currents during the first 10 cycles indicates the formation of a favorable interface layer, enhancing the interfacial contact between the electrodes and the electrolyte13. The cell was relaxed overnight, and subsequently continued for up to a total of 100 cycles, showing a lower initial plating and stripping current in cycle 16, which gradually increased each cycle. However, significant noise is observed during plating after cycle 51, possibly related to contact issues after several stripping/plating cycles (see Supplementary Fig. 7). After the first 100 cycles, the cell was relaxed overnight, and the voltage range between −0.5 to 2.5 V vs. Mg/Mg2+ was applied to the cell (see Supplementary Fig. 8). During the 1st cycle, an irreversible oxidation peak is observed at voltages above 1.2 V vs. Mg/Mg2+. This is assigned to the oxidation of the borohydride anion in agreement with other Mg(BH4)2 derivatives13,14,16,17. However, despite the oxidation of the electrolyte, it still allows for plating and stripping of Mg in the subsequent cycles, in agreement with the previous reports of Mg(BH4)2-1.6NH3 with MgO nanoparticles13. The oxidation of the electrolyte is less pronounced upon further cycling. The current density reaches a maximum in the 4th cycle and remains stable up to the 10th cycle, then gradually decreases to 50% and 25% of the maximum current density at the 20th and 30th cycles, respectively.Fig. 4: Electrochemical properties.a Cyclic voltammogram of a Mo|MI200|Mg cell at voltage range between −0.5 to 0.5 V at 70 °C with a scan rate of 10 mV s−1 and b a chronoamperogram of SS|MI200|SS at 70 °C with an applied voltage of 0.5 V.A chronoamperometry experiment was used to examine the ionic transport number (tion) of MI200 (Fig. 4b). By applying potential of 0.5 V to a cell using stainless steel (SS) blocking electrodes SS|MI200|SS at 70 °C, the steady state current (ie) of 0.727 nA was observed. The electronic conductivity was determined to be 7.93 × 10–11 S cm–1 using Ohm’s law resulting in an ionic transport number of 0.9999996. This indicates that the conducting species are almost exclusively ionic.Thermal stabilityThe thermal stability was determined using thermogravimetric analysis (TGA) and mass spectrometry (MS). Thermal decomposition of as-prepared Mg(BH4)2·1.47NH3 occurs with onset at Tonset ∼ 100 °C (see Supplementary Fig. 6), which is lower than the previously reported Tonset value for Mg(BH4)2·1.5NH3 (∼120 °C)13. This difference could be attributed to the lower heating rate of 2 °C/min compared to 5 °C/min in previous work. The thermal decomposition profiles of the confined samples exhibit a high thermal stability of more than 100 °C, see Supplementary Fig. 6. The mass loss of as prepared Mg(BH4)2·1.47NH3 in the temperature range of 100 to 190 °C is 4.6 wt%, while the mass loss of the confined samples ranges from 1.86 to 2.60 wt%, increasing with the Mg(BH4)2·1.47NH3 content in the sample. The different H2-release temperature between Mg(BH4)2·1.47NH3 and the confined samples (see Supplementary Fig. 6) could be an effect of nanoconfinement. Only hydrogen gas is observed by mass spectrometry, which suggests hydrogen elimination via di-hydrogen bonds in the solid-state.
