Promoted hydrogen release from alkali metal borohydrides in ionic liquids

Abstract

The use of alkali metal borohydrides for hydrogen storage has long been restricted by high dehydrogenation temperature and large endothermic dehydrogenation enthalpy. Here we report that the dehydrogenation properties of NaBH₄ and LiBH₄ can be significantly improved by the ionic liquid (IL) 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide (bmimNTf₂). The borohydrides form homogeneous solutions in bmimNTf₂, which release more than 70% of theoretical hydrogen below 180 °C, significantly lower than that in the solid state (370 °C for LiBH₄ and 500 °C for NaBH₄). The dehydrogenation reactions become highly exothermic in the IL, which is in contrast to the highly endothermic process in their solid states. The drastically changed dehydrogenation behaviour in IL is attributed to the destabilization of borohydrides due to the more favorable charge transfer from BH₄⁻ to the cation in the IL, which is in line with the established stability rule of metal borohydrides. The IL remains unchanged after dehydrogenation, which provides the possibility of its repeated use.

---

Introduction

Hydrogen is an outstanding next-generation energy source for its abundance, cleanliness and high energy capacity. However, safe and compact storage of hydrogen has long been a crucial and restrictive step in the widespread use of hydrogen energy.^1–3 For on-board applications, many properties including gravimetric/volumetric capacity, (de)hydrogenation thermodynamics and kinetics and reversibility should be considered. Researchers have investigated many hydrogen storage materials such as metal hydrides,⁴ borohydrides,⁵ amides/imides,⁶ alanates,⁷ ammonia borane (AB)⁸ and their combinations.⁹ Unfortunately, these solid state materials or their combinations can hardly meet the requirement of practical application.

Alkali metal borohydrides are promising candidates for hydrogen storage due to their high gravimetric hydrogen capacity (LiBH₄, 18.5 wt%; NaBH₄, 10.7 wt%). However, the use of these compounds through thermal dehydrogenation is predominantly hindered by the extraordinarily high dehydrogenation temperature.^10,11 This drawback mainly originates from the high dehydrogenation enthalpy of alkali metal borohydrides, i.e., the strong endothermic effect during dehydrogenation (Table 1). Therefore, modifying the dehydrogenation enthalpy by either destabilizing the reactants or stabilizing the products is a crucial and effective way to lower the dehydrogenation temperature.

Table 1 Dehydrogenation enthalpy and temperature of hydrogen storage materials containing alkali metal borohydride

Reactant	Product	T d (°C)	ΔrH (kJ mol^−1 H2)	Ref.
a These dehydrogenation temperatures are given by the extrapolation of the Van't Hoff plot. b These reaction enthalpies are calculated using the thermodynamic data of reactants and products.
NaBH4	Na + B + 2H2	534^a	123,^b 106.8	10 and 31
NaBH4	NaH + B + 3/2H2	—	133.32^b	10
LiBH4	LiH + B + 3/2H2	370^a	67,^b 74	11 and 12
2LiBH4 + MgH2	2LiH + MgB2 + 4H2	270	40.5	15
6LiBH4 + CaH2	6LiH + CaB6 + 10H2	350	59	32
2NaBH4 + MgH2	2Na + MgB2 + 5H2	515	90.9	15 and 31
3LiBH4 + TiF3	3LiF + TiB2 + B + 6H2	269	−157.1	17
3NaBH4 + YF3	3NaF + 3/4YB4 + 1/4YH2 + 23/4H2	423	51.2	19
NaBH4 (in IL)	NaBH1.04 + 1.48H2	179	−183.0	This work
LiBH4 (in IL)	LiBH1.16 + 1.42H2	162	−138.3	This work

---

There are many examples of the “stabilizing the products” strategy. By adding hydrides and/or other borohydrides containing alkali-earth or transition metals, stable borides can form, which significantly decreases the dehydrogenation enthalpy and lowers the dehydrogenation temperature. For instance, in the dehydrogenation reaction of a LiBH₄ + MgH₂ composite, a more stable phase, MgB₂, is formed as the final product.¹² As a result, the dehydrogenation temperature is lowered to 270 °C. In another example, LiBH₄ is ball-milled with CaH₂ and forms CaB₆ as the final product upon dehydrogenation.¹³ Many similar composites are investigated including LiBH₄ + CeH₂,¹³ LiBH₄ + Ca(BH₄)₂,¹⁴ NaBH₄ + MgH₂,¹⁵ NaBH₄–CaH₂/Ca(BH₄)₂,¹⁶ LiBH₄ + TiF₃,¹⁷ LiBH₄ + AlF₃,¹⁸ NaBH₄–YH₃,¹⁹etc. Although the “stabilizing the product” strategy shows certain effectiveness, the dehydrogenation temperature of most of these composites remains very high (typically >300 °C), with a dehydrogenation enthalpy larger than 40 kJ mol⁻¹ H₂. Moreover, the introduced additives significantly lower the hydrogen content and often cause complicated side reactions and formation of by-products.

An alternative strategy is to destabilize the reactants. This strategy requires thermodynamic modification on the borohydrides themselves. It is more difficult to apply as the thermodynamic stability of solid borohydrides is dominated by their intrinsic properties. High-energy ball-milling might cause some destabilization effects by introducing more defects. But the experimental investigation on ball-milled NaBH₄–MgH₂ and LiBH₄–MgH₂ composites shows little improvement in terms of both the dehydrogenation temperature and thermodynamics.^20,21

Here we report a novel approach to destabilize the alkali borohydrides NaBH₄ and LiBH₄ by dissolving the borohydrides into an ionic liquid (IL). There have been a few studies on using ILs for hydrogen storage. The IL methylguanidinium borohydride has been studied as a hydrogen storage material dissolved in ethylene diglyme.²² In this case, the IL behaves just like other solid borohydrides, which is catalytically decomposed to give hydrogen and a solid product. Imidazolium-based ILs have been shown to have a catalytic effect on dehydrogenation from ammonia borane (AB)^23–25 and ethylene diaminebisborane (EDB).²⁶ However, as the dehydrogenation of AB and EDB is thermodynamically favourable, similar or even better promotion effects can also be achieved by a large variety of solid catalysts.^27,28

For the thermodynamically highly stable alkali metal borohydrides, the effect of IL on the dehydrogenation is more drastic and unique. Dissolving in IL significantly lowers the dehydrogenation temperature and completely reverses dehydrogenation thermodynamics. Interestingly, NaBH₄ and LiBH₄ exhibit very similar dehydrogenation behaviour in the IL. Over 70% of theoretical hydrogen can be released from 160–180 °C. Moreover, the IL shows no structural changes after dehydrogenation and can be easily recycled, behaving as a pure catalyst in the very classic definition. The favourable dehydrogenation temperature of the liquid borohydride-IL hydrogen storage system will greatly advance the application of borohydrides as high capacity hydrogen storage materials.

Experimental section

Synthetic procedure

Materials. NaBH₄ (99.5%) and LiBH₄ (99.5%) are purchased from J&K Chemical Co. Ltd. 1-Butyl-3-methylimidazolium bromide (bmimBr), 1-butyl-2,3-dimethylimidazolium bromide (bmmimBr) and lithium bis(trifluoromethylsulfonyl)imide (LiNTf₂) are purchased from Lanzhou Institute of Chemistry and Physics. All the materials are used as-received unless stated otherwise.

Synthesis of bmimNTf₂ and bmmimNTf₂. BmimBr (12.98 g, 59 mmol) and LiNTf₂ (17.02 g, 59 mmol) are dissolved in 100 mL of deionized water respectively. Then the two clear solutions are mixed in a 500 mL flask, which rapidly develops a water–oil interface. The oil phase is separated and washed six times with deionized water until it becomes clear. The IL is extracted by dichloromethane, thoroughly washed with water three times and distilled under vacuum to remove dichloromethane. The IL is further kept at 120 °C for 4 days under dynamic vacuum and then transferred into an Ar-filled glovebox. The IL is characterized by using ¹H NMR. No residual water is detected on the NMR spectrum (Fig. 1a and d). The ionic structure of the bmimNTf₂ is shown in Fig. 1c. 1-Butyl-2,3-dimethylimidazolium bis(trifluoromethylsulfonyl)imide (bmmimNTf₂) is synthesized in the same way using bmmimBr (13.44 g, 58 mmol) and LiNTf₂ (16.56 g, 58 mmol).

Fig. 1 (a) ¹H NMR spectrum of the as-prepared bmimNTf₂ sample in CD₃CN. (b) ¹³C NMR spectrum of the as-prepared bmimNTf₂ sample in CD₃CN. The digits in (a) and (b) are corresponding to the atoms labelled in (c). The letters d, w and f represent CD₃CN, water and trifluoromethyl, respectively; (c) ionic structure of bmimNTf₂; (d) ¹H NMR spectrum of the as-prepared bmimNTf₂ sample without deuterated solvent.

Preparation of borohydride-IL samples. NaBH₄ (11.0 mg, 0.29 mmol) and bmimNTf₂ (11.99 g, 29 mmol) are milled in an agate mortar for 20 min to form a clear, colorless solution. The samples of bmimNTf₂-LiBH₄-1 mol% (denoted as LiBH₄-IL), bmmimNTf₂-NaBH₄-1 mol% (denoted as NaBH₄-bmmimNTf₂) and bmmimNTf₂-LiBH₄-1 mol% (denoted as LiBH₄-bmmimNTf₂) are prepared and stored using the same procedure. All these samples are clear and colourless solutions with relatively high viscosity. All the samples are prepared and stored in an Ar-filled glovebox.

Preparation of bmimBr/LiNTf₂–borohydride samples. BmimBr and LiNTf₂ are held at 150 °C under dynamic vacuum for 24 h to remove water before use. The bmimBr-NaBH₄-1 mol% sample (denoted as bmimBr-NaBH₄) is prepared by milling NaBH₄ (10.5 mg, 0.28 mmol) and bmimBr (5.99 g, 27.4 mmol) in an agate mortar for 15 min. The sample is collected and stored in a glass bottle. The samples of bmimBr-LiBH₄-1 mol% (denoted as bmimBr-LiBH₄), LiNTf₂–NaBH₄-1 mol% (denoted as LiNTf₂–NaBH₄) and LiNTf₂–LiBH₄-1 mol% (denoted as LiNTf₂–LiBH₄) are prepared using the same method. They are all white and fine powders. All the samples are prepared and stored in an Ar-filled glovebox.

Characterization techniques

NMR spectroscopy. ¹¹B NMR spectra are collected for liquid samples on a Bruker AVANCE III 500 MHz (for ¹H nuclei) spectrometer. All the ¹¹B NMR data is referenced to BF₃·Et₂O (at 0 ppm) as an external standard material. ¹H and ¹³C NMR spectra are recorded on the same spectrometer. All the NMR spectra are collected in acetonitrile-d₃ unless stated otherwise.

Temperature programmed desorption/mass spectrometry (TPD/MS) analysis. TPD/MS experiments are performed on a Quantachrome Autosorb iQ automatic gas sorption analyser. On heating, the carrier gas (Ar, purity >99.999%) flows through the sample tube and brings the released gas with it. The gas composition is further analysed by a Pfeiffer PrismaPlus™ Mass Spectrometer-Residual Gas Analyser (MS-RGA). Two needle valves, a vent sampling valve and a mass spectrometer input valve, are adjusted to introduce an appropriate amount of gas into the mass spectrometer. The multiple ion detection (MID) mode is used to analyse the temporal variation of the specific products. Temperature is simultaneously measured by a thermocouple installed next to the sample cell. The heating rate is 5 K min⁻¹ for all the samples. A gross leak test is performed before each experiment.

Quantification of the hydrogen released. Quantification experiments are performed on a Quantachrome Autosorb iQ automatic gas sorption analyser. A home-made glass tube is used as a container for the liquid sample (Fig. S1, ESI†). The sample is heated in a high purity argon (99.999%) flow of 50 mL min⁻¹. The generated gas is allowed to pass a cold trap cooled by using an ice bath before entering the sampling TCD and the thermal conductivity difference is measured to determine the quantity of the released gas. The acquired TCD signals are integrated to calculate the amount of hydrogen released. The coefficient used to convert the integration to the amount of hydrogen is determined by regression of the working curve using the external standard, pure MgH₂ (Fig. S2, ESI†). For all the measurement runs in this paper, no residue is found in the cold trap. Temperature is simultaneously measured by a thermocouple installed next to the sample cell. The heating rate is 5 K min⁻¹ for all the samples. A gross leak test is performed before each experiment.

Differential scanning calorimetry (DSC) measurements. DSC measurements are carried out on a Netzsch DSC HP 204 calorimeter. The sample (15.0 to 25.0 mg) is loaded in an aluminium crucible in an argon-filled glovebox. Another blank crucible is used as the reference. An aluminium lid with a pinhole on it covers the crucible, allowing the gaseous product to escape. The experiments are performed in an inert gas flow (high purity argon, 99.999%, 50 mL min⁻¹, 1.0 bar) with various heating rates.

Results and discussion

Characterization of bmimNTf₂ and the borohydride-bmimNTf₂ solution

The structure of the as-prepared bmimNTf₂ sample is investigated by ¹H and ¹³C NMR spectra (Fig. 1a and b). The resonance peaks on these spectra can be ascribed to the corresponding H and C atoms in the ion structure shown in Fig. 1c. However, a small resonance peak corresponding to water is found in the ¹H NMR spectrum. To find out whether this water residue is from bmimNTf₂ or deuterated solvent, we performed a ¹H NMR experiment of pure bmimNTf₂ without deuterated solvent (Fig. 1d). No peaks related to water are found in these spectra. Therefore, these results proved that the bmimNTf₂ prepared is pure and dry. Therefore, it can be used in the follow-up experiments.

After milling, the mixtures of MBH₄ and IL are clear and colourless solutions, as shown in Fig. 2b. The disappearance of borohydride solids during milling indicates dissolution of borohydrides in bmimNTf₂. The ¹H NMR spectra of NaBH₄-IL and LiBH₄-IL samples are shown in Fig. 2a. The multiplets from −0.5 to 0.5 ppm prove the existence of BH₄⁻ anions in the samples. In the ¹¹B NMR spectra (Fig. 2b), the quintets centred at −37.0 and −36.8 ppm can be ascribed to the resonance signals of BH₄⁻ in the samples. These spectroscopic features are very similar to those of BH₄⁻ in the solid state and other solvents.^29,30

Fig. 2 (a) ¹H NMR spectra of pristine bmimNTf₂, NaBH₄-IL and LiBH₄-IL samples; (b) ¹¹B NMR spectra of the NaBH₄-IL and LiBH₄-IL samples. The inset pictures are the photographs of the corresponding samples.

Furthermore, in both ¹H (Fig. 2a) and ¹³C NMR spectra of NaBH₄-IL and LiBH₄-IL samples (Fig. S3, ESI†), the resonance peaks corresponding to bmimNTf₂ do not change, indicating that the structure of IL does not change after dissolution of the borohydrides. These results indicate that the borohydrides can be dissolved into the IL without changing the structure of the IL or BH₄⁻ during the milling procedure.

Dehydrogenation properties

Fig. 3a and b show the dehydrogenation profile of NaBH₄-IL and LiBH₄-IL monitored by TPD/MS, respectively. The onset dehydrogenation temperature is 104 °C for NaBH₄-IL and 85 °C for LiBH₄-IL, while the peak temperature is 179 °C for NaBH₄-IL and 162 °C for LiBH₄-IL, respectively. No gaseous impurities are found under these conditions. The IL decomposes and emits some gaseous impurities above 200 °C. To avoid the decomposition of IL, the dehydrogenation temperature is limited to 180 °C for NaBH₄ and 160 °C for LiBH₄, respectively.

Fig. 3 TPD/MS curves of NaBH₄-IL (a) and LiBH₄-IL (b) samples in their dehydrogenation reaction. The MS signals for H₂ (m/z = 2, black), B₂H₆ (m/z = 27, red), and B₅H₉ (m/z = 59, green) are demonstrated in solid lines. The blue dashed curves show the temperature profile; isothermic dehydrogenation profiles of NaBH₄-IL (c) and LiBH₄-IL (d). For comparison, the hydrogen released at the same temperature from solid NaBH₄ and LiBH₄ is also shown.

Fig. 3c and d show the amount of H₂ released from both samples determined by TCD measurement. For the NaBH₄-IL sample, the dehydrogenated hydrogen content is 8.1% based on NaBH₄, covering 74% of the hydrogen in NaBH₄. For the LiBH₄-IL sample, the dehydrogenated content is 13.2 wt% based on LiBH₄, covering 71% of the hydrogen in LiBH₄. At the dehydrogenation temperature of NaBH₄-IL and LiBH₄-IL samples, the solid state NaBH₄ and LiBH₄ show no evidence of dehydrogenation. Such dehydrogenation properties are definitely not from a hydrolysis reaction since no residual water is detected in bmimNTf₂ (Fig. 1d).

The dehydrogenation products

After dehydrogenation, the samples appear as a clear solution with some precipitates at the bottom of the reaction tube. The solution is studied using ¹¹B, ¹H and ¹³C NMR spectra. ¹¹B NMR spectra of the solution indicate that the BH₄⁻ anion in borohydrides completely decomposes in the dehydrogenation reaction (Fig. 4a and b), which is in agreement with the absence of the B–H resonance peak in the ¹H NMR spectra in the 0.5 to −0.5 ppm region (Fig. 4c). The full range ¹¹B NMR spectra are shown in Fig. S4 (see the ESI†), showing no soluble boron-containing species in the solution phase. It means that all the boron-containing species are enriched in the solid phase.

Fig. 4 ¹¹B NMR spectra of the NaBH₄-IL (a) and LiBH₄-IL (b) samples before and after dehydrogenation. ¹H NMR (c) and ¹³C NMR (d) spectra of pristine bmimNTf₂ and dehydrogenated samples. The ¹³C NMR spectra are recorded without deuterated solvents.

Regarding the structure of IL after dehydrogenation, the ¹H NMR (Fig. 4c) spectra suggest that the chemical bonding of H atoms is not changed by heating or chemical reactions. Meanwhile, no evidence for the formation of C–B bonds or any other new C-related bonds is found in ¹³C NMR spectra (Fig. 4d). The above results indicate that the IL bmimNTf₂ is chemically stable and serves as a pure solvent in the dehydrogenation. Therefore, we may conclude that the dehydrogenated samples consist of a boron-containing solid phase and a liquid phase of pure IL. The stability of the IL during dehydrogenation makes recycling of the IL after dehydrogenation for repeated use possible.

The precipitate is also characterized using XRD (Fig. S5, ESI†) and FT-IR (Fig. S6, ESI†). The XRD patterns show broad peaks for both NaBH₄-IL and LiBH₄-IL samples, indicating that the dehydrogenation products are amorphous. In the IR spectra, only very weak peaks are detected in the B–H stretching region (2250–2400 cm⁻¹), indicating little residual B–H bond in the dehydrogenation products. This indicates rather complete elimination of the B–H bonds at moderate temperature in IL. According to the dehydrogenation capacity (Fig. 3c and d), the nominal dehydrogenation reaction of NaBH₄-IL and LiBH₄-IL can be described by eqn (1) and (2), respectively.

NaBH₄-IL → NaBH_1.04 + IL + 1.48H₂ (1)

LiBH₄-IL → LiBH_1.16 + IL + 1.42H₂ (2)

In their solid states, borohydrides usually exhibit multiple dehydrogenation steps. A stable intermediate species, M₂B₁₂H₁₂, is found to be the key intermediate in the dehydrogenation reaction pathway of alkali metal borohydrides.^33–35 However, in MBH₄-IL samples, the dehydrogenation capacity is much higher (70–75%) compared to the partial dehydrogenation of MBH₄ to M₂B₁₂H₁₂. The stoichiometry of the dehydrogenation products in the IL cannot be attributed to any known alkali metal hydroborates. Rather, it best matches a mixture of boron and metal hydride, which is the dehydrogenation product at high temperature for solid state LiBH₄ and NaBH₄.^10,11 This is also in agreement with the very low residual B–H bonding in the solid precipitate. Further characterization of the solid dehydrogenation product will be our next research focus.

Another interesting observation is the highly similar dehydrogenation behaviour in terms of both dehydrogenation temperature and the dehydrogenation capacity of NaBH₄-IL and LiBH₄-IL. In the solid state, there is a significant difference in the dehydrogenation temperature and dehydrogenation mechanism.⁵ The nearly metal independent dehydrogenation behaviour has important implications for the dehydrogenation mechanism, as will be discussed later.

Thermodynamic and kinetic investigation

Since the dehydrogenation temperatures of borohydride-IL samples are far lower than those calculated from thermodynamic parameters of solid state materials, the thermodynamic property of this system is very interesting. Therefore, we perform DSC measurements to compare the thermodynamic behaviour of NaBH₄ and LiBH₄ in IL and their solid states. In their crystalline states, NaBH₄ and LiBH₄ show very strong endothermic behaviour during dehydrogenation at high temperature (370 °C for LiBH₄ and 500 °C for NaBH₄) (Fig. 5b and d). The solid state LiBH₄ experiences an orthorhombic to tetragonal phase change (∼130 °C) and melting (∼280 °C) before dehydrogenation.³⁶ The dehydrogenation enthalpy is reported to be 108 kJ mol⁻¹ and 74 kJ mol⁻¹ for pristine NaBH₄ and LiBH₄, respectively.^10,11

Fig. 5 DSC curves of NaBH₄-IL and solid state NaBH₄ (a, b) and LiBH₄-IL and solid state LiBH₄ (c, d) at a heating rate of 2 K min⁻¹. The Kissinger's plots of NaBH₄-IL (black) and LiBH₄-IL (red) are demonstrated in (e). The apparent activation energy (E_a) is calculated from the slope of the fitted lines (the solid lines) according to the Kissinger's equation (eqn (3)).

Surprisingly, the dehydrogenation of NaBH₄-IL and LiBH₄-IL samples becomes highly exothermic (Fig. 5a and c). The NaBH₄-IL sample exhibits an exothermic peak at 134 °C, and the dehydrogenation enthalpy is −183.0 kJ mol⁻¹ H₂. The LiBH₄-IL sample also shows an exothermic peak at 127 °C, and the dehydrogenation enthalpy is −138.3 kJ mol⁻¹ H₂. In contrast to their very similar dehydrogenation temperature and dehydrogenation capacity, the dehydrogenation enthalpy of NaBH₄-IL and LiBH₄-IL is quite different. The dehydrogenation enthalpies are lower than those in most of the systems listed in Table 1,^12,15,31,32 but are similar to those observed in the borohydride–fluoride composites like LiBH₄–YF₃,¹⁷ which has an exothermic dehydrogenation enthalpy due to the formation of highly thermodynamically stable products such as LiF and TiB₂. However, the exothermic dehydrogenation is rather surprising for MBH₄-IL, as the IL is unchanged after dehydrogenation. The favourable dehydrogenation of MBH₄-IL, therefore, is a result of destabilized reactants. It is a much advantageous approach because no reactive additives are involved in the dehydrogenation reaction.

The activation energy of the dehydrogenation process is evaluated using the Kissinger's equation by measuring multiple DSC curves (Fig. S7 and S8, ESI†) at different heating rates (eqn (3)).³⁷

where β is the heating rate of DSC measurements (K min⁻¹), T_p is the peak temperature on the DSC curve, R is the ideal gas constant and E_a is the apparent activation energy.

The apparent activation energies for NaBH₄-IL and LiBH₄-IL samples are calculated to be 131.3 kJ mol⁻¹ and 91.4 kJ mol⁻¹, respectively (Fig. 5e). The high apparent activation energy shows that both NaBH₄-IL and LiBH₄-IL are thermodynamically unstable systems stabilized by kinetic barriers. It also provides the possibility to further lower the dehydrogenation temperature by introducing suitable catalysts in the MBH₄-IL system, which will be a very interesting field deserving further exploration.

Dehydrogenation mechanism study

Destabilization of MBH₄ by IL. According to the DSC results, dehydrogenation of the solid state MBH₄ (or melted LiBH₄) is highly endothermic:

Δ_rH(MBH₄, s) = Δ_fH(MBH) − Δ_fH(MBH₄, s) > 0 (4)

The solid dehydrogenation product is represented by “MBH”, which can be considered as a mixture of metal hydride and elemental boron, the high temperature dehydrogenation product of NaBH₄ and LiBH₄.

In contrast, DSC results also suggest that dehydrogenation of MBH₄ in the IL (eqn (1) and (2)) is highly exothermic:

Here the solid dehydrogenation product precipitated from the IL is also represented by “MBH”, according to the stoichiometry derived from the dehydrogenation capacity. The term Δ_fH(MBH₄-IL) describes the formation enthalpy of the MBH₄-IL samples. This term can be approximated as the sum of the inherent formation enthalpy of IL (Δ_fH(IL)) and the energy of the dissolved MBH₄ (Δ_fH(MBH₄ in IL)).

Δ_fH(MBH₄-IL) ≈ Δ_fH(MBH₄ in IL) + Δ_fH(IL) (6)

This is a rather good approximation for a dilute MBH₄-IL solution. In this case we can treat the IL as an infinitely large bath which is little affected by the dissolved borohydrides.

According to eqn (4), (5), and (6), the borohydride in IL becomes less stable compared to that in the solid state (eqn (7)),

Δ_fH(MBH₄ in IL) ≫ Δ_fH(MBH₄, s) (7)

Both NaBH₄ and LiBH₄ are typical ionic crystals.^38,39 Using the gas phase ions M⁺(g) and BH₄⁻(g) as the reference state, Δ_fH(MBH₄, s) is virtually the lattice enthalpy of MBH₄ (ΔU_MBH4). On the other hand, Δ_fH(MBH₄ in IL) is the solvation enthalpy (E_s) in the IL, determined by the interaction between the borohydrides and the IL. The destabilization effect (eqn (7)), therefore, is a result of the large difference between ΔU_MBH4 and E_s:

The energetics of the M–B–H species are summarized in Scheme 1. In this diagram, two approximations are made except that in deriving eqn (6): (1) in the case of pristine LiBH₄, we neglect the energy change due to phase transition and melting. However, these terms are small compared to the lattice enthalpy of LiBH₄ and can be included as a slight upshift of the position of LiBH₄ in the energy diagram. (2) We use the same dehydrogenation product “MBH” for both pristine MBH₄ and MBH₄-IL. This could be complicated by other dehydrogenation products, particularly in the case of low temperature dehydrogenation in the solid state. Nevertheless, the MBH₄ is the most stable M–B–H species compared to all the dehydrogenation products composed of M–B–H. For instance, one well known stable dehydrogenation intermediate M₂B₁₂H₁₂ is also an endothermic product from MBH₄.³⁴ Therefore, other M–B–H type solid state dehydrogenation products can be regarded as a band around the representative “MBH” species.

Scheme 1 A schematic energy diagram of the dehydrogenation of borohydrides in the solid state and IL solution.

The above approximations may cause slight adjustments on the energetic diagram, as shown in Fig. S9 (see the ESI†). However, the simplified diagram (Scheme 1) is sufficiently illustrative. The stability increases in the order: MBH₄ in IL < the dehydrogenation product (represented by “MBH”) < MBH₄(s), as suggested by the DSC results. This diagram clearly shows that the reversed dehydrogenation thermodynamics in the IL is caused by destabilization of MBH₄. This diagram also explains the limitation of thermodynamic modulation in the solid state. In pristine MBH₄, the thermodynamics is dominated by the large lattice enthalpy (ΔU_MBH4). Consequently, the dehydrogenation is always highly endothermic for M–B–H type products, as suggested by previous studies.^34,40

The interaction between BH₄⁻ and the cation in the IL. The destabilization of MBH₄ in IL is due to the interaction between the borohydrides and the IL. To understand this interaction, we study the effect of the cation and the anion in the IL separately (Fig. 6). The bmimBr-NaBH₄ and bmimBr-LiBH₄ samples dehydrogenate at 220 and 230 °C, respectively, which are also significantly lower than those in the solid state and only slightly higher than those of the corresponding MBH₄-IL samples. On the other hand, both LiNTf₂–borohydride composites show negligible dehydrogenation under 240 °C, as shown in Fig. 6. The above results suggest that bmim⁺ is the active component in the IL responsible for the promoted dehydrogenation properties of the borohydrides while NTf₂⁻ is ineffective.

Fig. 6 TPD curves of bmimBr-NaBH₄ (black, solid), bmimBr-LiBH₄ (red, solid), LiNTf₂–NaBH₄ (black, dash dot) and LiNTf₂–LiBH₄ (red, dash dot) samples. The temperature profile is demonstrated as blue dashed line.

At this stage, it is illustrative to recall the highly similar dehydrogenation temperature and capacity for LiBH₄ and NaBH₄ in the IL (Fig. 3), while a large difference is observed in their solid states. In dilute IL solution of borohydrides, BH₄⁻ is mostly surrounded by bmim⁺ cations rather than the metal cations. The almost cation independent dehydrogenation temperature and capacity, therefore, are a natural consequence of the new dehydrogenation mechanism in IL, i.e., dehydrogenation dominated by the bmim⁺/BH₄⁻ interaction.

Destabilization of borohydrides due to the bmim⁺/BH₄⁻ interaction is in well accordance with the stability rule observed in solid metal borohydrides.⁴¹ Nakamori et al. reported that the stability of borohydride is determined by the electronegativity of the metal in the metal borohydride. Metals with lower electronegativity (e.g. alkali metals) form more stable borohydrides.⁴² This rule is well explained by charge transfer between BH₄⁻ and metal cations.^41,43 Charge transfer from BH₄⁻ to the cations will lower the negative charge on BH₄⁻ and consequently destabilize the borohydride. In alkali metal borohydride such as NaBH₄ and LiBH₄, an ionic picture has been proposed by ab initio calculations.³⁹ The charge on the alkali metal cation is close to +1. In stable borohydrides such as LiBH₄, NaBH₄ and Ca(BH₄)₂, the charge transfer between Mⁿ⁺ and BH₄⁻ is relatively low, while in unstable borohydrides such as Zn(BH₄)₂ and CuBH₄, the charge transfer is relatively large.⁴⁴

The charge transfer strongly depends on the nature of the cation. Cations that are more inclined to accept negative charges favour the charge transfer and consequently destabilize the borohydride. For simple metal cations, this trend can be well described by the electronegativity of the corresponding metal. In our study, bmim⁺ is a large cation with a conjugated system. Such a structure is more favourable to accommodate extra negative charges, as the electron delocalization effect can stabilize the extra electron density. Indeed, both experimental and theoretical studies suggest that there is a notable charge transfer effect from NTf₂⁻ to bmim⁺ in bmimNTf₂ or similar ILs.⁴⁵ Such a charge transfer effect is expected to be more favourable from the less electron pulling BH₄⁻ (compared to NTf₂⁻). Therefore, the bmim⁺/BH₄⁻ interaction destabilizes the BH₄⁻ due to the enhanced electron transfer to bmim⁺, which explains the significantly lower dehydrogenation temperature and exothermic nature of the dehydrogenation process.

To further verify the charge transfer based destabilization mechanism, the dehydrogenation of borohydrides in another IL bmmimNTf₂ (bmmim = 1-butyl-2,3-dimethylimidazolium) is studied. The cation bmmim⁺ differs from bmim⁺ by the methyl group on the 2-C of the imidazolium ring. The 2-H in bmim⁺ may be deprotonated by strong bases. Methyl substitution eliminates this potentially active H species. Fig. 7 shows the TPD/MS curves of NaBH₄-bmmimNTf₂ and LiBH₄-bmmimNTf₂ samples. They show a very similar dehydrogenation temperature to that of NaBH₄-bmimNTf₂ and LiBH₄-bmimNTf₂ samples. Destabilization of BH₄⁻, therefore, is not associated with the 2-H of bmim⁺. Furthermore, this result can be well explained by the charge transfer mechanism, as the methyl substitution for H has little effect on the electron affinity of the imidazolium cation. The bulky imidazolium cation with the conjugated structure may stabilize an extra negative charge and thus cause more favourable charge transfer from BH₄⁻, which is little affected by the alkyl groups on the imidazolium ring.

Fig. 7 TPD/MS curves of NaBH₄-bmmimNTf₂ (a) and LiBH₄-bmmimNTf₂ (b) samples. The MS signals for H₂ (m/z = 2, black), B₂H₆ (m/z = 27, red), and B₅H₉ (m/z = 59, green) are demonstrated in solid lines. The blue dashed lines show the temperature profile of the dehydrogenation reaction.

The MBH₄-IL system represents a new type of liquid hydrogen storage system. Similar to the well-studied liquid organic hydrogen storage systems,^46,47 the MBH₄-IL shares many advantages in transport, storage and handling arising from the liquid form. Liquid hydrogen storage systems based on ILs are rare but not totally absent. Previous reports show that ILs can promote the dehydrogenation kinetics of B–N–H type compounds such as NH₃BH₃ and BH₃NH₂CH₂CH₂NH₂BH₃.^23–26 However, as the first step dehydrogenation of these materials is thermodynamically favourable, the role of the IL is mainly a kinetic promoter and toxic gas inhibitor. In fact, similar or even better promotion effects have also been achieved by many other solid catalysts.^8,27

On the other hand, the role of the IL for LiBH₄ and NaBH₄ is truly unique and indispensable. The promotion effect in terms of dehydrogenation temperature is unrivalled by any other additive reported so far. The strong ion pair interaction in MBH₄-IL dramatically changes the dehydrogenation temperature and thermodynamics of the borohydrides, which is an important advantage towards the application of NaBH₄ and LiBH₄ as high capacity hydrogen storage materials. Although inclusion of IL reduces the hydrogen capacity in one dehydrogenation cycle, repeated use of the IL is possible as the IL is unaffected during the dehydrogenation and can be easily separated from the solid state dehydrogenation residue. With engineering efforts to recycle the IL, high hydrogen capacity is attainable for the MBH₄-IL system.

Conclusions

The dehydrogenation of alkali metal borohydrides NaBH₄ and LiBH₄ in the ionic liquid bmimNTf₂ is studied. Three important observations are noted. (1) The dehydrogenation in IL occurs at 160–180 °C, which is similar for NaBH₄-IL and LiBH₄-IL and is much lower than that in their solid states. (2) The dehydrogenation process in the IL is strongly exothermic while it is endothermic in the solid state. (3) The IL remains unaffected after the dehydrogenation process and can be easily separated from the solid dehydrogenation residue of the borohydrides.

The distinct dehydrogenation behaviour originates from the interaction between BH₄⁻ and the cation in the IL which destabilizes BH₄⁻. The destabilization mechanism involves a more favourable charge transfer from BH₄⁻ to the bulky imidazolium cation, which is in line with the well-established stability rule of metal borohydrides. The borohydride-IL system is a promising liquid hydrogen storage system with high attainable hydrogen capacity at moderate temperature.

Acknowledgements

This study is supported by National Natural Science Foundation of China (no. U1201241, 11375020, 51431001 and 21321001). The authors acknowledge Prof. Taiwei Chu from Peking University for his assistance in synthesizing the ionic liquids.

Notes and references

1. L. Schlapbach and A. Zuttel, Nature, 2001, 414, 353–358.
2. J. Graetz, Chem. Soc. Rev., 2009, 38, 73.
3. S.-i. Orimo, Y. Nakamori, J. R. Eliseo, A. Zuettel and C. M. Jensen, Chem. Rev., 2007, 107, 4111–4132.
4. B. Sakintuna, F. Lamaridarkrim and M. Hirscher, Int. J. Hydrogen Energy, 2007, 32, 1121–1140.
5. A. Zuettel, A. Borgschulte and S.-I. Orimo, Scr. Mater., 2007, 56, 823–828.
6. P. Chen, Z. T. Xiong, J. Z. Luo, J. Y. Lin and K. L. Tan, Nature, 2002, 420, 302–304.
7. B. Bogdanovic and M. Schwickardi, J. Alloys Compd., 1997, 253, 1–9.
8. B. Peng and J. Chen, Energy Environ. Sci., 2008, 1, 479–483.
9. J. H. Luo, X. D. Kang and P. Wang, Int. J. Hydrogen Energy, 2013, 38, 4648–4653.
10. P. Martelli, R. Caputo, A. Remhof, P. Mauron, A. Borgschulte and A. Zuettel, J. Phys. Chem. C, 2010, 114, 7173–7177.
11. P. Mauron, F. Buchter, O. Friedrichs, A. Remhof, M. Bielmann, C. N. Zwicky and A. Zuettel, J. Phys. Chem. B, 2008, 112, 906–910.
12. J. J. Vajo, S. L. Skeith and F. Mertens, J. Phys. Chem. B, 2005, 109, 3719–3722.
13. S.-A. Jin, Y.-S. Lee, J.-H. Shim and Y. W. Cho, J. Phys. Chem. C, 2008, 112, 9520–9524.
14. J. Y. Lee, D. Ravnsbaek, Y.-S. Lee, Y. Kim, Y. Cerenius, J.-H. Shim, T. R. Jensen, N. H. Hur and Y. W. Cho, J. Phys. Chem. C, 2009, 113, 15080–15086.
15. J. F. Mao, X. B. Yu, Z. P. Guo, H. K. Liu, Z. Wu and J. Ni, J. Alloys Compd., 2009, 479, 619–623.
16. J. F. Mao, Z. P. Guo, X. B. Yu and H. K. Liu, J. Phys. Chem. C, 2011, 115, 9283–9290.
17. Y. H. Guo, X. B. Yu, L. Gao, G. L. Xia, Z. P. Guo and H. K. Liu, Energy Environ. Sci., 2010, 3, 465–470.
18. P. P. Yuan, B. H. Liu, H. P. Zhu, W. Y. Pan and Z. P. Li, J. Alloys Compd., 2013, 557, 124–129.
19. J. Zou, L. Li, X. Zeng and W. Ding, Int. J. Hydrogen Energy, 2012, 37, 17118–17125.
20. S. Garroni, C. Pistidda, M. Brunelli, G. B. M. Vaughan, S. Surinach and M. D. Baro, Scr. Mater., 2009, 60, 1129–1132.
21. K. Crosby and L. L. Shaw, Int. J. Hydrogen Energy, 2010, 35, 7519–7529.
22. A. Doroodian, J. E. Dengler, A. Genest, N. Roesch and B. Rieger, Angew. Chem., Int. Ed., 2010, 49, 1871–1873.
23. D. W. Himmelberger, L. R. Alden, M. E. Bluhm and L. G. Sneddon, Inorg. Chem., 2009, 48, 9883–9889.
24. R. K. Ahluwalia, J. K. Peng and T. Q. Hua, Int. J. Hydrogen Energy, 2011, 36, 15689–15697.
25. M. E. Bluhm, M. G. Bradley, R. Butterick, U. Kusari and L. G. Sneddon, J. Am. Chem. Soc., 2006, 128, 7748–7749.
26. S. Sahler, S. Sturm, M. T. Kessler and M. H. G. Prechtl, Chem. – Eur. J., 2014, 20, 8934–8941.
27. A. Staubitz, A. P. M. Robertson and I. Manners, Chem. Rev., 2010, 110, 4079–4124.
28. B. Peng and J. Chen, Energy Environ. Sci., 2008, 1, 479–483.
29. Y. Yan, A. Remhof, D. Rentsch and A. Zuettel, Chem. Commun., 2015, 51, 700–702.
30. Y. Yan, A. Remhof, D. Rentsch, Y.-S. Lee, Y. W. Cho and A. Zuettel, Chem. Commun., 2013, 49, 5234–5236.
31. C. Milanese, S. Garroni, A. Girella, G. Mulas, V. Berbenni, G. Bruni, S. Suriñach, M. D. Baró and A. Marini, J. Phys. Chem. C, 2011, 115, 3151–3162.
32. F. E. Pinkerton and M. S. Meyer, J. Alloys Compd., 2008, 464, L1–L4.
33. S. J. Hwang, R. C. Bowman, J. W. Reiter, J. Rijssenbeek, G. L. Soloveichik, J. C. Zhao, H. Kabbour and C. C. Ahn, J. Phys. Chem. C, 2008, 112, 3164–3169.
34. Y. Yan, A. Remhof, S.-J. Hwang, H.-W. Li, P. Mauron, S.-i. Orimo and A. Zuettel, Phys. Chem. Chem. Phys., 2012, 14, 6514–6519.
35. Y. Liu, S. Giri, J. Zhou and P. Jena, J. Phys. Chem. C, 2014, 118, 28456–28461.
36. J. P. Soulie, G. Renaudin, R. Cerny and K. Yvon, J. Alloys Compd., 2002, 346, 200–205.
37. H. E. Kissinger, Anal. Chem., 1957, 29, 1702–1706.
38. P. C. Aeberhard, K. Refson and W. I. F. David, Phys. Chem. Chem. Phys., 2013, 15, 8081–8087.
39. Z. Łodziana and M. J. van Setten, Phys. Rev. B: Condens. Matter, 2010, 81, 024117.
40. Y. S. Zhang, E. Majzoub, V. Ozolins and C. Wolverton, J. Phys. Chem. C, 2012, 116, 10522–10528.
41. Y. Nakamori, H. W. Li, K. Kikuchi, M. Aoki, K. Miwa, S. Towata and S. Orimo, J. Alloys Compd., 2007, 446, 296–300.
42. Y. Nakamori, K. Miwa, A. Ninomiya, H. W. Li, N. Ohba, S. I. Towata, A. Zuttel and S. I. Orimo, Phys. Rev. B: Condens. Matter, 2006, 74, 045126.
43. K. Miwa, N. Ohba, S.-i. Towata, Y. Nakamori and S.-i. Orimo, Phys. Rev. B: Condens. Matter, 2006, 69, 245120.
44. Y. Nakamori, H. Li, M. Matsuo, K. Miwa, S. Towata and S. Orimo, J. Phys. Chem. Solids, 2008, 69, 2292–2296.
45. O. Holloczki, F. Malberg, T. Welton and B. Kirchner, Phys. Chem. Chem. Phys., 2014, 16, 16880–16890.
46. R. H. Crabtree, Energy Environ. Sci., 2008, 1, 134–138.
47. D. Teichmann, W. Arlt, P. Wasserscheid and R. Freymann, Energy Environ. Sci., 2011, 4, 2767–2773.

---

Footnote

† Electronic supplementary information (ESI) available: Additional figures including illustrations, calibration curves, energetic diagrams, detailed NMR spectra and characterization of precipitations. See DOI: 10.1039/c6qi00167j

---