Dehydrogenation studies of the bimetallic borohydrides

Abstract

One of the major issues associated with the use of borohydride complexes for hydrogen storage is the thermodynamic stability of these materials, with the Group I and II complexes also requiring the most demanding temperatures to facilitate dehydrogenation. In recent years, the idea of modulating thermodynamic properties by combining metals with different stabilities (as monocation borohydrides) has come to light. By incorporating a cation with ionic bonding characteristics, it has been proposed that the volatile Sc borohydrides can be stabilized to an extent that diborane release is prevented. We show, using in situ IR and TG analysis, that these complexes do indeed release significant quantities of diborane, which is an irrevocable barrier to reversibility. Our findings suggest that the Group I/Sc bimetallic borohydrides are not suitable for hydrogen storage.

---

Introduction

The lightweight Group I borohydrides (i.e. Li, Na, K) have some of the highest hydrogen weight capacities amongst the various complex hydrides that have been investigated for hydrogen storage. These complexes, however, have thermodynamic and kinetic barriers that prevent fast dehydrogenation under mild conditions. For example, lithium borohydride, LiBH₄, liberates 80% of its hydrogen at a minimum temperature of 653 K,¹ significantly higher than the typical operating temperatures of PEM fuel cells. Attempts to enhance the kinetics of dehydrogenation, including the incorporation of additives^1–3 and nanostructuring,^4–6 have proven to be effective ways to lower dehydrogenation temperatures as well as increase dehydrogenation rates. The stabilities of these borohydrides, however, are intrinsic properties that cannot be altered using the same techniques that have demonstrated success towards improving dehydrogenation kinetics.

It has been shown that the dehydrogenation of Group I borohydrides can be enhanced by addition of a second component that stabilizes the dehydrogenated state thus reducing the enthalpy of dehydrogenation. This was demonstrated by Vajo et al. where LiBH₄ and MgH₂ were mechanically milled with 2–3 mol% TiCl₃ as a catalyst.⁷ Dehydrogenation proceeded according to eqn (1):

LiBH₄ + 1/2MgH₂ ⇔ LiH + 1/2MgB₂ + 2H₂ (1)

The formation of MgB₂ was found to stabilize the dehydrogenated state, consequently destabilizing the initial hydrogenated state, and effectively lowering the enthalpy of dehydrogenation by 25 kJ mol⁻¹. These results initiated a new direction in the search for a hydrogen storage material: thermodynamic tuning through the use of additives.

With regards to the necessary thermodynamic properties for solid state hydrogen storage, the transition metal borohydrides have largely been regarded as impractical. The volatility of these complexes under ambient temperatures and pressures has been known since the middle of the 20^th century.⁸ While alkali metals form highly stable ionic interactions with BH₄⁻ and require high energy input in order to facilitate dehydrogenation, the transition metals bond covalently with the borohydride anion.⁹ As a result, the transition metal borohydrides are often highly reactive with much lower dehydrogenation temperatures than the Group I complexes.⁸ This issue, along with the evolution of diborane (B₂H₆) during decomposition, has eliminated the majority of the transition metal borohydrides from the pool of complex hydrides with hydrogen storage potential.

Although scandium borohydride (Sc(BH₄)₃) has not been isolated due to its volatility and lack of stability, the potential for scandium borohydride to dehydrogenate at mild temperature (<300 °C) based on first-principles calculations¹⁰ has encouraged investigations into the dehydrogenation properties of bimetallic borohydride complexes incorporating scandium along with a Group I metal. The choice of the Group I metal has typically been based on the correlation between the electronegativity of the borohydride cation and the stability of the complex which has been recognized since 1955 ¹¹ but was more recently revived by Nakamori et al.¹⁰ The enthalpy of formation, ΔH_f, of the borohydride complex and the Pauling electronegativity, χ_p, of the metal cation, were calculated to follow a linear correlation described as ΔH_f = 248.7χ_p − 390.8. This relationship has also been linked to the dehydrogenation temperature of the metal borohydride.

It was then speculated that a bimetallic complex incorporating two metals with greater differences in electronegativities, i.e. a highly stable metal, such as Li or Na, with an overly volatile one, such as a transition metal, might result in a species with an appropriate stability for reversible hydrogen storage. Over recent years, several studies have been published on the synthesis and dehydrogenation of mixed metal borohydrides.^12–18 Vibrational spectroscopy revealed that scandium has a unique tendency to form anionic borohydride complexes,¹⁴ i.e. Sc(BH₄)₄⁻ which then can be coordinated to an alkali metal cation to form a neutral bimetallic complex. The anionic transition metal borohydrides have been found to be less volatile than their neutral counterparts^14,18 and dehydrogenation of these more stable borohydrides may avoid the release of B₂H₆.

Studies investigating the dehydrogenation behavior of the Group I scandium borohydrides have revealed that upon heating, these compounds decompose into ScB₂ ^15,17 and the corresponding alkali borohydride.^15,18 While the predicted decomposition pathway of LiSc(BH₄)₄ consisted of numerous steps, B₂H₆ was not found to be a thermodynamically favorable intermediate or product.¹⁷ Multiple hydrogen releasing steps were also detected during the thermal dehydrogenation of NaSc(BH₄)₄ but the decomposed material was not characterized.¹⁵ In the case of KSc(BH₄)₄, two steps involving mass loss were observed which released a lower weight percent in total than the calculated hydrogen content of the starting material.¹⁸ This convinced the authors that the evolution of B₂H₆ was unlikely.

Although the results of these studies suggested that B₂H₆ formation was unlikely, quantitative analysis of the dehydrogenation products is needed as verification. As well, the decomposition pathways of the Group I scandium borohydrides are poorly understood. While in some cases the terminal products have been identified, evidence for multiple steps during dehydrogenation suggests that intermediate species may be formed. The presence of these intermediates may provide possible ways to manipulate the dehydrogenation path in order to achieve reversibility. We present a detailed characterization of the dehydrogenation products of the complex alkali scandium borohydrides under different dehydrogenation temperatures by nuclear magnetic resonance (NMR) spectroscopy and in situ infrared (IR) spectroscopy. Recently, the gaseous decomposition products of borohydride complexes have been analyzed by a combination of Fourier-transform infrared (FTIR) spectroscopy and thermogravimetry (TG) that was developed to circumvent the conventional issues preventing the quantitative measurement of gases by mass spectrometry.¹⁹ We utilize this method to determine whether B₂H₆ is released during decomposition of the Group I scandium borohydride complexes. The release of B₂H₆ is a critically important factor in determining the viability of a potential hydrogen storage material and thus its detection is an invaluable tool in evaluating the usefulness of the bimetallic borohydrides.

Experimental

NMR spectroscopy

Solid state Magic angle spin (MAS) NMR spectra were obtained on a Varian Inova spectrometer equipped with a 3.2 mm HX cross-polarization magic-angle spinning probe (Varian Chemagnetics, Ft. Collins, CO) at 128.3 and 97.2 MHz for ¹¹B and ⁴⁵Sc. Samples were packed into 3.2 mm zirconium oxide rotors and spun at 12 kHz. Single-pulse excitation was used with pulse widths of 1.0 and 6.0 μs for ¹¹B and ⁴⁵Sc respectively. Frequencies are reported with respect to boric acid for ¹¹B and aqueous ScCl₃ for ⁴⁵Sc (set at 0 ppm).

In situ TG–IR

Samples were loaded into a magnetic suspension balance (Rubotherm, Bochum) modified to allow for the measurements to take place under controlled gas flow. The system was connected to an infrared gas analyser (Bruker Alpha spectrometer equipped with a 8 cm gas cell at a resolution of 0.9 cm⁻¹). Dehydrogenation reactions were conducted under 150 mL min⁻¹ hydrogen flow at 1 bar. A detailed description of the experimental set-up can be found in a previous study.¹⁹

Synthesis of complex borohydrides

All sample preparation and manipulation was performed in an argon glovebox. Anhydrous ScCl₃ (Alfa Aesar, 99.9%) and either lithium (Alfa Aesar, 95%), sodium (Sigma-Aldrich, 99%), or potassium (Alfa Aesar, 98%) borohydride were added to 80 mL stainless steel vessels and milled under an argon atmosphere in a Fritsch Pulverisette 7 planetary mill with a 10 mm ball to sample mass ratio of approximately 35 : 1. Conditions for ballmilling are given in Table 1. The ratios of borohydride complex to ScCl₃ varied amongst the three different complexes. These individual ratios were found to give the highest yield of the respective bimetallic complex. Samples were characterized after ballmilling by IR spectroscopy and ¹¹B and ⁴⁵Sc MAS NMR.

Table 1 Mechanical milling conditions

Mixture	Molar ratio	Milling speed (rpm)	Milling time (h)
LiBH4 + ScCl3	4 : 1	350	10
NaBH4 + ScCl3	2 : 1	350	20
KBH4 + ScCl3	1.5 : 1	350	10

---

Thermal dehydrogenation studies

Alkali metal scandium borohydride complexes were dehydrogenated isothermally for NMR analysis on a Suzuki Shokan PCT-2SDWIN Sievert type apparatus. The samples were heated to 473 K with a customized heating sleeve in an initially evacuated fixed volume reactor for 24 hours.

Results and discussion

NMR characterization of starting complexes and dehydrogenated products

The preparation of bimetallic complexes by mechanical milling followed literature methods^19,20,22 and crystal structures have been determined for all of the complexes. This technique has been found to be the most successful way to prepare solvent-free materials.

LiBH₄–ScCl₃

The ¹¹B MAS NMR spectrum of the ballmilled sample is given in Fig. 1 along with the spectra for the dehydrogenated materials after trials conducted at 373, 473, and 573 K. Prior to dehydrogenation, the ballmilled powder consisted predominantly of LiSc(BH₄)₄, depicted as a strong resonance at −17 ppm in the ¹¹B NMR analysis with a secondary species representing residual LiBH₄ at −34 ppm.¹⁷ After dehydrogenation for 1 day at 373 K, very little change was observed in the ¹¹B MAS NMR of the products but dehydrogenation at 473 K found that the materials decomposed to LiBH₄. Dehydrogenation for 1 week at 573 K yielded the same result.

Fig. 1 ¹¹B MAS NMR of (a) LiSc(BH₄)₄ and then after dehydrogenation at (b) 373 K, 1 day; (c) 473 K, 1 day; (d) 573 K, 1 week.

The presence of a single resonance at 105 ppm in the ⁴⁵Sc MAS NMR (Fig. 2a) confirmed the complete reaction of ScCl₃ to form LiSc(BH₄)₄.¹⁷ Dehydrogenation at 373 K also did not present any change in the ¹¹B or ⁴⁵Sc MAS NMR results (Fig. 1b and 2b) but the powder exhibited a distinct color change from white to orange. A change in color is indicative of an electronic transition in the Sc and although the Sc spectra appear identical before and after dehydrogenation at 373 K, a small quantity of a species such as ScB_x was most likely formed.

Fig. 2 ⁴⁵Sc MAS NMR of (a) LiSc(BH₄)₄; (b) after dehydrogenation at 373 K.

At a higher dehydrogenation temperature of 473 K, however, the boron is completely converted to LiBH₄ (Fig. 1c). Prolonged dehydrogenation at 573 K produced similar results, confirming that the complex simply decomposes to the thermally stable Group I borohydride upon heating. Initial attempts to observe a ⁴⁵Sc signal were unsuccessful most likely due to quadrupolar broadening of the signal, an effect that would also account for the absence of a boride resonance in the ¹¹B MAS NMR. Subsequent analysis by high field ⁴⁵Sc MAS NMR at 850 MHz revealed the presence of two Sc species, possibly different ScB_x polymorphs.

The results of characterization by NMR confirmed that the starting material consisted of a mix of lithium scandium borohydride and lithium chloride¹⁴ (as well as unreacted LiBH₄) in the following reaction:

4LiBH₄ + ScCl₃ → LiSc(BH₄)₄ + 3LiCl (2)

Low temperature dehydrogenation succeeded in initiating enough decomposition to effect a change in color, again suggestive of a new Sc species such as ScB_x. Once the dehydrogenation temperature reached 473 K, the ¹¹B MAS NMR spectrum indicated that LiSc(BH₄)₄ had decomposed to LiBH₄. The decomposition very likely follows a path similar to that of LiZn₂(BH₄)₅ ¹⁹ which was found to disproportionate into the corresponding Group I and transition metal borohydrides,¹⁹ or LiBH₄ and the unstable Sc(BH₄)₃ in our case, followed by rapid decomposition of Sc(BH₄)₃. From the NMR data alone, it cannot be determined whether B₂H₆ is released. If no B₂H₆ were to form, the dehydrogenation reaction can be described as:

LiSc(BH₄)₄ → LiBH₄ + Sc(BH₄)₃ → LiBH₄ + ScB_x + 6H₂ (3)

The decomposition of Sc(BH₄)₃ to ScB_x is supported by first principles calculations.¹⁰

Dehydrogenation at both 673 and 723 K of ballmilled ScCl₃ and LiBH₄ has previously been found to produce a mixture of ScB₂, B₁₂H₁₂²⁻, and LiBH₄.^17,21 Our findings show that even the low temperature dehydrogenation promotes the decomposition of LiSc(BH₄)₄ to the more stable LiBH₄ along with the highly volatile Sc(BH₄)₃ species which immediately dehydrogenates to scandium boride. With higher temperatures, the mixture of LiBH₄ and scandium boride proceeds to decompose according to the well-studied lithium borohydride dehydrogenation pathway^1,22–24 which includes the formation of LiB₁₂H₁₂.

The powder that had been dehydrogenated for 1 day at 473 K was returned to the reaction vessel and hydrogenated for 2 days at 523 K under 12 MPa H₂ pressure. Surprisingly, the ¹¹B MAS NMR spectrum showed a small amount of B₁₂H₁₂²⁻ (−15 ppm,²¹ Fig. 3b), a species known to be a thermally stable decomposition product. The metal cation could be either Sc²⁺ or Li⁺ as both these species as B₁₂H₁₂²⁻ complexes have a similar chemical shift in ¹¹B MAS NMR.²¹ The formation of B₁₂H₁₂²⁻ may arise from the presence of a variety of boron compounds that are present in the ¹¹B MAS spectrum of dehydrogenated LiSc(BH₄)₄, appearing as a large broad feature that overlaps with the LiBH₄ resonance (Fig. 4). These unresolved signals likely represent an array of boron hydrides with similar chemical shifts which, upon heating, condense to form B₁₂H₁₂²⁻.

Fig. 3 ¹¹B MAS NMR of (a) LiSc(BH₄)₄ dehydrogenated 473 K, 1 day; (b) rehydrogenated at 523 K, 2 days, 12 MPa.

Fig. 4 ¹¹B MAS NMR of LiSc(BH₄)₄ dehydrogenated 473 K, 1 day with expanded scale.

Reversible dehydrogenation of LiSc(BH₄)₄ does not appear possible under these thermal conditions. While the decomposition of the complex occurs readily, the resulting formation of LiBH₄ will then require much higher temperatures to continue decomposing. Hydrogenation to regenerate the original borohydride is extremely unlikely as this would first require the decomposition of LiBH₄ and scandium boride to allow Sc(BH₄)₄⁻ to form. The presence of Li₂B₁₂H₁₂ after rehydrogenation at 523 K further emphasizes the preference for the system to convert to more thermodynamically stable species. Once Li₂B₁₂H₁₂ has formed, it is essentially impossible to hydrogenate this species to LiBH₄ and then to LiSc(BH₄)₄ under practical operating conditions.

NaBH₄–ScCl₃

Two primary resonances are evident in the ¹¹B MAS NMR spectra (Fig. 5a) for the ballmilled mixture, NaBH₄–ScCl₃, representing NaBH₄ at −37 ppm and a larger signal centered around −18 with a small shoulder that has been attributed to two different B environments in NaSc(BH₄)₄.¹⁵ More complete conversion to NaSc(BH₄)₄ has been demonstrated by ballmilling under the same conditions¹⁵ but in our case, even ballmilling an additional 20 hours did not increase the yield of NaSc(BH₄)₄. While the ¹¹B MAS NMR confirmed that more of the boron in the sample was found in NaSc(BH₄)₄ than in NaBH₄, the ⁴⁵Sc MAS NMR (Fig. 6a) showed that the major portion of the Sc content was bound in Na₃ScCl₆ (219 ppm) rather than NaSc(BH₄)₄ (115 ppm).¹⁵ Trace levels of an unidentified Sc species at 190 ppm was present as well. From the above results it appears that the ballmilling preparation produced:

4NaBH₄ + 2ScCl₃ → NaSc(BH₄)₄ + Na₃ScCl₆ (4)

although a significant amount of unreacted NaBH₄ remained. According to ¹¹B MAS NMR (Fig. 5b), the dehydrogenation at 373 K resulted in an increase in NaBH₄ with a concomitant decrease in NaSc(BH₄)₄. The ⁴⁵Sc MAS NMR spectrum also depicted a noticeable decrease in the Sc associated with NaSc(BH₄)₄ which, without evidence for any new Sc species, either formed more Na₃ScCl₆ or decomposed to scandium boride. The same change in color from white to orange that occurred when LiSc(BH₄)₄ was decomposed at 373 K was observed for this sample as well.

Fig. 5 ¹¹B MAS NMR of (a) NaSc(BH₄)₄; (b) after dehydrogenation at 373 K, 1 day; (c) after dehydrogenation at 473 K, 1 day; (d) rehydrogenation of powder from (c) at 523 K, 5 days, 12 MPa.

Fig. 6 ⁴⁵Sc MAS NMR of (a) NaSc(BH₄)₄; (b) after dehydrogenation at 373 K, 1 day; (c) after dehydrogenation at 473 K, 1 day; (d) rehydrogenation of powder from (c) at 523 K, 5 days, 12 MPa. Asterisks denote spinning sidebands.

At 473 K, dehydrogenation promoted the complete decomposition of NaSc(BH₄)₄ to NaBH₄ with a trace amount of an oxidized boron species at about 8 ppm (Fig. 5c), indicating some exposure of the sample to atmosphere. The NaSc(BH₄)₄ resonance in the ⁴⁵Sc MAS NMR spectrum was no longer visible (Fig. 6c) and only Na₃ScCl₆ was evident after dehydrogenation at 473 K. It is difficult to determine from the ⁴⁵Sc MAS NMR what the fate of the Sc released from NaSc(BH₄)₄ upon decomposition may have been. It is unlikely that it formed more Na₃ScCl₆ as this would require a Cl⁻ source and the ⁴⁵Sc MAS NMR of the starting material (Fig. 6a) gave no evidence for residual ScCl₃. In the case of LiSc(BH₄)₄, the Sc was ultimately decomposed to a species that could not be detected by NMR. If the same is assumed for NaSc(BH₄)₄ then decomposition proceeds via a similar route:

NaSc(BH₄)₄ → NaBH₄ + Sc(BH₄)₃ → NaBH₄ + ScB₃ + 6H₂ (5)

Rehydrogenation of NaSc(BH₄)₄ (dehydrogenated at 473 K, 1 day) for 5 days at 523 K under 12 MPa H₂ pressure yielded little change in the ⁴⁵Sc spectrum (Fig. 6d). In the ¹¹B spectrum, however, new species was present at −12 ppm. Similar to the rehydrogenation of LiSc(BH₄)₄, this represented B₁₂H₁₂²⁻ which has not been observed experimentally in dehydrogenation studies of NaBH₄. In this case, the counter ion must be Na⁺ since no additional resonances appeared in the ⁴⁵Sc MAS NMR (Fig. 6d). The formation of the species at −12 ppm may have emerged from hydrogenation of trace quantities of a variety of polyboranes at undetectable levels. The absence of NaSc(BH₄)₄ in either the ¹¹B or ⁴⁵Sc spectra verifies that the rehydrogenation pathway, at least under moderate conditions, does not lead to regeneration of the original bimetallic borohydride.

KBH₄–ScCl₃

The mixture of KBH₄–ScCl₃ after ballmilling consisted primarily of the new species KSc(BH₄)₄ at −22 ppm ¹⁸ and residual unreacted KBH₄ at −37 ppm as revealed in the ¹¹B MAS NMR data (Fig. 7a). In the ⁴⁵Sc MAS NMR, two predominant species were present, corresponding to K₃ScCl₆ at 218 ppm and KSc(BH₄)₄ at 103 ppm ¹⁸ (Fig. 8a). A third and much smaller resonance centered at 165 ppm was unidentified but has been observed in a previous reported synthesis of KSc(BH₄)₄.¹⁸ Mechanical milling appeared to have facilitated an analogous reaction to that of the preparation of NaSc(BH₄)₄ in spite of the different ratio of the initial Group I borohydride to ScCl₃, although for the KBH₄–ScCl₃ mixture less ScCl₃ may have been more appropriate.

Fig. 7 ¹¹B MAS NMR of (a) KSc(BH₄)₄; (b) after dehydrogenation at 373 K, 1 day; (c) after dehydrogenation at 473 K, 1 day; (d) rehydrogenation of powder from (c) at 523 K, 5 days, 12 MPa.

Fig. 8 ⁴⁵Sc MAS NMR of (a) KSc(BH₄)₄; (b) after dehydrogenation at 373 K, 1 day; (c) after dehydrogenation at 473 K, 1 day; (d) rehydrogenation of powder from (c) at 523 K, 5 days, 12 MPa. Asterisks denote spinning sidebands.

The only change in the ¹¹B MAS NMR spectrum after dehydrogenation at 373 K (Fig. 7b) was a trace signal at 0 ppm, possibly due to exposure to air in either the PCT reactor or NMR rotor resulting in the oxidation of a small amount of dehydrogenated KSc(BH₄)₄. In the ⁴⁵Sc MAS NMR, the resonance for K₃ScCl₆ at 218 ppm appeared to resolve into two overlapping peaks (Fig. 8b) attributable to the close chemical shifts of K₃ScCl₆ and residual ScCl₃ from an excess in the initial mixture.

After dehydrogenation at 473 K, any evidence of KSc(BH₄)₄ (−22 ppm) was absent in the ¹¹B MAS NMR spectrum and only a resonance for KBH₄ (−37 ppm) was seen (Fig. 7c). The KSc(BH₄)₄ also disappeared from the ⁴⁵Sc spectrum (Fig. 8c), leaving K₃ScCl₆ and ScCl₃ (as an overlapping shoulder) as the primary Sc species at 218 ppm. Rehydrogenation of the powder did not change the Sc characterization by MAS NMR but a small increase in the shoulder at −33 ppm, downfield from the KBH₄ resonance (Fig. 7d), was observed as well as the appearance of a resonance representative of a polyborane species at −15 ppm. Similar to what was observed from the Li and Na complexes, this new boron species most likely corresponded to a B₁₂H₁₂²⁻ salt.

The results of dehydrogenation and rehydrogenation trials on KSc(BH₄)₄ resembled those of NaSc(BH₄)₄. While dehydrogenation at 373 K caused a color change from white to orange, little change was seen in the NMR spectra. Heating at 473 K resulted in the complete decomposition of the bimetallic species to KBH₄ and either K₃ScCl₆ or ScB_x. The dehydrogenation process mostly likely involved initial decomposition to KBH₄ and Sc(BH₄)₃ followed by formation of ScB_x accompanied by the release of H₂. For this mixture, it is difficult to evaluate from the ⁴⁵Sc MAS NMR whether the Sc bound in KSc(BH₄)₄ formed additional K₃ScCl₆ or if it decomposed to a ScB_x material that could not be detected by either ¹¹B or ⁴⁵Sc MAS NMR. Rehydrogenation resulted in the formation of K₂B₁₂H₁₂, an indication once again of the lack of reversibility within these bimetallic complexes.

In situ TG–IR

In Fig. 9a–c the mass loss rates for the samples are presented. The mass loss rates are computed from the TG data (time derivative of mass loss, dotted line) and the time evolution of IR intensities lines (solid line). Specifically, for the TG data, the mass loss refers to the emission of all gaseous species from the sample; for the IR data, the mass loss is related to a specific IR peak, therefore to a specific compound, in this case, B₂H₆.

Fig. 9 TG data plotted as a function of rate of mass loss vs. time and temperature (open squares) with IR data (solid lines) representing diborane evolution for (a) LiSc(BH₄)₄; (b) NaSc(BH₄)₄; (c) KSc(BH₄)₄. The time evolution of the diborane IR signal for LiSc(BH₄)₄ is plotted in (d).

The IR analysis of LiSc(BH₄)₄ is plotted as a function of temperature (and therefore time) during linear ramping to 450 K and given in Fig. 9d. Although only analysis for LiSc(BH₄)₄ is shown, similar results were obtained for all three Sc complex borohydrides. Diborane was the only gas phase decomposition species detected in the IR spectra besides impurity levels of CO₂. Bands characteristic of B₂H₆ appeared at 1174 and 1604 cm⁻¹,²⁵ reaching a maximum in intensity by about 400 K followed by signal decay as the temperature continued to increase to 450 K. The gas-phase decomposition products of LiSc(BH₄)₄ were also analysed by mass spectrometry which confirmed that only B₂H₆ and H₂ were released upon heating to 453 K (ESI, Fig. S1†).

From Fig. 9a it is apparent that B₂H₆ was released quite early during the dehydrogenation period, beginning at about 350 K for both the IR and TG data. The initial loss of mass in the sample can be attributed to the release of B₂H₆ resulting in a weight loss of up to 0.5% of total sample mass. Between 400 and 450 K the B₂H₆ IR signal decayed to zero but the TG data reached a maximum within this temperature range. By 450 K B₂H₆ evolution was complete and continued heating was not associated with any B₂H₆ formation. The TG data, however, confirmed that the sample continued to lose mass after 450 K, reaching a second maximum in rate of weight loss at approximately 490 K. The rate of mass loss is still above zero at the cutoff temperature of 573 K, reaching a total mass loss of 4.75%.

At 373 K the ¹¹B and ⁴⁵Sc MAS NMR for LiSc(BH₄)₄ did not show any change but B₂H₆ evolution had already begun according to Fig. 9a. The initial step in decomposition must be representative of a reaction such as.

LiSc(BH₄)₄ → Li⁺ + Sc(BH₄)₄⁻ → LiBH₄ + B₂H₆ + ScB_x + 3H₂ (6)

where B₂H₆ formed due to the rapid decomposition of the unstable Sc(BH₄)₄⁻ anion. By 473 K, B₂H₆ was no longer formed and, according to ¹¹B MAS NMR, the primary boron species detected was LiBH₄ although this result was obtained after 24 hours of dehydrogenation which may have provided sufficient time for kinetically unfavorable processes to complete. Thus, during the TG–IR analysis, during the heat ramp to 573 K, there may have still been intact LiSc(BH₄)₄ present at 473 K. We confirmed this by ¹¹B MAS NMR analysis of LiSc(BH₄)₄ decomposed at 473 K and immediately cooled. The data exhibited a distinct resonance remaining for LiSc(BH₄)₄. The second step observed between 450 to 573 K (Fig. 9a) may simply represent the decomposition of the remaining LiSc(BH₄)₄.

Another possible source of mass loss that was not correlated to B₂H₆ release during decomposition of LiSc(BH₄)₄ may be due to kinetically fast processes. It was discussed earlier that there may have been an variety of polyborane species observed as a broad feature in the baseline of the ¹¹B MAS NMR spectrum for the decomposed LiSc(BH₄)₄ (Fig. 4) which, upon rehydrogenation, condensed to form B₁₂H₁₂²⁻ (Fig. 3b). These polyboranes may be a result of the pyrolysis of B₂H₆ which is slow below 473 K but becomes faster with higher temperatures.²⁶ The first step in thermal decomposition is the splitting of B₂H₆ into two BH₃ units²⁷ followed by the interaction of BH₃ with B₂H₆ to condense into larger polyboranes such as B₃H₇, B₄H₁₀, and B₅H₁₁ ²⁶ while releasing H₂. These species are all gaseous polyboranes but were not detected by IR analysis because of rapid kinetics in the gas phase, resulting in the condensation of these polyboranes into the solid state species that were detected in the baseline of ¹¹B MAS NMR spectrum (Fig. 4) after decomposition at 473 K.

The TG–IR data for NaSc(BH₄)₄ (Fig. 9b) indicated that B₂H₆ was again released early in the decomposition although the intensity of the IR signal was considerably lower than the in the case of LiSc(BH₄)₄. The maximum B₂H₆ evolution occurred around 373 K at which an increase in the amount of NaBH₄ was observed relative to NaSc(BH₄)₄ in the ¹¹B MAS NMR (Fig. 5b). It is interesting to note that that while the ¹¹B MAS NMR showed more decomposition occurring for NaSc(BH₄)₄ than for LiSc(BH₄)₄ after dehydrogenation at 373 K, a smaller mass loss was observed for NaSc(BH₄)₄. Two subsequent steps occurred after B₂H₆ evolution ceased, both associated with greater rates of mass loss than the initial one.

At temperatures greater than 450 K, B₂H₆ was no longer detected for the decomposition of NaSc(BH₄)₄. As discussed for LiSc(BH₄)₄, the subsequent change in mass after B₂H₆ evolution ceased may result from the decomposition of B₂H₆ that is facilitated by higher temperatures. The two steps that occur between 400–573 K in the TG data probably represent the decomposition (resulting in H₂ release) of different temperature-dependent borane species formed upon condensation.

For KSc(BH₄)₄, decomposition occurred over two steps, the initial step being entirely associated with B₂H₆ release, while the subsequent step observed at higher temperature was associated with much higher rates of mass loss. The initial B₂H₆ evolution is very similar to that of NaSc(BH₄)₄ with regards to both temperature range, intensity of the IR signal, and rate of mass loss. Overall, KSc(BH₄)₄ lost the least mass at 1.9 wt% versus 2.5 wt% and 4.8 wt% for NaSc(BH₄)₄ and LiSc(BH₄)₄, respectively. This may reflect in part the stabilities of the Group I borohydrides: LiBH₄ has the lowest stability of the three complexes and also exhibited the greatest mass loss. The variations seen in the TG data given in Fig. 9 may also arise from differences in how the metal cations stabilize polyborane species produced upon decomposition.

The combination of results from NMR, TG and IR experiments confirm that B₂H₆ is formed from all three Sc complexes and decomposition occurs as according to eqn (6). The B₂H₆ evolution does not necessarily stop at higher temperatures but B₂H₆ may decompose rapidly to form higher boranes, both gas and solid phase, releasing H₂ in the process. The low overall weight loss of all three complexes suggests that even if the majority of the loss was due to dehydrogenation, the quantity is insufficient to provide a useful amount of hydrogen for applications where an energy dense fuel source is needed.

Conclusions

In comparing the three bimetallic Sc complexes examined in our study, LiSc(BH₄)₄ released the most B₂H₆ and lost the most mass. This suggests that the Li cation had the least stabilizing effect on Sc(BH₄)₄⁻ and the complex readily decomposed to LiBH₄ and the volatile Sc(BH₄)₃ while releasing B₂H₆. The Na and K complexes, both of which form more stable monocation borohydrides than Li, appeared to suppress the emission of B₂H₆ but also lost less mass. As well, all three complexes decomposed to the stable Group I borohydride, the thermodynamic driving force for the reaction, leading to the formation of the tetrahedrally distorted Sc(BH₄)₃ and finally B₂H₆.

While all three bimetallic Sc complexes demonstrated onset decomposition temperatures that fell between that of the parent Group I and transition metal borohydrides, B₂H₆ was consistently released within the initial period of decomposition, eliminating any chance of reversibility. The formation of B₂H₆ is a critical factor in determining whether reversible dehydrogenation is a possibility and our results confirm that this species is a major component of the decomposition products. In addition, the formation of stable alkali metal borohydrides and transition metal borides make rehydrogenation improbable, at least within the limits of practical application. Our results highlight the importance of gas phase analysis in characterizing the dehydrogenation products of borohydride complexes. These findings also emphasize the need for multiple analytical tools for the thorough evaluation of materials with regards to practical hydrogen storage.

Acknowledgements

This work was supported by the DOE DEERE, by the European Commission, Grant agreement no. FP7-284522 (infrastructure program H2FC), by the Swiss Federal Office of Energy through the project ACH and by CCEM and Swisselectric research through the HyTech project. The research leading to these results has received funding from the Fuel Cells and Hydrogen Joint Undertaking under BOR4STORE Grant 303428.

Notes and references

1. A. Züttel, S. Rentsch, P. Fischer, P. Wenger, P. Sudan, P. Mauron and C. Emmenegger, J. Alloys Compd., 2003, 356–357, 515.
2. S. Orimo, Y. Nakamori, G. Kitahara, K. Miwa, N. Ohba, S. Towata and A. Züttel, J. Alloys Compd., 2005, 404–406, 427.
3. M. Au, A. R. Jurgensen, W. A. Spencer, D. L. Anton, F. E. Pinkerton, S.-J. Hwang, C. Kim and R. C. Bowman, J. Phys. Chem. C, 2008, 112, 18661.
4. A. F. Gross, J. J. Vajo, S. L. Van and G. L. Olson, J. Phys. Chem. C, 2008, 112, 5651.
5. Z. Z. Fang, P. Wang, T. E. Rufford, X. D. Kang, G. Q. Lu and H. M. Cheng, Acta Mater., 2008, 56, 6257.
6. S. Cahen, J.-B. Janot and J.-M. Tarascon, J. Power Sources, 2009, 189, 902.
7. J. J. Vajo, S. L. Skeith and F. Mertens, J. Phys. Chem. B, 2005, 109, 3719.
8. H. R. Hoekstra and J. J. Katz, J. Am. Chem. Soc., 1949, 71, 2488.
9. T. J. Marks and G. R. Kolb, Chem. Rev., 1977, 77, 263.
10. Y. Nakamori, K. Miwa, A. Ninomiya, H. Li, N. Ohba, S. Towata, A. Züttel and S. Orimo, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 1.
11. G. N. Schrauzer, Naturwissenschaften, 1955, 42, 438.
12. J. R. Eliseo, Synthesis and Characterization of Novel Anionic Transition Metal Borohydrides, Masters Thesis, University of Hawaii, 2007.
13. D. B. Ravnsbæk, Y. Filinchuk, Y. Cerenius, H. J. Jakobsen, F. Besenbacher, J. Skibsted and T. R. Jensen, Angew. Chem., Int. Ed., 2009, 48, 6659.
14. H. Hagemann, M. Longhini, J. W. Kaminski, T. A. Wesolowski, R. Cerný, N. Penin, M. H. Sørby, B. C. Hauback, G. Severa and C. M. Jensen, J. Phys. Chem. A, 2008, 112, 7551.
15. R. Cerný, G. Severa, D. B. Ravnsbæk, Y. Filinchuk, V. D. Anna, H. Hagemann, C. M. Jensen and T. R. Jensen, J. Phys. Chem. C, 2010, 114, 1357.
16. H.-W. Li, S. Orimo, Y. Nakamori, K. Miwa, N. Ohba, S. Towata and A. Züttel, J. Alloys Compd., 2007, 446–447, 315.
17. C. Kim, S. Hwang, R. C. Bowman, J. W. Reiter, J. A. Zan, J. G. Kulleck, H. Kabbour, E. H. Majzoub and V. J. Ozolins, J. Phys. Chem. C, 2009, 113, 9956.
18. R. Cerný, D. B. Ravnsbæk, G. Severa, Y. Filinchuk, V. D. Anna, H. Hagemann, J. Skibsted, C. M. Jensen and T. R. Jensen, J. Phys. Chem. C, 2010, 114, 19540.
19. A. Borgschulte, E. Callini, B. Probst, A. Jain, S. Kato, O. Friedrichs, A. Remhof, M. Bielmann, A. J. Ramirez-Cuesta and A. Züttel, J. Phys. Chem. C, 2011, 115, 17220.
20. G. Severa, H. Hagemann, J. W. Kaminski, T. A. Wesolowski and C. M. Jensen, J. Phys. Chem. C, 2010, 114, 15516.
21. S.-J. Hwang, R. C. Bowman, J. J. W. Reiter, J. Rijssenbeek, G. L. Soloveichik, J.-C. Zhao, H. Kabbour and C. C. Ahn, J. Phys. Chem. C, 2008, 3164.
22. N. Ohba, K. Miwa, M. Aoki, T. Noritake, S. Towata and A. Züttel, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 075110.
23. S.-J. Orimo, Y. Nakamori, N. Ohba, K. Miwa, M. Aoki, S. Towata and A. Züttel, Appl. Phys. Lett., 2006, 89, 021920.
24. A. Züttel, A. Borgschulte and S.-J. Orimo, Scr. Mater., 2007, 56, 823.
25. W. E. Anderson and E. F. Barker, J. Chem. Phys., 1950, 18, 698.
26. M. Söderlund, P. Mäki-Arvela, K. Eränen, T. Salmi, R. Rahkola and D. Y. Murzin, Catal. Lett., 2005, 105, 191.
27. E. L. Muetterties and W. H. Knoth, Polyhedral Boranes, M. Dekker, New York, 1968, p. 197.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra10352a

‡ Current address: Hawaii Natural Energy Institute, University of Hawaii at Manoa, 1680 East West Road, Honolulu, HI, 96822, USA.

---