Regeneration of Ammonia Borane Spent Fuel by Direct Reaction with Hydrazine and Liquid Ammonia

See allHide authors and affiliations

Science  18 Mar 2011:
Vol. 331, Issue 6023, pp. 1426-1429
DOI: 10.1126/science.1199003


Ammonia borane (H3N-BH3, AB) is a lightweight material containing a high density of hydrogen (H2) that can be readily liberated for use in fuel cell–powered applications. However, in the absence of a straightforward, efficient method for regenerating AB from dehydrogenated polymeric spent fuel, its full potential as a viable H2 storage material will not be realized. We demonstrate that the spent fuel type derived from the removal of greater than two equivalents of H2 per molecule of AB (i.e., polyborazylene, PB) can be converted back to AB nearly quantitatively by 24-hour treatment with hydrazine (N2H4) in liquid ammonia (NH3) at 40°C in a sealed pressure vessel.

A critical factor in realizing a hydrogen (H2) economy is storage of the molecule for controlled delivery, presumably to an energy-producing fuel cell (1). Options for H2 storage include compressed hydrogen, metal hydrides (2), and porous sorbent materials (3). In addition to these, there is also the possibility of using chemical hydrides wherein the H2 is stored in E–H bonds (E: C, B, N, or O). In this case, hydrogen can be released thermally or through the use of an acid, base, or metal catalyst. Ammonia borane (H3N-BH3, AB) is a particularly appealing chemical hydrogen storage medium, owing to its high gravimetric capacity of H2 (19.6 weight percent), low molecular mass (30.7 g mol−1), and overall chemical properties (1). AB has both hydridic and protic hydrogens, facilitating H2 release under mild conditions. Methods explored for promoting release include acid or base treatment, as well as transition-metal catalysis (48). At the same time, the viability of any storage system is critically dependent on efficient recyclability, but reports on the latter subject are sparse and the processes generally energy intensive (1, 913).

The composition of spent fuel depends on the dehydrogenation method used (1, 14-16). Initial transition metal–catalyzed dehydrogenation of AB with rhodium catalysts such as [Rh(cod)Cl]2 (cod: 1,5-cyclooctadiene) produced the intermediates cyclotriborazine and μ-aminodiborane before forming borazine, polyborazylene (PB, Fig. 1), and other oligomers as the final dehydrogenated material (17, 18). More recent transition-metal catalyst systems based on Ir(H2)POCOP [POCOP: κ3-1,3-(OPtBu2)2C6H3] form oligomeric polyaminoboranes (19) as the exclusive product, but these are limited by the release of only one equivalent of H2 (5). However, transition metals supported by N-heterocyclic carbenes (NHC), e.g., Ni(NHC)2 complexes, can release greater than two equivalents of H2 and give PB as a single spent-fuel component (6). We have predominantly pursued methods for regenerating PB as this is the form of spent fuel that results from loss of greater than two equivalents of H2 per molecule of AB. Nonetheless, the presence of residual catalyst in PB mixtures could complicate the AB regeneration process, so we have used a purer form of PB that is derived from dehydrogenation of borazine. PB is not a simple polymeric material, instead consisting of a mixture of oligomers. Consequently, PB can exhibit variation in molecular weight distribution and empirical formulae depending on the conditions used for its synthesis, and is composed of partially cross-linked structures and structures containing fused or polycyclic ring systems. As a result of these factors, the 11B NMR (nuclear magnetic resonance) spectrum of PB exhibits a broad, featureless resonance centered around 31 parts per million in tetrahydrofuran (THF) due to the range of boron environments present (20, 21).

Fig. 1

A representative structure of PB.

An important consideration here is that both catalytic and thermal (i.e., noncatalytic) routes to PB from AB (based on the volume of H2 evolved) yield a hydrogen-depleted material in which greater than two equivalents of hydrogen are released from AB. The PB generated in either methodology to this point appears to be indistinguishable from the other. This leads to the general acceptance of the formulation of PB as “BNHx” where x is greater than zero, less than two, but typically one (20). This material thus corresponds to a spent fuel derived from the loss of greater than two equivalents of H2 per molecule of AB, but still allows for the preservation of some of the energetic B–H bonds. This latter property confers both solubility for subsequent recycle chemistry as well as reduction of the overall energy required to regenerate the desired BH3 fragments. Although synthetically prepared PB derived from borazine is a surrogate for real spent fuel, it represents a scenario in which as much H2 has been liberated downstream of AB before ceramic BN is formed. This makes it an ideal material with which to demonstrate the conversion of this spent fuel form to AB.

Owing to the thermodynamics of AB dehydrogenation, in which the energy required to dehydrogenate AB to PB, liberating 2.5 equivalents of hydrogen, is −57.7 kJ mol−1 (2225), direct hydrogenation of spent fuel would require a substantial energy input (4, 2225). Because a simple on-board direct regeneration is unlikely, a more viable approach would be to perform the regeneration off-board in a series of reactions that are more thermodynamically reasonable. We recently described (13) the use of ortho-benzenedithiol (BDT) for recycling PB. This approach avoided the formation of thermodynamically stable B–O bonds, moieties that require high-energy reducing agents to reform the BH3 units (recycling B–O bonds is somewhat analogous to recycling CO2 to gasoline). This sulfur-based reagent was selected because B–S bonds are weaker than the analogous B–O bonds, and the acidity of the S–H moiety was likely to facilitate breakdown of the polymeric PB structure into molecular fragments for subsequent reduction chemistry. The process proved successful on a 50-mg scale with subsequent conversion of the (BDT)B-H species back into AB, which was accomplished with Sn-H reagents (26). To understand how this approach might translate from the laboratory to an industrial plant environment (100 metric tons/day recycle of AB), an economic analysis was carried out on the thiolate/tin based chemistries (27). This assessment showed that the mass of the tin reductants (e.g., Bu3SnH, 291.06 g mol−1) was a major contributor to the overall plant cost due to the physical transport of this high molecular weight material. It was therefore apparent from this analysis that one way to increase the process efficiency would be to use a substantially lighter reductant.

On the basis of this information, we thus endeavored to find a simplified, more energy-efficient process for AB recycle involving a minimum number of steps. In considering potential reductants for this purpose, hydrazine (N2H4, 32.05 g mol−1) appeared to be a good candidate because of its low molecular weight. In the sense of a redox reaction, hydrazine can decompose as N2H4 → N2 + 2H+ + 2H resulting in two protons and hydridic hydrogen atoms and diatomic nitrogen gas (N2) as the likely potential by-product of any reduction chemistry, which lends itself well to potential capture and recycle. Because PB is soluble in polar organic solvents, our initial studies using N2H4 were carried out in THF solution at room temperature (28). The PB was converted by the hydrazine to species that only contained BH3 units. However, the major product was hydrazine-borane (N2H4-BH3, HzB), with smaller amounts of AB as determined by 11B{1H} NMR even when substoichiometric amounts of hydrazine were used and residual PB remained. The development of HzB as a potential H2 storage material has been investigated. The first report on the thermal decomposition of HzB was by Ricker and Goubeau in 1961 (29), and more recently Lentz and co-workers improved the H2-release behavior by addition of LiH (30). However, one main problem highlighted in this later report is the inability to identify the products after dehydrogenation and the subsequent lack of regeneration strategies for this unknown material. It therefore became apparent that conversion of HzB into AB would be a key reaction to achieve appreciable regeneration of a H2 storage material. When HzB was dissolved in a vast excess of liquid NH3 at room temperature (~150 psi at 25°C), no exchange of the N2H4 for NH3 was observed; rather, only HzB was observed in the 11B NMR spectrum, even after 14 days. Surprisingly, direct heating of HzB in a 0.5 M NH3 solution in dioxane gave improved yields of AB, up to 25% as indicated by 11B NMR spectroscopy. Thus, we conjectured that energy in the form of heat was necessary to aid the scission of the B–N bond in HzB with subsequent formation of AB, and we attempted the reaction of HzB in a sealed vessel with liquid NH3 (60°C) for 24 hours. After the NH3 was allowed to evaporate, a trace amount of N2H4 remained, and the predominant boron-containing species was AB (~85% yield by 11B NMR spectroscopy), the only other observable boron-containing species being HzB. If the reaction work-up conditions were changed to include removal of the volatiles (including remaining hydrazine, boiling point 114°C, vapor pressure 14.4 torr at 25°C), under high vacuum the yield could be improved to 95% isolated ammonia borane. The residual hydrazine does play a role in the conversion of AB to HzB. The exchange in neat hydrazine shows complete conversion to HzB after 12 hours at 60°C, and this reaction is slowed as the temperature is lowered (2% after 12 hours at 25°C) (28), consistent with some form of kinetic control. These results are consistent with high-level CCSD(T)–complete basis set calculations, which show that the B–N dative bond energy in AB is 4.5 kcal/mol less than the B–N dative bond energy in HzB at 298 K. Using a self-consistent reaction field approach (31) with the COSMO (conductor-like screening model) parameterization (32) as implemented in two different electronic structure packages (Gaussian and ADF) (3335), we calculated the energy of the reaction in NH3, THF, and dioxane. The two computational implementations differ slightly, but in both cases the free energy for the exchange reaction is substantially decreased in NH3 as a solvent with free-energy differences of 1.9 to 1.0 kcal/mol. As the dipole moment of the solvent decreases, the free energies approach the gas-phase value. Thus, BH3NH3 is more stabilized in the polar solvents, leading to a decrease in the free energy.

This result led us to hypothesize that, if in THF solution HzB was the predominant product in the reaction of PB with N2H4, the coordinated N2H4 moiety should be readily displaced from any N2H4-BH3 formed in situ with liquid NH3 at elevated temperatures. This would drive the formation of the desired AB product while freeing N2H4 for further reduction of the B–H bonds in unreacted PB. This procedure would not only enable us to minimize the amount of N2H4 required to recycle AB, but also greatly improve production and recycle costs by reducing separation steps, the only by-product being N2, which itself could potentially be recovered. Indeed, carrying out the reaction of PB and N2H4 in liquid NH3 under slightly milder conditions (40°C) for 24 hours resulted in quantitative conversion to BH3-containing species. 11B NMR spectroscopic investigation of the crude material after removal of NH3 and the remaining N2H4 under dynamic vacuum revealed quantitative conversion to BH3 compounds, with ~92% being AB with the remainder as HzB (Fig. 2). Differential scanning calorimetry on the AB produced by this method afforded identical results to that performed on an authentic sample of AB (36). In an ideal situation, 4 “BNH” + 5N2H4 → 4H3N-BH3 + 5N2, meaning that every 1 mol of PB requires 1.25 mol of N2H4 to fully regenerate AB from spent fuel. This reaction energy can be estimated to be −64.7 kcal/mol for the production of solid AB from liquid “BNH” [see (28) for details]. As a result, a slight excess of hydrazine was used (1.35 equivalents of N2H4) to compensate for differing extents of cross-linking in PB. After AB regeneration, the NH3/N2H4 that is removed from the reaction can be readily recycled and, if necessary, a straightforward separation of these two components can be carried out.

Fig. 2

11B{1H} NMR spectrum in THF for crude material isolated from one-pot PB/N2H4/NH3 reaction after heating at 40°C for 24 hours, resulting in HzB (8%) and AB (92%).

To confirm the necessity of hydrazine as a reductant, we also heated PB in liquid NH3 at 60°C in the absence of N2H4 for 14 days. We observed no conversion to AB, and only starting material was recovered at the end of the reaction period.

To further understand the mechanism of this conversion, we calculated the energies of the intermediates for the addition of N2H4 molecules to a reasonable model system [i.e., “BNH2” (borazine)] at the G3MP2B3 level (37) of theory (28). The overall reaction sequence is complicated owing to the presence of many reaction steps (more than a hundred reactions). The computational results indicate that the initial reactions that lead to decomposition of the ring are near thermoneutral and that subsequent reactions are exothermic. The overall reaction for conversion of “BNH2” isc-B3N3H6 + 3N2H4 → 3BH3NH3 + 3N2↑and the reaction is slightly endothermic (ΔH = 5.7 kcal/mol) in the gas phase and exothermic (ΔH = −52.5 kcal/mol) if the c-B3N3H6 and BH3NH3 are in the solid state (using the best available values).

For this route to become an industrially viable process for H2 storage, the worldwide production volume of hydrazine would need to be either increased or produced via a new process. Hydrazine is produced primarily by the Olin-Raschig process (from NH3, NaOH, and Cl2) as an aqueous solution for less than $4 a kilogram and has a wide variety of industrial uses, including in nuclear fuel reprocessing, as a fuel in rocket propulsion, as a boiler feedwater deoxygenating agent, and in the manufacture of foamed plastics, pharmaceuticals, and biodegradable pesticides and herbicides (38). However, with worldwide production capacity estimated to be ~200,000 metric tons/year in 2007, the limited production volume and limited applications make current pricing less relevant. Hydrazine production colocated with recycle plants would reduce the transportation and storage costs and reduce the cost of hydrazine further, although a more efficient method for hydrazine synthesis would be of greater benefit.

The overall reaction cycle, including the production of hydrazine, is summarized in Fig. 3. In this scheme, the N2 released from the N2H4 can be rehydrogenated to NH3 and subsequently converted to N2H4. Although synthetic procedures such as the Olin-Raschig process produce hydrazine in high yields and great efficiency, we can begin to look to the future and think about a new route to hydrazine production. In conjunction with an AB-regeneration plant, in situ N2H4 production from NH3 does not require high conversion rates to produce a N2H4/NH3 feed for AB regeneration. In this scheme, the only material consumed is hydrogen, the production of which is widely recognized as the second major technical challenge for the hydrogen economy.

Fig. 3

Ideal overall reaction scheme for AB (H3N-BH3) regeneration from PB (“BNH”) with hydrazine (N2H4).

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Tables S1 to S4


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. This work was funded by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy. The authors and Los Alamos National Laboratory have filed a patent on the methods presented herein. We thank L. Sneddon for invaluable discussions.

Stay Connected to Science

Navigate This Article