Chemical activation of MgH₂; a new route to superior hydrogen storage materials

Abstract

We report the discovery of a new, chemical route for ‘activating’ the hydrogen store MgH₂, that results in highly effective hydrogen uptake/release characteristics, comparable to those obtained from mechanically-milled material.

---

Hydrogen is widely regarded as a promising alternative to carbon-based fuels; it can be produced from a variety of renewable resources and, when coupled with fuel cells, offers the prospect of near-zero emission of pollutants and greenhouse gases.¹ However, the development of hydrogen as a major energy carrier will require solutions to many scientific and technological challenges.² Foremost among these is the issue of hydrogen storage; the lightest of all the chemical elements has an excellent energy content per unit weight, but a low energy content per unit volume.³

The current generation of solid state hydrogen stores employ metal hydrides (e.g. LaNi₅H₆) that have excellent volumetric storage densities—higher than for both compressed gas and liquid hydrogen—but which have poor gravimetric storage densities (1.37 wt% for LaNi₅H₆), thereby precluding their use for mobile storage applications in hydrogen fuel-cell vehicles, for which a capacity of 6–7 wt% is regarded as a minimum requirement.² Magnesium hydride offers the highest energy density of all reversible hydrides applicable for hydrogen storage.⁴ However, hydrogen adsorption/desorption kinetics are too slow to form the basis of a practical hydrogen store. It has been demonstrated previously by a number of groups that the hydrogen sorption kinetics of Mg and Mg-based alloys can be greatly improved if the materials are prepared with a nanocrystalline microstructure through high velocity ball milling.⁵ However, this technique is energy intensive, especially for batch-milling large quantities of material.

Recently, a number of new hydrogen storage systems have been proposed⁶ based on the interaction between light metal hydrides and amides. Luo reported⁷ that a 1 ∶ 2 mixture of MgH₂ and LiNH₂ desorbed hydrogen reversibly at around 200 °C. Here we report a dramatic enhancement in the absorption/desorption kinetics of MgH₂ through reaction with small amounts of LiBH₄.⁸

MgH₂ and LiBH₄ (Sigma-Aldrich) were mixed together in a mole ratio of 1 ∶ 0.1 in an argon atmosphere glove box (O₂ content < 10 ppm) and placed in a quartz tube. The tube was capped with a Young's tap, via an Ultra Torr fitting, removed from the glove box and placed on a vacuum line, where it was evacuated to 10⁻⁶ mbar before being sealed with a gas torch. The sealed evacuated tube was then placed in a muffle furnace and heated at 300 °C for 12 hours.

For comparison, MgH₂ (typical size ∼ 100 µm) was milled at 300 rpm (PM400 Retsch planetary ball mill) for 15 hours in an argon atmosphere in a stainless steel milling pot (250 ml) using 8 mm balls with a 10 ∶ 1 ball to powder ratio. All samples were prepared for characterization in the glove box.

The hydrogen desorption kinetics of MgH₂/LiBH₄ after five hydrogen desorption and reabsorption cycles are shown in Fig. 1, compared to MgH₂ and MgH₂ milled for 15 h (i, ii and iii respectively).⁹ It is apparent that MgH₂/LiBH₄ desorbed hydrogen significantly faster than both the MgH₂ and milled MgH₂. The average desorption rates, therefore, were 0.28 × 10⁻² wt% s⁻¹, 0.01 × 10⁻² wt% s⁻¹ and 0.12 × 10⁻² wt% s⁻¹ for MgH₂/LiBH₄, MgH₂ and milled MgH₂ respectively.¹⁰

Fig. 1 Hydrogen desorption kinetics at 300 °C under 10 mbar H₂ of i) MgH₂/LiBH₄, ii) as-received MgH₂ and iii) MgH₂ milled for 15 h; all samples measured on the sixth hydrogen desorption.

The subsequent sixth hydrogen absorption at 300 °C and 10 bar H₂ for the same three materials is shown in Fig. 2. Again the MgH₂/LiBH₄ material reabsorbed hydrogen faster than MgH₂, and at a rate comparable to MgH₂ milled for 15 hours. The wt% values of hydrogen absorbed at 60 min for MgH₂/LiBH₄, MgH₂ and ball milled MgH₂ were 5.86, 4.86 and 6.01 respectively, and the absorption rates within this time scale were 1.03 × 10⁻² wt% s⁻¹, 0.16 × 10⁻² wt% s⁻¹ and 1.24 × 10⁻² wt% s⁻¹ respectively.¹¹

Fig. 2 Hydrogen absorption kinetics at 300 °C under 10 bar H₂ of i) MgH₂/LiBH₄, ii) MgH₂ and iii) MgH₂ milled for 15 h; all samples measured on the sixth hydrogen absorption.

Clearly, the reaction of MgH₂ with quite modest amounts of LiBH₄ resulted in a spectacular increase in the kinetics of hydrogen desorption/absorption. This improvement, however, was not apparent in the ‘as-prepared’ material: importantly, the rate of hydrogen desorption/absorption increased with each successive cycle for the first 4–5 cycles before stabilizing. Both the heat treatment and the initial cycling were necessary to activate the material: cycling a simple mixture of MgH₂ and LiBH₄ did not result in such an ‘active’ material with fast kinetics.

There are a number of possible factors that could contribute to the observed enhancement in hydrogen absorption/desorption properties. Powder XRD studies revealed that MgH₂/LiBH₄ contained only phases with the tetragonal α-MgH₂ structure (rutile).¹² Evidently a change in the crystal structure of MgH₂ is not responsible. The XRD pattern of the as-prepared material exhibited narrow, split diffraction lines, indicating that the material contained two different phases with the same structure, present in roughly equal amounts. After cycling, the active material was apparently single phase but with noticeably broader lines, and a unit cell slightly expanded (∼0.8%) from that of as-received MgH₂.¹³ This expansion could contribute to enhanced diffusion of H atoms in Mg/MgH₂.

Previously observed enhancements in the hydrogen absorption/desorption kinetics of Mg and Mg-based alloys after ball milling have been attributed to a nanocrystalline microstructure with an increased volume of grain boundary material, which acts as a preferred pathway for hydrogen diffusion.¹⁴ Changes in sample morphology and particle size may also be important. SEM images of MgH₂/LiBH₄, before and after six hydrogen desorption/absorption cycles, are shown in Fig. 3, along with those of as-received and milled MgH₂.¹⁵ Before cycling, MgH₂/LiBH₄ (Fig. 3i(c)) had a very similar appearance to MgH₂ (Fig. 3i(a)) consisting of large particles of up to 100 µm with a smooth surface. After six cycles of repeated crystallization of Mg and MgH₂, some finer particles were visible in MgH₂, but a significant proportion of larger lumps remained (Fig. 3ii(a)). MgH₂/LiBH₄ (Fig. 3ii(c)) in contrast consisted of a fine, apparently porous material of coral-like appearance with characteristic length scale ca. 2 µm. This is roughly the same as the particle size of MgH₂ milled for 15 hours (Fig. 3i(b) & ii(b)), although the morphologies were noticeably different.

Fig. 3 Scanning electron microscope images, i) before hydrogen desorption of a) MgH₂, b) MgH₂ milled for 15 h and c) MgH₂/LiBH₄ and ii) taken after the sixth hydrogen reabsorption of a) MgH₂, b) MgH₂ milled for 15 h and c) MgH₂/LiBH₄.

The fine powder structure observed may be a factor in the fast kinetics of both MgH₂/LiBH₄ and milled MgH₂ materials, but this may not be the most important factor in the former, as it differs chemically, owing to the presence of LiBH₄. Analysis of MgH₂/LiBH₄ after the sixth hydrogen absorption by XPS¹⁶ indicated that lithium and boron were still present in the sample. The FTIR spectrum¹⁷ of MgH₂/LiBH₄ before cycling exhibited peaks at 1126 cm⁻¹, which corresponds to a BH₂ deformation, and four peaks at 2225 cm⁻¹, 2238 cm⁻¹, 2291 cm⁻¹ and 2386 cm⁻¹, corresponding to B–H_t (terminal) stretching.¹⁸ This may be compared to LiBH₄, which showed peaks at 1120 cm⁻¹ (BH₂ deformation) and 2220 cm⁻¹, 2231 cm⁻¹, 2282 cm⁻¹ and 2381 cm⁻¹ (B–H_t stretches). The spectrum of the same material after six hydrogen desorption/absorption cycles contained no B–H vibrations, suggesting that BH₄⁻ anions did not persist at this stage. This conclusion was supported by proton-decoupled solid state ¹¹B NMR spectra¹⁹ of the MgH₂/LiBH₄ before and after cycling. Before hydrogen desorption the spectrum was very similar to that of LiBH₄ with a single central transition at ca. −41 ppm. After the first hydrogen desorption the spectrum was transformed with peaks at 12.7 ppm, 0.9 ppm, −16.0 ppm and −43.3 ppm, and retained multiple peaks on subsequent absorption/desorption cycles.

The decomposition of BH₄⁻ after the first hydrogen desorption at 300 °C is to be expected as compounds such as LiBH₄ and NaBH₄ are well known to lose hydrogen at temperatures lower than this.²⁰ The fact that the anion did not reform on subsequent hydrogen adsorption, however, is significant, indicating that LiBH₄ served as a convenient method for introducing both lithium and boron into the system, but that BH₄⁻ compounds were not present in the active material. In fact, it was possible to make a material with identical properties starting from a mixture of magnesium powder and LiBH₄. Although the presence of both Li and B in MgH₂/LiBH₄ after cycling is not in doubt, their exact chemical role in the fast hydrogen absorption/desorption has not yet been identified. One suggestion is that these elements might have a catalytic effect at the surface of the magnesium or magnesium hydride particles, but preliminary XPS data suggest that boron enters the subsurface layers of the material and is not just confined to the surface.

Doping of boron and/or lithium into MgH₂ could also be consistent with the small increase observed in lattice parameters and could change the thermodynamic properties of the compound. So far, however, there is little evidence for significant thermodynamic differences from MgH₂, as there was no observed shift in the hydrogen plateau pressure at 300–350 °C. The observation that use of NaBH₄ in the reaction instead of LiBH₄ resulted in no improvement in the hydrogen absorption/desorption kinetics, however, suggests that lithium doping of MgH₂ could be significant. The presence of Li⁺ in the MgH₂ lattice might lead to hydride anion vacancies which could facilitate diffusion of hydrogen in the hydride phase. More detailed characterization including measurements of thermodynamic parameters and microstructural studies of MgH₂/LiBH₄ is in progress.

We gratefully acknowledge The European Framework V (FUCHSIA) and EPSRC (SUPERGEN) for funding and we thank colleagues at the Johnson Matthey Technology Centre, UK for technical support.¹⁶

Notes and references

1. P. Hoffman, Tomorrow's Energy: Hydrogen, Fuell Cells and The Proposals for a Cleaner Planet, MIT Press, Cambridge, MA, 2001.
2. G. W. Crabtree, M. S. Dresselhaus and M. V. Buchanan, Phys. Today, 2004,(December), 39.
3. (a) L. Schlapbach and A. Züttel, Nature, 2001, 414, 353–358; (b) W. Grochala and P. P. Edwards, Chem. Rev., 2004, 104, 1283–1315; (c) I. R. Harris, D. Book, P. A. Anderson and P. P. Edwards, Fuel Cell Rev., 2004,(June/July), 17–23.
4. B. Bogdanovic, K. Bohmhammel, B. Christ, A. Reiser, K. Schlichte, R. Vehlen and U. Wolf, J. Alloys Compd., 1999, 282, 84–92.
5. (a) W. Oelerich, T. Klassen and R. Bormann, J. Alloys Compd., 2001, 315, 237–242; (b) O. Gutfleisch, N. Schlorke-de Boer, N. Ismail, M. Herrich, A. Walton, J. Speight, I. R. Harris, A. S. Pratt and A. Zuttel, J. Alloys Compd., 2003, 356, 598–602.
6. (a) P. Chen, Z. Xiong, J. Luo, J. Lin and K. L. Tan, Nature, 2002, 420, 302–304; (b) Y. Nakamori, G. Kitahara and S. Orimo, J. Power Sources, 2004, 138, 309–312; (c) F. E. Pinkerton, G. P. Meisner, M. S. Meyer, M. P. Balogh and M. D. Kundrat, Phys. Chem. B Lett., 2005, 109, 6–8.
7. W. Luo, J. Alloys Compd., 2004, 381, 284–287.
8. UK patent application no GB 0415561.0.
9. Hydrogen absorption/desorption properties were measured on a constant pressure thermogravimetric balance (Hiden IGA). The desorption kinetics were monitored at 10 mbar of hydrogen and 300 °C, absorption at 10 bar and 300 °C. Samples were transferred quickly from the glove box to the IGA to minimize contact with the atmosphere.
10. The desorption rates were calculated as the ratio between the amount desorbed and time taken to go from 80 to 20% of full capacity. The error in the wt% was ±0.05, while the error in the desorption rate was ±0.01 × 10⁻² wt% s⁻¹.
11. The absorption rates were calculated as the ratio between the amount absorbed and time taken to go from 20 to 80% of capacity measured at 60 min. The error in the wt% was ±0.05, while the error in the absorption rate was ±0.01 × 10⁻² wt% s⁻¹.
12. In addition to α-MgH₂, a number of other crystal structures, observed and predicted, have been reported for MgH₂; see M. Bortz, B. Bertheville, G. Bøttger and K. Yvon, J. Alloys Compd., 1999, 287 L4. α-MgH₂ is sometimes referred to in the literature as the ‘β-MgH₂ phase’, which is not to be confused with the proposed high temperature, high pressure β-MgH₂ structure.
13. Structural characterisation of the samples was carried out on a Siemens D5000 X-ray diffractometer, using a CuK_α source. The structure of α-MgH₂ is tetragonal (space group P4₂mnm) with a = 4.508 ± 0.001 Å, c = 3.017 ± 0.001 Å; MgH₂/LiBH₄ as-prepared contained two phases with a = 4.510 ± 0.001 Å, c = 3.017 ± 0.001 Å and a = 4.502 ± 0.001 Å, c = 3.011 ± 0.001 Å; MgH₂/LiBH₄ after cycling, a = 4.522 ± 0.003 Å, c = 3.025 ± 0.003 Å.
14. G. Liang, J. Huot, S. Boily and R. Schulz, J. Alloys Compd., 2000, 305, 239–245.
15. SEM images were recorded on a JEOL 6300 electron microscope.
16. XPS measurements were carried out at the Johnson Matthey Technology Centre, UK on a VG ESACLAB 250 instrument. A monochromatic Al K_α X-ray source at 1464.6 eV was used.
17. FTIR spectra were acquired at a resolution of 2 cm⁻¹, from KBr disks containing the samples, on a Nicolet Magna FTIR spectrometer, equipped with a liquid nitrogen cooled MCTB detector.
18. (a) T. J. Marks, W. J. Kennelly, J. R. Kolb and L. A. Shimp, Inorg. Chem., 1972, 11, 2540; (b) J. A. A. Ketelaar and C. J. H. Schutte, Spectrochim. Acta, 1961, 17, 1240; (c) T. C. Waddington, J. Chem. Soc., 1958, 4783.
19. Solid state ¹¹B NMR spectra were acquired on a 400 MHz Varian Inova spectrometer. Samples were packed into 3.2 mm o.d. rotors and spun at the magic angle at rates of 10 to 15 KHz. A pulse width of 0.4 to 1.4 µs was employed for data acquisition.
20. A. Züttel, S. Rentsch, P. Fischer, P. Wenger, P. Sudan, Ph. Mauron and Ch. Emmenegger, J. Alloys Compd., 2003, 356, 515–520.

---

Footnote

† Electronic supplementary information (ESI) available: XPS and ¹¹B NMR spectra. See http://www.rsc.org/suppdata/cc/b5/b503085d/

---