Synthesis and crystal structure of Li₄BH₄(NH₂)₃

Abstract

The solid solution, (LiNH₂)_x(LiBH₄)_(1−x), formed through the reaction of the two potential hydrogen storage materials, LiNH₂ and LiBH₄, is dominated by a compound that has an ideal stoichiometry of Li₄BN₃H₁₀ and forms a body-centred cubic structure with a lattice constant of ca. 10.66 Å.

---

The issue of storage is one of the major technical obstacles to the implementation of hydrogen as a fuel for mobile transport and much research effort has been devoted to developing new hydrogen storage methods and materials.¹ It has been shown that vast improvements in the hydrogen storage properties of known materials can be achieved by catalytic or chemical modification, such as in NaAlH₄ and MgH₂.^2,3 Although these developments are encouraging, it has also been recognized that an answer to the hydrogen storage challenge is likely to involve new, more complex materials systems.

LiNH₂ and LiBH₄ have both been extensively studied with regard to their potential hydrogen storage properties.^4,5 Recently, it was shown that, in the presence of lithium hydride, lithium amide reversibly desorbs hydrogen.^4,6,7 It has also been reported^8–10 that lithium amide and lithium borohydride, heated or ball milled in a 2 ∶ 1 mole ratio, form a new body centred cubic phase of composition Li₃BN₂H₈,⁸ but no further structural information was provided. This new quaternary hydride was shown to have much improved hydrogen desorption properties compared to the separate reactants. It is clear that the mixing of hydrides can provide alternative decomposition pathways promoting the release of hydrogen,^10,11 which may result in new, improved hydrogen storage systems. Here we report the synthesis and structure of the new quaternary hydride, whose stoichiometric composition is in fact Li₄BN₃H₁₀, but which tolerates a wide variation in stoichiometry.

LiNH₂ and LiBH₄ (Sigma-Aldrich) were ground together by hand in the desired mole ratio in an argon atmosphere glovebox (O₂ content < 10 ppm) and placed in a quartz tube. The tube was sealed with a Young's tap via an Ultra-Torr fitting and removed from the glovebox, and the reactants were then heated in a tube furnace under a mixture of H₂ and N₂ gas (1 ∶ 9) at 1 bar.

In a series of experiments, we heated different mixtures of LiNH₂ and LiBH₄ over the entire composition range and identified that a body centred cubic phase with a unit cell of around 10.66 Å was always present across the full compositional range. Synchrotron X-ray powder diffraction experiments showed that the lattice parameters varied between 10.6664 and 10.7080 Å, and that the 80 ∶ 20 composition contained an amide impurity while LiBH₄ was detected in the 70 ∶ 30 composition. We concluded that the most likely stoichiometric composition for this structure was 75 ∶ 25, giving the empirical formula Li₄BN₃H₁₀. Subsequently we observed that a mixture in a 3 ∶ 1 mole ratio heated at around 100 °C (near the orthorhombic–tetragonal structural transition temperature of LiBH₄)⁵ gave a pure¹² phase after annealing at 180 °C.¹³

The FTIR spectrum¹⁴ of Li₄BN₃H₁₀ suggests that NH₂⁻ and BH₄⁻ anions remain intact within the structure and that the compound should be regarded as Li₄BH₄(NH₂)₃. There were no absorptions observed in positions expected for bridging B–H–B vibrations, again consistent with the BH₄⁻ anion remaining intact.¹⁵ B–H stretching modes were shifted relative to LiBH₄ by around 8 cm⁻¹ to the higher frequencies of 2387, 2293, 2237 (shoulder) and 2225 cm⁻¹. BH₂ deformation bands at 1120 and 1092 cm⁻¹ in LiBH₄ were observed at 1126 and 1082 cm⁻¹ in Li₄BH₄(NH₂)₃. In contrast, the observed symmetric and asymmetric stretches of the amide anion¹⁶ were reduced from 3312 and 3259 cm⁻¹ in LiNH₂ to 3303 and 3243 cm⁻¹ in Li₄BH₄(NH₂)₃.

Recrystallization of the compound was attempted from a number of different solvents¹⁷ in order to produce a single crystal for structural determination, but unfortunately Li₄BH₄(NH₂)₃ proved to be similar to LiNH₂ in its lack of solubility in common solvents. Instead, the crystal structure was solved using a combination of high resolution synchrotron X-ray and neutron powder diffraction measurements.¹⁸ Extremely sharp diffraction peaks were observed in the X-ray data, indicating a high degree of crystallinity (see Fig. 1(a)). The regular spacing of peaks is indicative of a cubic crystal structure, which was verified using the computer program DASH;¹⁹ the refined lattice constant and unit cell volume were 10.66445(1) Å and 1212.875(3) ų, respectively. Consideration of the volumes of a single formula unit of LiNH₂ (V = 32.6 ų) and LiBH₄ (V = 54.2 ų) from diffraction data collected on ID31 leads to a 3 ∶ 1 ratio and 7.98 formula units in the unit cell. The closeness of this value to 8 is a strong indication that the ideal stoichiometry is indeed Li₄BN₃H₁₀. The most probable space groups were also determined using DASH by means of a Bayesian algorithm²⁰ that evaluates the relative probabilities of different space group extinction symbols; the most probable symbol was found to be I- - - which is associated with the space groups I23, I2₁3, Im3, I432, I4̄3m and Im3m. Unless there is substantial disorder in the crystal structure, the only two space groups that are consistent with 8 formula units per unit cell are I23 and I2₁3. Both were assessed in the structure solution process. The contributions of nitrogen and boron together correspond to over half the X-ray scattering in Li₄BH₄(NH₂)₃ and thus to simplify structure solution only these atoms were considered initially in the solution process which was undertaken using the computer program TOPAS.²¹ Solutions of similar quality were found for both space groups I23 and I2₁3. Examination of the resulting structures indicated that the only difference between the two space groups was the permutation of N and B atoms over the same set of sites in the unit cell²² which explains the similar qualities of fit.

Fig. 1 Final Rietveld plots for (a) X-ray and (b) neutron diffraction data showing observed (purple and green respectively), calculated (red) and difference (grey) plots. The inserts show the low d-spacing region at a magnified scale. Peak positions for Li₄BH₄(NH₂)₃ and Li₂O are indicated.

The B and N sites occupy half of the possible positions at approximately ([2ℓ + 1]/8, [2m + 1]/8, [2n + 1]/8) where ℓ,m,n = {0,1,2,3}; this arrangement is reminiscent of the nitrogen positions in LiNH₂ (Fig. 2(b)). Building upon the apparent close similarity between Li₄BH₄(NH₂)₃ and LiNH₂, all sites at (ℓ/4, m/4, n/4) (where ℓ,m,n = {0,1,2,3}) were considered to be possible lithium candidate positions. The next stage involved the refinement of the occupancies of each of these sites. At this point, I2₁3 became a better option than I23. The final stage of the structure solution process involved neutron diffraction data.¹⁸ Usually deuteration and ¹¹B enrichment are required for neutron powder diffraction experiments because of problems with high background and absorption. The high count-rate of GEM, however, means that natural Li₄BH₄(NH₂)₃ could be used without isotopic enrichment (see Fig. 1(b)). Four hydrogen atoms were placed in tetrahedral coordination around boron initially at 1.25 Å while two hydrogen atoms were located around each nitrogen atom at a distance of 1 Å in a configuration similar to LiNH₂. Initial bond length and angle restraints were used for both BH₄⁻ and NH₂⁻ anions and all atomic positions were refined. In the final refinement using the neutron diffraction data (Table 1), all restraints were removed and an excellent fit was obtained over the full d-spacing range down to 0.44 Å. The resulting crystal structure is shown in Fig. 2(a). The corresponding fits to the X-ray and neutron diffraction data are shown in Figs. 1 (a) and (b) respectively.

Table 1 Refined crystallographic data: space group I2₁3 a = 10.66445(1) Å

Atom	x/a	y/a	z/a	B iso/Å^2
a X-ray data^18 – Rwp = 13.8, Rexp = 10.6, χ^2 = 1.7. Neutron data^23 – Rwp = 0.821 (11.9%), Rexp = 0.726 (10.5%), χ^2 = 1.3.
Li1	0.25	−0.0428(14)	0.0	1.5(1)
Li2	−0.2622(16)	0.0	0.25	1.5(1)
Li3	0.2722(12)	0.2722(12)	0.2722(12)	1.5(1)
B	0.1354(5)	0.1354(5)	0.1354(5)	1.4(2)
H1	0.0721(15)	0.0721(15)	0.0721(15)	4.1(2)
H2	0.0969(10)	0.1361(9)	0.2480(8)	4.1(2)
N	−0.1559(3)	0.1101(2)	0.3668(3)	1.8(1)
H3	−0.1533(12)	0.0533(15)	0.4394(12)	5.6(2)
H4	−0.0773(12)	0.1672(12)	0.3768(14)	5.6(2)

---

Fig. 2 (a) Crystal structure of Li₄BH₄(NH₂)₃; (b) LiNH₂ for comparison. Li atoms are depicted in red, N in blue, H in grey and BH₄⁻ as green tetrahedra.

Interestingly, the B–H bonds in Li₄BH₄(NH₂)₃ are shorter than in LiBH₄; however, as with LiBH₄, one B–H bond is significantly (1.17 Å) shorter than the other three (1.27 Å).²³ N–H bond lengths are close to the N–D bond lengths observed in LiND₂.²⁴ These observations agree with the FTIR data, which showed a strengthening and weakening of the B–H and N–H bonds relative to LiBH₄ and LiNH₂, respectively.

The Li₄BH₄(NH₂)₃ structure accommodates a wide range of stoichiometries. Direct substitution of NH₂⁻ ions by the larger BH₄⁻ ions (anion volumes of 0.43 and 0.66 nm³, respectively) results in an increase in observed lattice parameter.²⁵ Annealing experiments are currently in progress to determine the exact limits of this non-stoichiometry; preliminary results indicate that the composition range for (LiNH₂)_x(LiBH₄)_(1−x) may extend from x = 0.5 to x = 0.8. It is worth noting that the replacement of an NH₂⁻ anion in the unit cell of LiBH₄(NH₂)₃ by an additional BH₄⁻ necessarily results in shorter distances between BH₄⁻ ions than in the stoichiometric compound and that non-stoichiometric compositions may be less stable as a result. In samples containing less than 40 mol% LiNH₂, we also observed a second body centred cubic phase with a larger unit cell size of ∼ 11.1 Å, which co-exists with the Li₄BH₄(NH₂)₃ structure and an excess of LiBH₄. Early indications suggest that this phase may represent a second preferred composition within the (LiNH₂)_x(LiBH₄)_(1−x) solid solution range. Further studies to determine the stoichiometry of this second phase are ongoing, but this work is complicated by its lack of stability over time with respect to LiBH₄(NH₂)₃ and LiBH₄.

When Chen et al. observed that the intimate mixing of LiH with LiNH₂ suppresses the usual release of ammonia in favour of hydrogen evolution at a much reduced decomposition temperature, it was suggested that this could be due to an interaction between partly positively charged hydrogen atoms in LiNH₂ and the negatively charged hydride anions in LiH,²⁶ and a mechanism for hydrogen desorption has recently been discussed.²⁷ We hypothesize that a similar interaction may occur in Li₄BH₄(NH₂)₃, as H atoms in BH₄⁻ are also partially negative.²⁸ Significantly, in Li₄BH₄(NH₂)₃ the BH₄⁻ ions are close to the NH₂⁻ on a molecular level, obviating the need for mechanical mixing. Initial findings confirm that the major gaseous decomposition product of Li₄BH₄(NH₂)₃ above 260 °C is indeed H₂ with ammonia present only in trace amounts. Similar findings have been reported for the decomposition of the 2 ∶ 1 LiNH₂/LiBH₄ mixture.^8,10 Remarkably, this new quaternary complex hydride, constituted mainly of—and with striking structural similarities to—LiNH₂, has a completely different decomposition pathway from the pure amide, favouring H₂ evolution over that of ammonia.

We wish to thank Andy Fitch for assistance in using beamline ID31 at the ESRF, Grenoble, and Ron Smith and Richard Ibberson for assistance at ISIS. SRJ acknowledges the EPSRC (SUPERGEN), and PAC the EPSRC for funding.

Notes and references

1. See L. Schlapbach and A. Züttel, Nature, 2001, 414, 353; R. Harris, D. Book, P. Anderson and P. Edwards, Fuel Cell Rev., 2004, 1, 17; A. Züttel, Naturwissenschaften, 2004, 91(4), 157; W. Grochala and P. P. Edwards, Chem. Rev., 2004, 104, 1283; A. M. Seayad and D. M. Antonelli, Adv. Mater., 2004, 16, 765; F. Schüth, B. Bogdanović and M. Felderhoff, Chem. Commun., 2004, 2249 and references therein.
2. B. Bogdanović and M. Schwickardi, J. Alloys Compd., 1997, 253, 1.
3. S. R. Johnson, P. A. Anderson, P. P. Edwards, I. Gameson, J. W. Prendergast, M. Al-Mamouri, D. Book, I. R. Harris, J. D. Speight and A. Walton, Chem. Commun., 2005, 22, 2823.
4. P. Chen, X. Xiong, J. Luo, J. Lin and K. L. Tan, Nature, 2002, 420, 302.
5. A. Züttel, P. Wenger, S. Rentsch, P. Sudan, Ph. Mauron and Ch. Emmenegger, J. Power Sources, 2003, 118, 1.
6. Y. H. Hu and E. Ruckenstein, J. Phys. Chem. A, 2003, 107, 9737.
7. T. Ichikawa, S. Isobe, N. Hanada and H. Fujii, J. Alloys Compd., 2004, 356, 271.
8. F. E. Pinkerton, G. P. Meisner, M. S. Meyer, M. P. Balogh and M. D. Kundrat, J. Phys. Chem. B, 2005, 109, 6.
9. M. Aoki, K. Miwa, T. Noritake, G. Kitahara, Y. Nakamori, S. Orimo and S. Towata, Appl. Phys. A, 2005, 80, 1409.
10. Y. Nakamori, A. Ninomiya, G. Kitahara, M. Aoki, T. Noritake, K. Miwa, Y. Kojima and S. Orimo, J. Power Sources, 2005 (in press).
11. J. Lu and Z. Z. Fang, J. Phys. Chem. B, 2005, 109, 20830.
12. Li₂O (Fm3̄m, a = 4.609(5) Å) present from the LiNH₂ starting material was also observed at ca. 6%.
13. Initially a small amount of lithium amide was present, which was incorporated into the structure by repeated grinding and annealing at 180 °C.
14. FTIR spectra were acquired from samples contained in pressed KBr disks on a Nicolet Magna FTIR spectrometer equipped with a liquid nitrogen cooled MCTB detector, at a resolution of 2 cm⁻¹.
15. T. J. Marks and J. R. Klob, Chem. Rev., 1997, 77, 263.
16. J.-P. O. Bohger, R. R. Eβmann and H. Jacobs, J. Mol. Struct., 1995, 348, 325.
17. Solvents tried were diethyl ether, THF, DMF and DMSO.
18. X-ray diffraction data were collected on the ID31 diffractometer at the ESRF, Grenoble at a wavelength and stepsize of 0.80102 Å and 0.003° respectively. 354 reflections were used in the structural refinement. Neutron diffraction data were collected on GEM at the ISIS spallation neutron source at the Rutherford Appleton Laboratory. 1395 reflections were used in the refinement. CCDC 225200 & 294505. For crystallographic data in CIF or other electronic format see DOI: 10.1039/b518243c.
19. W. I. F. David, K. Shankland and N. Shankland, Chem. Commun., 1998, 8, 931.
20. A. J. Markvardsen, W. I. F. David, J. C. Johnson and K. Shankland, Acta Crystallogr., Sect. A, 2001, 57, 47.
21. Bruker AXS (2004): TOPAS V3.0: General profile and structure analysis software for powder diffraction data. Bruker AXS, Karlsruhe, Germany.
22. See Electronic Supplementary Information.
23. J.-Ph. Soulié, G. Renaudin, R. Černý and K. Yvon, J. Alloys Compd., 2002, 346, 200.
24. M. Nagib and H. Jacobs, Atomkernenergie, 1973, 21(4), 275.
25. H. D. B. Jenkins, H. K. Roobottom, J. Passmore and L. Glasser, Inorg. Chem., 1999, 38, 3609.
26. P. Chen, Z. Xiong, J. Luo, J. Lin and K. L. Tan, J. Phys. Chem. B, 2003, 107, 10967.
27. S. Isobe, T. Ichikawa, S. Hino and H. Fujii, J. Phys. Chem. B, 2005, 109, 14855.
28. Q. Ge, J. Phys. Chem. A, 2004, 108, 8682.

---

Footnote

† Electronic supplementary information (ESI) available: FTIR spectra, a comparison of symmetry groups and a virtual reality image of Fig. 2(a). See DOI: 10.1039/b518243c

---