Ammonia triborane: A new synthesis, structural determinations, and hydrolytic hydrogen-release properties

Article abstract of DOI:10.1021/ja808045p

Iodine oxidation of B3H8 - in glyme solution to produce (glyme)B3H7, followed by displacement of the coordinated glyme by reaction with anhydrous ammonia provides a safe and convenient preparation of ammonia triborane, NH3B3H7 (1). X-ray crystallographic

Full text of DOI:10.1021/ja808045p

Published on Web 12/15/2008
Ammonia Triborane: A New Synthesis, Structural
Determinations, and Hydrolytic Hydrogen-Release Properties
Chang Won Yoon, Patrick J. Carroll, and Larry G. Sneddon*
Department of Chemistry, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104-6323
Received October 17, 2008; E-mail: lsneddon@sas.upenn.edu
Abstract: Iodine oxidation of B3H8- in glyme solution to produce (glyme)B3H7, followed by displacement of
the coordinated glyme by reaction with anhydrous ammonia provides a safe and convenient preparation of
ammonia triborane, NH3B3H7 (1). X-ray crystallographic determinations and DFT computational studies of
both NH3B3H7 and the NH3B3H7 ·18-crown-6 adduct demonstrate that while computations predict a symmetric
single bridging-hydrogen conformation, NH3B3H7 has a highly asymmetric structure in the solid-state that
results from intermolecular N-H+ · · ·H--B dihydrogen bonding interactions. Studies of its hydrolytic reactions
have shown that upon the addition of acid or an appropriate transition metal catalyst, aqueous solutions of
1 rapidly release hydrogen, with 6.1 materials wt % H2-release being achieved from a 22.7 wt % aqueous
solution of 1 at room temperature in the presence of 5 wt % Rh/Al2O3 (1.1 mol% Rh). The rate of H2-
release was controlled by both the catalyst loadings and temperature.
Introduction
materials wt % H2) or thermolysis, eq 2, (17.7 materials wt %
H2) also makes it an attractive candidate for chemical hydrogen
Owing to their high hydrogen densities, boron-based com-
pounds, such as sodium borohydride (NaBH4) and ammonia
borane (NH3BH3), are now being intensively investigated as
chemical hydrogen storage materials that can release hydrogen
by either hydrolytic1,2 or thermolytic processes.3
storage.
NH3B3H7 + 6H2O f NH+4 + 3BO2 + 2H+ + 8H2
NH3B3H7f“B3N” + 5H2
(1)
(2)
The high hydrogen release capacity that could potentially be
achieved by ammonia triborane oxidative-hydrolysis, eq 1, (9.7
Although Kodama first synthesized4 NH3B3H7 over 50 years
ago, owing to the lack of a suitable method for its efficient and
safe synthesis, its reactivities and properties have not been
intensively explored. Of the fewer than 30 previous publications
on 1, many were stimulated by the apparent contradiction
between the computational studies5 that predict a symmetric
single hydrogen-bridged CS-symmetric structure, and the early
single crystal X-ray determination6 of 1 that showed an
asymmetric structure with perhaps two bridging-hydrogens.
In this paper we present (1) a new, efficient preparation of 1
that now makes this compound easily available; (2) a new
crystallographic study of the solid-state structure of 1, along
with a structural determination of the 1·18-crown-6 adduct, that
(1) For examples of NaBH4 hydrolytic H2-release reactions, see: (a)
Amendola, S. C.; Sharp-Goldman, S. L.; Janjua, M. S.; Kelly, M. T.;
Petillo, P. J.; Binder, M. J. Power Sources 2000, 85, 186–189. (b)
Amendola, S. C.; Sharp-Goldman, S. L.; Janjua, M. S.; Spencer, N. C.;
Kelly, M. T.; Petillo, P. J.; Binder, M. Int. J. Hydrogen Energy 2000,
25, 969–975. (c) Kojima, Y.; Suzuki, K.; Fukumoto, K.; Sasaki, M.;
Yamamoto, T.; Kawai, Y.; Hayashi, H. Int. J. Hydrogen Energy 2002,
27, 1029–1034. (d) Dong, H.; Yang, H.; Ai, X.; Cha, C. Int. J.
Hydrogen Energy 2003, 28, 1095–1100. (e) Jeong, S. U.; Kim, R. K.;
Cho, E. A.; Kim, H.-J.; Nam, S.-W.; Oh, I.-H.; Hong, S. -A.; Kim,
S. H. J. Power Sources 2005, 144, 129–134. (f) Krishnan, P.; Yang,
T.-H.; Lee, W.-Y.; Kim, C.-S. J. Power Sources 2005, 143, 17–23.
(g) Wee, J.-H.; Lee, K.-Y.; Kim, S. H. Fuel Process. Technol. 2006,
87, 811–819.
(2) For examples of amineborane hydrolytic H2-release reactions, see: (a)
Kelly, H. C.; Marriott, V. B. Inorg. Chem. 1979, 18, 2875–2878. (b)
Shvets, I. B.; Erusalimchik, I. G. Elektrokhimiya 1984, 20, 535–537.
(c) D’Ulivo, A.; Onor, M.; Pitzalis, E. Anal. Chem. 2004, 76, 6342–
6352. (d) Storozhenko, P. A.; Svitsyn, R. A.; Ketsko, V. A.; Buryak,
A. K.; Ul’yanov, A. V. Zh. Neorg. Khim. 2005, 50, 1066–1071. (e)
Chandra, M.; Xu, Q. J. Power Sources 2006, 156, 190–194. (f)
Chandra, M.; Xu, Q. J. Power Sources 2006, 159, 855–860. (g) Xu,
Q.; Chandra, M. J. Power Sources 2006, 163, 364–370. (h) Mohajeri,
N.; Adebiyi, O.; Baik, J.; Bokerman, G.; T-Raissi, A. Prepr. Pap.-
Am. Chem. Soc., DiV. Fuel Chem. 2006, 51, 520–521. (i) Chandra,
M.; Xu, Q. J. Power Sources 2007, 168, 135–142. (j) Cheng, F.; Ma,
H.; Li, Y.; Chen, J. Inorg. Chem. 2007, 46, 788–794. (k) Mohajeri,
N.; T-Raissi, A.; Adebiyi, O. J. Power Sources 2007, 167, 482–485.
(l) Clark, T. J.; Whittell, G. R.; Manners, I. Inorg. Chem. 2007, 46,
7522–7527. (m) Kalidindi, S. B.; Indirani, M.; Jagirdar, B. R. Inorg.
Chem. 2008, 47, 7424–7429. (n) Yan, J.-M.; Zhang, X.-B.; Han, S.;
Shioyama, H.; Xu, Q. Angew. Chem., Int. Ed. 2008, 47, 2287–2289.
(o) Yoon, C. W.; Sneddon, L. G. J. Am. Chem. Soc. 2006, 128, 13992–
13993.
(3) For examples of amineborane thermolytic H2-release reactions, see:
Stephens, F. H.; Pons, V.; Baker, R. T. Dalton Trans. 2007, 2613–
2626, and references therein.
(4) (a) Kodama, G. Ph.D. Dissertation 1957, University of Michigan. (b)
Kodama, G.; Parry, R. W. Proc. XVI Int. Congress Pure Appl. Chem.
1957, 482–489. (c) Kodama, G.; Parry, R. W.; Carter, J. C. J. Am.
Chem. Soc. 1959, 81, 3534–3538.
(5) (a) Brown, L. D.; Lipscomb, W. N. Inorg. Chem. 1977, 16, 1–7. (b)
McKee, M. L. Inorg. Chem. 1988, 27, 4241–4245. (c) Mebel, A. M.;
Musaev, D. G.; Morokuma, K. Chem. Phys. Lett. 1993, 214, 69–76.
(d) Es-sofi, A.; Serrar, C.; Ouassas, A.; Jarid, A.; Boutalib, A.; Nebot-
Gil, I.; Toma´s, F. J. Phys. Chem. A 2002, 106, 9065–9070. (e) Nguyen,
V. S.; Matus, M. H.; Nguyen, M. T.; Dixon, D. A. J. Phy. Chem. C
2007, 111, 9603–9613. (f) Subotnik, J. E.; Sodt, A.; Head-Gordon,
M. Phys. Chem. Chem. Phys. 2007, 9, 5522–5530. (g) Sundberg,
M. R.; Sanchez-Gonzalez, A. Inorg. Chem. Commun. 2007, 10, 1229–
1232.
(6) Nordman, C. E.; Reimann, C. J. Am. Chem. Soc. 1959, 81, 3538–
3543.
9
10.1021/ja808045p CCC: $40.75
2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 855–864 855

A R T I C L E S
Yoon et al.
Table 1. Crystallographic Data Collection and Structure
resolves the contradictions with computational structural predic-
tions; and (3) a description of the hydrolytic hydrogen release
properties of 1.
Refinement Information
1
1·18-crown-6
C12B3H34NO6
320.83
empirical formula
fw
B3H10N
56.52
Experimental Section
All manipulations were carried out using standard high-vacuum
or inert-atmosphere techniques as described by Shriver.7
cryst class
space group
Z
a, Å
b, Å
monoclinic
P21/n (no. 14)
4
10.358(3)
4.8041(12)
9.967(3)
115.141(6)
448.9(2)
0.836
monoclinic
P21/n (no. 14)
8
21.961(3)
8.5678(9)
21.910(3)
113.555(3)
3779.1(8)
1.128
Materials. Aldrich NaBH4 (98%), iodine, 18-crown-6, RhCl3,
[Rh(COD)(µ-Cl)]2 (COD ) 1,5-cyclooctadiene), Pd/Al2O3, Pt/C,
Pt (nanosize, activated), Pd (sub-micrometer), Ru powder, and NiCl2
from Aldrich or Strem Rh/Al2O3, Ru/Al2O3, and RuCl3 were stored
under inert atmosphere and used as received. Diglyme (Acros
Organics) was distilled over Na prior to use. Glyme, dimethoxy-
ethane, (Aldrich) was dried by passing through an activated alumina
c, Å
ꢀ, deg
V, Å3
Dcalc, g/cm3
µ, cm-1
λ, Å (Mo K)
cryst size, mm
F(000)
0.41
0.71073
0.83
0.71073
column prior to use. Bu4N+B3H8 was prepared according to the
-
0.44 × 0.30 × 0.20 0.42 × 0.12 × 0.10
procedure described by Ryschkewitsch.8
128
1408
5.16-50.02
143
-25 e h e 26
-10 e k e 7
-22 e l e 26
22 010
Physical Measurements. 11B NMR spectra at 128.4 MHz and
1H NMR spectra at 400.1 MHz were obtained on a Bruker DMX-
400 spectrometer equipped with appropriate decoupling accessories.
1H NMR spectra at 500.1 MHz were obtained on a Bruker AM-
500 spectrometer. All 11B chemical shifts are referenced to external
BF3 ·O(C2H5)2 (0.0 ppm) with a negative sign indicating an upfield
2θ angle, deg
T, K
hkl collected
7.48-50.00
143
-12 e h e 12
-5 e k e 4
-11 e l e 11
3496
no. reflections measured
no. unique reflns
no. observed reflns (F > 4σ)
no. reflections used in refinement 791
791 (Rint ) 0.0201) 6590 (Rint ) 0.0420)
722
1
shift. All H chemical shifts were measured relative to internal
5869
6590
671
R1 ) 0.0498
wR2 ) 0.1208
R1 ) 0.0560
wR2 ) 0.1270
1.086
residual protons from the lock solvents (99.9% CD2Cl2 or 99.5%
C6D6) and are referenced to (CH3)4Si (0.0 ppm). Diffuse reflectance
infrared Fourier transform (DRIFT) spectra were obtained on a
Perkin-Elmer 100 FT-IR spectrometer with DRIFT accessory.
Melting points were determined using a standard melting point
apparatus and are uncorrected. All pH measurements were per-
formed using an Orion 230 pH meter.
no. of params
78
a
R1 ) 0.0411
wR2 ) 0.1078
R1 ) 0.0438
wR2 ) 0.1098
1.091
R
indices
(F > 4σ)
a
R
indices
(all data)
b
GOF
final difference peaks, e/Å3
+0.159, -0.173
+0.237, -0.258
Preparation of NH3B3H7 (1) from Bu4N+B3H8-. After a
a
R1 ) Σ|Fo| - |Fc|/Σ|Fo|; wR2 ) {Σw(Fo - Fc2)2/Σw(Fo )2}1/2
.
-
2
2
solution of Bu4N+B3H8 (1.0 g, 3.53 mmol) in dry glyme (10.0
b
GOF ) {Σw(Fo - Fc2)2/(n - p)}1/2; n ) the number of reflections; p
) the number of parameters refined.
2
mL) was cooled at -70 °C for 15 min under an inert atmosphere,
a solution of iodine (0.44 g, 1.73 mmol) in dry glyme (10.0 mL)
was added, and the resulting solution stirred for 10 min at this
temperature. The reaction mixture was then allowed to warm to
room temperature, during which time the purple color of the solution
faded, indicating consumption of the I2. Analysis of the solution
by 11B NMR showed the formation of (glyme)B3H7. The solution
was again cooled at -70 °C and anhydrous ammonia was bubbled
through the solution with vigorous stirring. The solution was then
warmed slowly to room temperature, and the white precipitate that
had formed was removed by filtration. The filtrate was dried in
vacuo and separation of the resulting residue by silica gel
chromatography with CH2Cl2 as an eluent gave a white crystalline
powder, NH3B3H7 (0.16 g, 2.83 mmol) in 80% yield. The melting
point and spectroscopic data match the literature values for
NH3B3H7.9 Mp ≈ 73 °C. 1H NMR (400.13 MHz, CD2Cl2) 3.58 (t,
3, NH3), 1.62 (br, 7, B3H7). 11B NMR (128.4 MHz, CD2Cl2) -7.7
(br, s, 2, B2 and B3), -32.8 (br, s, 1, B1). IR (cm-1) 3321(m),
3251(w), 2494(s), 2435(s), 1599(m), 1388(s), 1164(m), 1122(w),
1011(w), 882(w).
Computational Studies. DFT/GIAO/NMR calculations were
performed using the Gaussian 03 program.10a The geometries were
fully optimized at the B3LYP/6-31G(d) level without symmetry
constraints. A vibrational frequency analysis was carried out on
each optimized geometry using the same level of theory with a
true minimum found for each structure (i.e., possessing no
imaginary frequencies). The energies of the optimized structures
of 1, MeNH2B3H7, and Me2NHB3H7 and their corresponding dimers
were computed both at the B3LYP/6-311+G(2d,p) level and at the
MP2/6-311+G(2d,p) level based on the B3LYP/6-31G(d) optimized
structures. In all cases, zero point energies obtained at the B3LYP/
6-31G(d) level were scaled by 0.980410b and then used for the final
energy correction.
The 11B NMR chemical shifts were calculated at the B3LYP/
6-311G(d) level using the GIAO option within Gaussian 03. The
11B NMR GIAO chemical shifts are referenced to BF3 ·OEt2 using
an absolute shielding constant of 101.58, which was obtained from
the GIAO NMR calculated shift of BF3 ·OEt2 at the B3LYP/6-
311G(d)//B3LYP/6-31G(d) level of theory.
Crystallographic Data. Single crystals of 1 (Penn 3323) were
obtained from a concentrated toluene solution stored at -20 °C.
Crystals of the 1·18-crown-6 (C12H24O6) adduct (Penn 3304) were
grown from 1:1 mixtures of 1 and 18-crown-6 in THF/hexanes at
room temperature.
Collection and Reduction of the Data. Crystallographic data
and structure refinement information are summarized in Table 1.
X-ray intensity data were collected on a Rigaku Mercury CCD area
detector employing graphite-monochromated Mo KR radiation
(λ ) 0.71073 Å). Preliminary indexing was performed from a series
of 12 0.5° rotation images with exposures of 30 s. Rotation images
were processed using CrystalClear,11a producing a listing of
unaveraged F2 and σ(F2) values which were then passed to the
CrystalStructure11b program package for further processing and
The reaction residue after the ammonia treatment could be
alternatively purified by sublimation at 50 °C under reduced
pressure, but this method took longer and afforded lower yields
(<50%). (CAUTION: after sublimation, immediate deactiVation
of the oily residue is required; otherwise the residue and/or gases
from the oily residue can enflame upon exposure to air).
(7) Shriver, D. F.; Drezdon, M. A. Manipulation of Air SensitiVe
Compounds, 2nd ed.; Wiley: New York, 1986.
(8) (a) Nainan, K. C.; Ryschkewitsch, G. E. Inorg. Nucl. Chem. Lett. 1970,
6, 765–766. (b) Ryschkewitsch, G. E.; Nainan, K. C. Inorg. Synth.
1974, 15, 113–114.
(9) Dodds, A. R.; Kodama, G. Inorg. Chem. 1976, 15, 741–743.
(10) (a) Frisch, M. J. et al. Gaussian 03, Revision B.05; Gaussian, Inc.:
Pittsburgh, 2003. (b) Foresman, J. B. and Frisch, E. Exploring
Chemistry with Electronic Structure Methods, 2nd ed.; Gaussian, Inc.:
Pittsburgh, 1996; p 64.
(11) (a) CrystalClear: Rigaku Corporation, 1999. (b) Crystal Structure:
Crystal Structure Analysis Package, Rigaku Corp. Rigaku/MSC, 2002.
9
856 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Ammonia Triborane
A R T I C L E S
structure solution on a Dell Pentium III computer. The intensity
data were corrected for Lorentz and polarization effects and for
absorption.
Solution and Refinement of the Structure. The structures were
solved by direct methods (SIR97).12 Refinements were made by
full-matrix least-squares based on F2 using SHELXL-97.13 All
reflections were used during refinement (F2’s that were experimen-
tally negative were replaced by F2 ) 0). Non-hydrogen atoms were
refined anisotropically and hydrogen atoms were refined isotropically.
General Procedures for Hydrolytic H2-Release Measure-
ments from Aqueous NH3B3H7 Solutions. Samples of 1 were
added to neutral water in a two-necked round-bottom flask equipped
with a sidearm containing acid or metal catalysts. The flask was
connected to a gas buret via the second arm, closing the system.
When the sidearm was inverted, the acid or catalyst was mixed
with the aqueous NH3B3H7 solution and H2 was evolved. The
volume of evolved H2 was quantitatively measured versus time
employing the gas buret with pressure equalization against atmo-
spheric pressure. For the reactions at 30 and 50 °C, the final volumes
were measured after cooling the reaction mixtures to room
temperature.
Stability of Solid and Aqueous Ammonia Triborane. Solid 1
stored in air at room temperature showed no changes by 11B NMR
analysis after 2 months. The stability of a 33 wt % (8.8 M) aqueous
solution (prepared by dissolving 0.33 g of 1 in 0.66 mL of distilled
water) of 1 was determined by comparing the integration ratio of
1 (-32.9 ppm) and its NH3BH3 (-22.0 ppm) decomposition product
in the 11B NMR spectra taken versus time (Table S1).
mg (0.085 mmol) of 1. More 1 was then added to the solution at
three additional times and the evolved hydrogen measured (Table
S7).
Extended Catalyst Lifetime Studies of 5 wt % Rh/Al2O3 (7
mol % Rh) in the Presence of Buffer Solutions. A borate-buffered
solution was prepared by adding CO2-free distilled water to a
volumetric flask containing 3.05 g of Na2B4O7 ·10H2O and then
adjusting the pH of the solution to 7.2. The 5 wt % Rh/Al2O3 (1.25
mg Rh, 0.0121 mmol Rh) catalyst was added to 2.0 mL of the
buffer solution containing 9.0 mg (0.159 mmol) of 1 and the evolved
H2 was measured with the gas buret. More 1 was periodically added
to the solution at 10 additional times and the evolved hydrogen
measured (Table S8).
Hydrolysis of 1 with Rh/Al2O3 at Different Temperatures.
In separate experiments, 40 mg samples of Rh/Al2O3 (5 wt % Rh/
Al2O3, 0.019 mmol Rh, 1.1 mol % Rh) were added to 4.9 wt %
aqueous solutions of 1 (0.1 g NH3B3H7 + 1.9 mL H2O + 40 mg
Rh/Al2O3) at 0, 20, 35, and 50 °C and the evolved hydrogen was
then measured versus time (Table S9).
Hydrolysis of a Concentrated Aqueous NH3B3H7 Solution.
A 40 mg sample of Rh/Al2O3 (5 wt % Rh/Al2O3, 0.019 mmol Rh,
1.1 mol % Rh) was added to a 22.7 wt % aqueous solution of 1
(0.1 g NH3B3H7 + 0.3 mL H2O + 40 mg Rh/Al2O3) at room
temperature. The evolved hydrogen was quantitatively measured
versus time with the gas buret (Table S10). The weight percent of
the hydrogen produced was calculated by materials wt % H2
released ) [H2 wt produced/(NH3B3H7 + H2O + Rh/Al2O3 wts)]
× 100.
For the 0.5 wt % solutions (0.09 M) with various pH values, a
series of 10.0 mg samples of 1 were dissolved in 2.0 mL of distilled
water or the desired buffers. The initial pH of the unbuffered 1
(0.5 wt %) solution was 8.7 by pH meter. The phosphate buffers
(sodium dihydrogenphosphate, 99%, Aldrich) of 1.0 M concentra-
tions were prepared by dissolving 12.0 g of NaH2PO4 and 14.6 g
of NaCl in 90.0 mL of CO2-free distilled water in a volumetric
flask, and then adjusting the solution to the desired pH’s using a
minimum amount of HCl and/or NaOH. Additional CO2-free
distilled water was added into the volumetric flask to 100.0 mL.
The degree of degradation of 1 was again measured by integrating
the ratio of 1 and NH3BH3 in the 11B NMR spectrum at each time
(Table S1).
Hydrolysis of 1 with HCl. Concentrated HCl (11.8 N, 1.0 mL)
was added to solid 1 (12.0 mg, 0.212 mmol) at room temperature,
and the evolved hydrogen was quantitatively measured versus time
with the gas buret (Table S2). In separate experiments, 1.9 mL of
each of the standardized 0.93 (1.0 equiv), 2.79 (3.0 equiv), and
4.66 M (5.0 equiv) solutions of HCl was added to the three samples
of 1 (0.1 g, 1.77 mmol) and the released hydrogen was quantitatively
measured versus time with the gas buret (Table S3).
Results and Discussion
Synthesis. Kodama4 originally synthesized ammonia triborane
(1) by a two-step process (eqs 3 and 4) involving (1) the
cleavage reaction of tetraborane, B4H10, with ethers to initially
form LB3H7 (L ) tetrahydrofuran or tetrahydropyran) adducts
(plus B2H6), followed by (2) displacement of the ethers by
reaction with anhydrous ammonia. Unfortunately, since tet-
raborane is both thermally unstable and explosive in air, a large-
scale synthesis based on this method is not feasible.
B4H10 + L f LB3H7 + 1 ⁄ 2B2H6
(3)
(4)
LB3H7 + anhydrous NH3 f NH3B3H7 + L
Kodama also showed 1 could be obtained from the reaction
of Na+B3H8- with NH4Cl, but in much lower yields (typically
20-30%) than the tetraborane-based route. The air- and
-
moisture-stable B3H8 anion is an attractive starting material
since it can be readily made by the iodine oxidation of sodium
borohydride (eq 5).8 Binder14 previously reported that I2
Hydrolysis of 1 with Metal Catalysts. Experiments were carried
out at room temperature by adding the desired amounts (Table S4)
of the metal catalysts to samples of 1 dissolved in neutral water
(1.0 or 2.0 mL) and then measuring the evolved hydrogen versus
time with the gas buret (Table S5).
oxidation of Me4N+B3H8 in noncoordinating solvents (e.g.,
-
CH2Cl2) yielded the B3H7 dimer, B6H14. We found that iodine
oxidation (with slow warming from -70 to 20 °C) of the air-
stable Bu4N+B3H8 salt in glyme (1,2-dimethoxyethane) pro-
-
vided an efficient method for preparing solutions of the
(glyme)B3H7 adduct (eq 6).
Extended Catalyst Lifetime Studies of [Rh(COD)(µ-Cl)]2
(22.7 and 7.9 mol %) in the Absence of Buffer Solutions. The
[Rh(COD)(µ-Cl)]2 (12 mg, 0.024 mmol, 22.9 mol %) precatalyst
was added to 1.0 mL of the neutral water containing 4.6 mg (0.081
mmol) of 1 and the evolved H2 was measured with the gas buret.
More 1 was then added to the solution at four additional times,
and the evolved hydrogen measured (Table S6). In a separate
experiment, the [Rh(COD)(µ-Cl)]2 (3.6 mg, 0.0073 mmol, 7.9 mol
%) catalyst was added to 1.0 mL of neutral water containing 4.8
diglyme
3Na+BH-4 + I2 f Na+B3H8- + 2NaI + 2H2
(5)
110oC
glyme
NBu4+B3H-8 + 1 ⁄ 2I2
f
(glyme)B3H7 + 1 ⁄ 2H2 + NBu4I
-70-23oC
(6)
(7)
glyme
f
(glyme)B3H7 + NH3(g)
NH3B3H7 + glyme
(12) Altomare, A.; Burla, M.; Camalli, M.; Cascarano, G.; Giacovazzo,
C.; Guagliardi, A.; Moliterni, A.; Polidori, G.; Spagna, R. J. Appl.
Crystallogr. 1999, 32, 115–119.
(13) Sheldrick, G. M. SHELXL-97: Program for the Refinement of Crystal
Structures; University of Go¨ttingen: Go¨ttingen, Germay, 1997.
-70-23oC
(14) Brellochs, B.; Binder, H. Angew. Chem., Int. Ed. Engl. 1988, 27,
262–3.
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 857

A R T I C L E S
Yoon et al.
In contrast to the computationally predicted structures, the
early X-ray crystallographic determination reported6 by Nord-
man and Reimann (NR) showed that 1 is highly distorted in
the solid-state with H1b located between B1 and B3, and with
H23 in an asymmetric position closer to B2 than B3. NR
suggested that 1 has a structure close to a double bridging
hydrogen conformation, but with the caveat that the single
bridging-hydrogen conformation shown in Figure 2a could not
be entirely excluded. The fact that the NR data were not of
sufficient quality to allow location of the NH3 hydrogen positions
also raises uncertainty about structural conclusions based on
their determination.
In order to help resolve the inconsistency between the
computational predictions and the NR structure, we redetermined
the structure of 1. The higher quality of the diffraction data in
this redetermination enabled the location and refinement of all
hydrogens. As shown in Figure 2b, the geometrical parameters
for the redetermined structure are, in fact, consistent with those
of the original NR structure but reveal an even more asymmetric
geometry, with H1b closer to B3, (B3-H1b, 1.685(13) Å), than
was found in the NR structure, (1.75(3) Å). This B3-H1b
distance is likewise much shorter than B2-H1a (2.266(13) Å),
causing the H1b-B1-B3 (66.0(7)°) angle to be much more
acute than H1a-B1-B2 (99.1(7)°), clearly indicating interac-
tions between B3 and H1b. Consistent with this interpretation,
B1-B3 (1.798(2) Å) is shorter than B1-B2 (1.819(2) Å) but
longer than B2-B3 (1.743(2) Å). Also as in the NR structure,
H23 is asymmetrically positioned between B2 and B3 with
B3-H23 (1.347(14) Å) being significantly longer than B2-H23
(1.179(14) Å). Therefore, in agreement with the previous NR
structure, the solid-state structure of 1 has a triborane unit with
a second partially hydrogen-bridging conformation, rather than
the computationally predicted single hydrogen-bridged structure.
A comparison of the bond distances and angles for the DFT-
optimized geometry of 1 with those of the redetermined structure
is given in the first two columns of Table 2, where it is apparent
that there are substantial differences in the two structures, not
only in the distances and angles involving H1b and H23, but
also with the B1-N (1.642 vs 1.585(2) Å); B1-B2 (1.849 vs
1.819(2) Å) and B1-B3 (1.849 vs 1.798(2) Å) distances, as
well as the B3-B1-N (108.1° vs 115.28(9)o) angle.
Figure 1. 11B{1H} NMR spectra for the reaction progression leading to
the formation of 1: (a) B3H8-, (b) (glyme)B3H7 formed after I2 oxidation
of B3H8 in glyme, (c) 1 formed following ammonia displacement of the
-
coordinated glyme (before purification), and (d) 1 after silica gel filtration.
Figure 2. (a) DFT (B3LYP/6-31G(d)) optimized structure of 1 and (b)
crystallographically redetermined structure of 1.
When the molecular geometry observed in the crystal
structure of 1 was computationally optimized at B3LYP/6-
31G(d), it reverted to that of the symmetric single-bridged
structure in Figure 2a. The question that then arises is: Why is
the solid-state structure of 1 significantly different from that of
the computationally predicted structure?
As shown in Figure 1, upon treatment with iodine at low
-
temperature, the B3H8 anion was oxidized to give a new
compound that exhibited broad 11B NMR resonances in a 2:1
ratio at -6.8 and -9.5 ppm, respectively, in agreement with
the DFT/GIAO calculated chemical shifts for (glyme)B3H7, -3.6
(B2,3) and -11.1 ppm (B1). Treatment of the glyme solution
with anhydrous ammonia at -70 °C then afforded 1. Any
unreacted B3H8- was easily removed by filtration of the reaction
solution through silica gel to give pure 1 in 80% yield. 1 could
also be purified by vacuum sublimation at 50 °C, but some
decomposition occurred lowering the yield to less than 50%.
All spectroscopic data for 1 are in agreement with the literature
values.9
Crabtree15 has shown that intermolecular dihydrogen bonding
between hydridic B-H (δ-) and protonic N-H (δ+) hydrogens
is important for ”BNH” compounds. The B-H· · ·H-N dihy-
drogen bonding distances are typically ∼1.7-2.2 Å, and, in
contrasttoclassicalhydrogen-bondinginteractions,theB-H· · ·H-N
unit is not linear, with the H-H-N angles usually larger,
117-171°, than the B-H-H angles, 95-120°.
The ORTEP plot presented in Figure 3a shows the relative
orientation of the molecular units in the solid-state crystal
structure of 1, where it can be seen that the three ammonia N-H
Structural Studies. There have been a number of computa-
tional studies that have predicted that 1 should have a CS-
symmetric structure with a single bridging hydrogen.5 The DFT-
optimized geometry along with selected bond distances for this
structure is shown in Figure 2a. In this structure, both H1a and
H1b are bound only to B1, and the H23 bridging hydrogen is
symmetrically located on the B2-B3 edge.
(15) (a) Richardson, T. B.; de Gala, S.; Crabtree, R. H. J. Am. Chem. Soc.
1995, 117, 12875–12876. (b) Crabtree, R. H.; Siegbahn, P. E. M.;
Eisenstein, O.; Rheingold, A. L.; Koetzle, T. F. Acc. Chem. Res. 1996,
29, 348–354. (c) Klooster, W. T.; Koetzle, T. F.; Siegbahn, P. E. M.;
Richardson, T. B.; Crabtree, R. H. J. Am. Chem. Soc. 1999, 121, 6337–
6343.
9
858 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Ammonia Triborane
A R T I C L E S
Table 2. Comparisons of Selected Bond Distances (Å) and Angles
(deg) from the DFT-Optimized Geometry of 1, the
Crystallographically Redetermined Structure of 1, the Crystal
Structure of 1·18-Crown-6 (Molecule a), and the DFT-Optimized
Geometry of 1·18-Crown-6 (Molecule a)
parameters
1 (DFT)
1 (exptl)
1·18-crown-6 (exptl) 1·18-crown-6 (DFT)
B1-N
1.642
1.849
1.849
1.729
1.324
1.324
1.203
1.203
2.157
2.157
1.208
1.195
1.208
1.195
1.022
1.021
1.022
55.8
1.585(2)
1.819(2)
1.798(2)
1.743(2)
1.179(14)
1.347(14)
1.097(13)
1.106(14)
1.685(13)
2.266(13)
1.087(14)
1.105(14)
1.080(14)
1.085(14)
0.87(2)
1.582(3)
1.829(4)
1.842(4)
1.725(5)
1.33(3)
1.27(3)
1.12(3)
1.09(3)
2.192(26)
1.933(26)
1.05(3)
1.14(2)
1.14(3)
1.09(3)
0.98(3)
0.95(3)
0.81(3)
1.607
1.857
1.859
1.725
1.335
1.321
1.211
1.209
2.176
2.091
1.204
1.199
1.205
1.199
1.026
1.026
1.025
B1-B2
B1-B3
B2-B3
B2-H23
B3-H23
B1-H1a
B1-H1b
B3-H1b
B2-H1a
B2-H2a
B2-H2b
B3-H3a
B3-H3b
N-H4a
N-H4b
N-H4c
0.91(2)
0.87(2)
B2-B1-B3
B1-B2-B3
B1-B3-B2
B2-B1-N
B3-B1-N
B2-H23-B3 81.6
H23-B2-B3 49.2
H23-B3-B2 49.2
H1a-B1–B2
H1b-B1–B3
H4a-O1
H4b-O3
H4c-O5
57.61(7)
60.58(7)
61.81(7)
110.98(9)
115.28(9)
87.0(9)
50.5(7)
42.5(6)
99.1(7)
66.0(7)
56.1(2)
62.4(2)
61.6(2)
55.3
62.1
62.1
108.1
108.1
62.4
62.3
112.6
111.5
81.0
49.1
49.9
83.1
87.6
113.7(2)
110.0(2)
83.0(5)
46.8(12)
50.1(13)
78.0(14)
93.4(14)
2.011(28)
2.068(22)
2.195(23)
87.3
87.3
-
-
-
-
2.027
2.043
2.046
-
-
hydrogens are each directed toward different B-H hydrogens
on adjacent triborane fragments, thus suggesting the possibility
of significant intermolecular B-H· · ·H-N dihydrogen bonding
interactions. The schematic diagram in Figure 3b provides more
details of the intermolecular interactions between the ammonia
hydrogens, H4a, H4b, and H4c, of a central NH3B3H7 and the
nearest B-H hydrogens, H1a, H2b, and H3a, on adjacent B3H7
fragments.
Figure 3. (a) ORTEP view showing the relative orientation of the molecular
units and potential intermolecular B-H· · ·H-N interactions in the solid-
state structure of 1 and (b) a schematic diagram showing the closest
intermolecular interactions in the solid-state structure of 1.
The H4b-H2b distance (2.13(3) Å) and the N-H4b-H2b
(146(2)°) and H4b-H2b-B2 (123(1)°) bond angles are all
within the ranges previously found for molecules having strong
B-H· · ·H-N dihydrogen bonding interactions. The H4b-H2b
distance is slightly longer than the closest N-H· · ·H-B distance
(2.02(3) Å)15c found in the neutron diffraction study of the solid-
state NH3BH3 structure. On the other hand, while having bent
B-H· · ·H and H · · ·H-N angles typical of dihydrogen bonding
values, the H1a-H4a (2.41(2) Å) and H4c-H3a (2.43(2) Å)
distances are close to normal nonbonding H · · ·H contacts (>2.4
Å), suggesting only weak interactions between these hydrogens.
These observations suggest that dihydrogen bonding of the
H4b and H2b hydrogens may give rise to the distorted solid-
state structure of 1. That is, the H4b-H2b interaction may
weaken the bonding between H2b and B2 and thereby enhance
B2-H23 bonding resulting in the shortening of B2-H23
(1.179(14) Å) and elongation of B3-H23 (1.347(14) Å). This
change could then induce a new bonding interaction between
B3 and H1b that results in a shortening of B3-H1b (1.685(13)
Å) compared to B2-H1a (2.266(13) Å).
structure as lowest energy. In order to provide computational
support for the effects of intermolecular dihydrogen bonding
on the solid-state structure of 1, several dimeric structures that
might serve as models for the types of intermolecular interac-
tions observed in the solid state were optimized by DFT at
B3LYP/6-31G(d). Two energetically favored dimeric structures
were located (Figure 4). The lower energy side-by-side structure
12 shows, as found in the solid-state structure of 1, dimeric
interactions (1.97 Å) between one of the terminal hydrogens
on B2 (but, in this case H2a′ instead of H2b) and one hydrogen
on the NH3 (but, in this case H4a′ instead of H4b) with the
bent B2′-H2a′-H4a′ (100.5°) and H2a′-H4a′-N4′ angles
(142.2°) again in the ranges observed for dihydrogen-bonding.
The 1.97 Å dihydrogen distance is somewhat longer than was
found in the PCI-80/B3LYP computational study15a of dihy-
drogen bonding in NH3BH3 dimers (1.82 Å). The MP2/6-
311+G(2d,p) calculated 11.0 kcal/mol stabilization of 12 relative
to 2 equiv of 1 (2.5 kcal/mol at B3LYP/6-311+G(2d,p)) is
similar to that found for the NH3BH3 dimer (12.1 kcal/mol)15a
at PCI-80/B3LYP. Even with the four (but longer) dihydrogen
interactions, the face-to-face dimer 12′ (Figure 4b) was found
to be 1.3 kcal/mol less stable than 12.
The stabilizing effects of these intermolecular interactions
are not included in the molecular gas-phase structural calcula-
tions of 1; therefore, such computations yield the CS-symmetric
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 859

A R T I C L E S
Yoon et al.
Table 3. Comparisons of Selected Bond Distances and Angles
Observed in the Solid-State Structure of 1 with Those Calculated
for the DFT-Optimized Structure of 12
1
12
B-N
1.585(2)
1.819(2)
1.798(2)
1.743(2)
1.097(13)
1.087(14)
1.685(13)
2.266(13)
1.179(14)
1.347(14)
1.086(14)
1.106(11)
1.081(11)
1.084(14)
0.87(2)
1.630
1.843
1.841
1.729
1.201
1.208
1.979
2.283
1.294
1.362
1.208
1.202
1.203
1.194
1.027
1.021
1.022
55.99
61.96
62.05
107.42
112.23
81.20
51.10
47.70
94.86
77.87
B1-B2
B1-B3
B2-B3
B1-H1a
B1-H1b
B3-H1b
B2-H1a
B2-H23
B3-H23
B2-H2a
B2-H2b
B3-H3a
B3-H3b
N4-H4a
N4-H4b
0.91(2)
0.87(2)
N4-H4c
B2-B1-B3
B1-B2-B3
B1-B3-B2
B2-B1-N
B3-B1-N
B2-H23-B3
H23-B2-B3
H23-B3-B2
H1a-B1-B2
H1b-B1-B3
57.61(7)
60.58(7)
61.81(7)
110.98(9)
115.28(9)
87.0(9)
50.5(7)
42.5(6)
99.1(7)
66.0(7)
Figure 4. B3LYP-optimized dimeric structures of 1: (a) 12 and (b) 12′.
The MP2/6-311+G(2d,p)-calculated energies are in kcal/mol and are based
on the B3LYP/6-31G(d)-optimized structures relative to two separated
NH3B3H7 molecules. Selected intermolecular distances and angles. 12:
H2a′-H4a′, 1.97 and 1.97 Å; B2-H2a′-H4a′, 100.5° and 100.5°;
H2a′-H4a′-N4′, 142.2° and 142.2°. 12′: H2′′-Ha′′, 2.15 and 2.15 Å;
H3′′-Hc′′, 2.14 and 2.14 Å; B2′′-H2′′-Ha′′, 118.9° and 119.4°;
H2′′-Ha′′-N4′′, 146.0° and 146.4°; B3′′-H3′′-Hc′′, 118.9° and 118.9°;
H3′′-Hc′′-N4′′, 146.7° and 147.0°.
bridge that is located in a symmetric position across the B2-B3
edge. The computational studies identified two dimeric
(MeNH2B3H7)2 (Me-12, -12.1 and Me-12′, -8.9 kcal/mol) and
one dimeric (Me2NHBH3)2 (Me2-12, -12.8 kcal/mol) structures
(Figure 7) that, as a result of their dihydrogen bonding
interactions, (B-H· · ·H-N distances: 1.98 Å, Me-12; 2.01 and
2.17 Å, Me-12′; 2.04 and 2.14 Å, Me2-12) were more stable
than 2 equiv of their respective monomers. As was the case for
1, while the DFT geometry optimizations of the monomeric
MeNH2B3H7 and Me2NHB3H7 structures yielded CS-symmetric
single hydrogen-bridging geometries, the optimized Me-12 and
Me2-12 structures exhibited asymmetric structures (Figure 8),
similar to those of both 12 and the solid-state 1 structure, again
confirming the structural consequences of intermolecular
B-H· · ·H-N dihydrogen bonding.
Further evidence for the effects of intermolecular dihydrogen
bonding on the solid-state structure of 1 comes from the
Figure 5. Comparison of (a) the solid-state structure of 1 with (b) the
DFT-optimized 12 structure. Dashed lines indicate the positions of the
dihydrogen bonding interactions.
While 12′ exhibited a nearly CS-symmetric NH3B3H7 frame-
work with a single-bridging hydrogen on each triborane
fragment, it is significant, as illustrated in Figure 5 and Table
3 that the lower energy structure 12 had an asymmetric structure
that resembles the observed solid-state structure of 1, with both
H1b moving closer to B3, and H23 moving away from B3
toward B2 such that it is asymmetrically bonded at the B2-B3
edge. Likewise, the B2-B1-N (107.42°) and B3-B1-N
(112.23°) angles are not equivalent.
Similar structural changes were computationally observed
between the DFT-optimized monomeric structures of
MeNH2B3H7 and Me2NHB3H7 and their dihydrogen-bonded
dimers. As shown in Figure 6, the DFT-optimized structures of
both MeNH2B3H7 and Me2NHB3H7 have only one hydrogen
Figure 6. DFT-optimized structures of MeNH2B3H7 (a, Me-1) and
Me2NHB3H7 (b, Me2-1).
9
860 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Ammonia Triborane
A R T I C L E S
Figure 7. DFT-optimized dimeric structures of (a) (MeNH2B3H7)2, Me-
12, (b) (MeNH2B3H7)2, Me-12′, and (c) (Me2NHB3H7)2, Me2-12. The MP2/
6-311+G(2d,p)-calculated energies are in kcal/mol and are based on the
B3LYP/6-31G(d)-optimized structures relative to 2 equiv of MeNH2B3H7
and Me2NHB3H7, respectively.
structural determination of the 1·18-crown-6 adduct. Ammonia
borane has previously been shown to form NH3BH3 ·18-crown-6
adducts, where the three ammonia hydrogens are hydrogen-
bonded to the 18-crown-6 oxygens.16 We likewise found that
the crystalline NH3B3H7 ·18-crown-6 adduct formed from a
THF/hexanes solution containing a 1:1 mixture of 1 and 18-
crown-6 (eq 8).
Figure 8. Selected bond distances in the model dimeric structures (a)
(MeNH2B3H7)2, Me-12, (b) (MeNH2B3H7)2, Me-12′, and (c) (Me2NHB3H7)2,
Me2-12 structures. The dashed lines indicate the positions of the dihydrogen
bonding interactions.
NH3B3H7 + 18-crown-6 f 1·18-crown-6
(8)
A structural determination showed two independent molecules
of 1·18-crown-6 in the asymmetric unit, but since there were
no significant differences in the two structures, only the structure
of molecule a, given in Figure 9, is discussed. The ammonia
hydrogens of the ammonia triborane unit each hydrogen bond
to one of the oxygen atoms of the crown ether with the longest
O-H distance (H4a-O1, 2.011(28) < H4b-O3, 2.068(22) <
H4c-O5, 2.195(23) Å) correlating with the shortest N-H
(N4-H4c (0.81(3) < N4-H4a (0.98(3) Å), N4-H4b (0.95(3)
Å). As a result of these favored O · · ·H-N hydrogen-bonding
interactions, there are no B-H· · ·H-N interactions less than
6.8 Å.
In contrast to the asymmetric structure observed in the solid-
state structure of pure 1, the ammonia triborane fragment in
1·18-crown-6 has a much more symmetric single hydrogen-
bridging framework that is more in line with the computational
results for 1. Moreover, unlike for 1, the DFT-optimized
structure of 1·18-crown-6 (Figure 10 and Table 2) is in good
agreement with the 1·18-crown-6 crystallographic determina-
tion. In both the crystal and DFT-optimized structures, H23 is
bonded in a symmetric fashion at the B2-B3 edge, and the
long (>1.93 Å) H1a-B2 and H1b-B3 distances clearly indicate
that H1a and H1b are bound only to B1. The slight asymmetry
observed in both the crystal and DFT-optimized structures (e.g.,
H1a-B2 < H1b-B3) of 1·18-crown-6 undoubtedly arises
because of the asymmetric ammonia conformation that is
dictated by its hydrogen-bonding interactions with the oxygen
atoms of the 18-crown-6. Thus, in the absence of any significant
B-H· · ·H-N dihydrogen interactions, the ammonia triborane
fragment in 1·18-crown-6 adopts the single hydrogen-bridging
conformation that is computationally favored for 1.
Hydrolytic Hydrogen Release. As described earlier by eq 1,
oxidative hydrolysis of 1 could release 8 equiv of H2 making
such a process potentially competitive with either NaBH4 or
NH3BH3 based hydrolytic H2-release systems. Two factors that
are critical for such applications are aqueous solubility and
stability. In this regard, owing most probably to its N-H· · ·O
hydrogen-bonding properties discussed in the previous section,
1 proved to be remarkably soluble in water, with solutions of
at least 33 wt % being attained. Furthermore unlike aqueous
NaBH4, which is stable only in strongly alkaline solutions,1a,b
NH3B3H7 was reasonably stable in neutral aqueous solutions,
as evidenced by the 11B NMR studies shown in Figure 11 of a
20 wt % solution of 1 which showed that, even after 11 days,
only a small amount had decomposed to ammonia borane and
borates. Likewise, in agreement with Kodama’s previous report
of the air stability of 1,4 11B NMR studies showed solid 1 stored
in air was not changed even after 2 months at room temperature.
Efficient H2-release from aqueous solutions of 1 was obtained
upon the addition of either acids or appropriate metal catalysts.
Acid hydrolysis of NH3BH3 has previously been demonstrated,2
(16) (a) Colquhoun, H. M.; Jones, G.; Maud, J. M.; Stoddart, J. F.; Williams,
D. J. J. Chem. Soc., Dalton Trans. 1984, 63–66. (b) Alston, D. R.;
Stoddart, J. F.; Wolstenholme, J. B.; Allwood, B.; Williams, D. J.
Tetrahedron 1985, 41, 2923–2926.
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 861

A R T I C L E S
Yoon et al.
Figure 9. (top) X-ray crystal structure of 1·18-crown-6 (molecule a) and
(bottom) selected distances in the ammonia triborane fragment of 1·18-
crown-6 (molecule a).
Figure 10. (top) DFT-optimized structure of 1·18-crown-6 (molecule a);
(bottom) selected calculated distances in the ammonia triborane fragment
of 1·18-crown-6 (molecule a).
and as shown by the quantitative gas buret measurements of
H2 release presented in Figure 12a, analogous reactions of 12
mg (0.21 mmol) of 1 with an excess of aqueous HCl (1 mL of
12.1 M HCl) gave a near-theoretical value (eq 9) of 7.85 H2
equiv (1.65 mmol) over ∼1 h.
-
NH3B3H7 + HCl + 6H2O f NH4+ + Cl- + “ 3BO2
+
3H+” + 8H2 (9)
The rate of acid hydrolysis was dependent upon the amount
of acid used. Thus, when 5.0 equiv of HCl was reacted with a
4.9 wt % solution of 1, 7.0 H2 equiv were liberated in 95 min
(Figure 12b), whereas a reaction with 3.0 equiv of HCl liberated
5.7 H2 equiv after 310 min and a reaction with only 1.0 equiv
of HCl gave less than 1.0 H2 equiv after 350 min.
Figure 11. 11B NMR spectra of a 20 wt % aqueous solution of 1 at room
temperature after (a) 0 h, (b) 24 h, (c) 96 h, (d) 147 h, (e) 11 days (b,
NH3B3H7; 1, NH3BH3).
Previous computational studies5f,17 of the acid-induced de-
hydrogenation of ammonia borane and ammonia triborane both
support a mechanism involving the initial elimination of 1 equiv
of H2 via the reaction of the acid proton with a hydridic B-H
hydrogen of the amine boranes. In aqueous solutions, subsequent
electrophilic association of the resulting cationic species (i.e.,
Hydrogen release from NaBH4 and NH3BH3 has previously
been attained by metal-catalyzed hydrolysis, with Ru catalysts1a,b
for NaBH4 and Pt catalysts2e for NH3BH3 exhibiting the highest
reactivities. A variety of these and other potential catalysts were
screened for activating the hydrolysis of 1, including Rh(0),
Ru(0), Pd(0), and Pt(0) supported on alumina or carbon; RhCl3,
[Rh(COD)Cl]2, RuCl3, and NiCl2 metal halides; and nano- or
microsized metal or NiB2 powders. As shown in the examples
in Figure 13, catalytic activity was found to be highly dependent
upon the particular metal, kind of supports, and degree of
dispersed catalytic sites.
+
NH3BH2 and NH3B3H6+) with water would then be expected
to lead to the observed rapid H2 release and formation of borates.
More rapid hydrogen release was achieved without acids by
using metals to catalyze the hydrolysis reaction (eq 10).
NH3B3H7 + 6H2O. Metal catalyst8 NH4+ + “3BO2- + 2H+” +
(17) Stephens, F. H.; Baker, R. T.; Matus, M. H.; Grant, D. J.; Dixon,
8H2(g) (10)
D. A. Angew. Chem., Int. Ed. 2007, 46 (5), 746–749.
9
862 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Ammonia Triborane
A R T I C L E S
Figure 14. Hydrogen evolution following repeated additions (11 cycles)
of 1 to a borate-buffered solution containing 5 wt % Rh/Al2O3.
afforded a measurable amount of hydrogen in 2 h. Although 5
wt % Pd/Al2O3, Ru(0) supported on an ion-exchange resin and
nickel boride also showed significant catalytic activity, their
reactivities were much less than either 5 wt % Rh/Al2O3 or
RhCl3, with less than 7.0 H2 equiv being evolved in 40 min.
Ru powder (22.6 mol%) was much less active, exhibiting less
than 3 H2 equiv in 400 min. Ni, Pd, and Pt powders and NiCl2
showed negligible H2 release in 2 h.
Unlike the immediate reaction observed with RhCl3, the
hydrolysis reaction of 1 with 7.1 mol % of RuCl3 showed an
induction period of ∼5-10 min. The formation of black
particles during this time suggests reduction of the RuCl3 to
form catalytically active Ru(0). A similar induction period was
previously observed for NaBH4 catalyzed by Ru supported on
carbon (2 wt % Ru).1d
Figure 12. Hydrogen release from (a) a 0.45 wt % aqueous solution of 1
upon reaction with excess HCl and (b) a 4.9 wt % aqueous solution of 1
upon reaction with 9, 5.0; b, 3.0; or 2, 1.0 equiv of hydrochloric acid.
The pH of a 0.45 wt % aqueous solution of 1 increased to above
8 after hydrolysis in the presence of [Rh(COD)Cl]2, and if more 1
was then added to this solution, then a decrease in the rate of H2
release was observed. On the other hand, if a borate buffer was
used to maintan the solution pH, then the Rh/Al2O3 catalyst
exhibited an extended lifetime. As indicated in Figure 14, hydrogen
evolution measurements following periodic additions of ∼9 mg
(∼0.16 mmol) of solid 1 to a 2 mL aqueous borate-buffered (pH
maintained between 7.2 and 8.0) solution containing 1.3 mg (0.012
mmol Rh) of 5 wt % Rh/Al2O3 showed little change in the
hydrogen release rates over 11 cycles.
Figure 13. Metal-catalyzed hydrolytic H2 release from aqueous (0.45 wt
%) solution of 1 containing (9) RhCl3 (6.9 mol %); (b) 5 wt % Rh/Al2O3
(7.0 mol % Rh); (1) RuCl3 (7.1 mol %) (inset); (2) [Rh(COD)µ-Cl]2 (7.1
mol %); (star) NixB (35-41 mol %); ([) 5 wt % Ru supported on ion
exchange resins (6.3 mol %); (left-pointing triangle) 5 wt % Pd/Al2O3 (6.7
mol %); (right-pointing triangle) Ru powder (22.6 mol %); (•) 5 wt % Rh/C
(5.3 mol %), NiCl2 (6.9 mol %), Ni powder (nanosize, 38.6 mol %), Pd
powder (sub-micrometer size, 9.0 mol %), Pt powder (nanosize, 6.5 mol
%), 5 wt % Ru/Al2O3 (7.3 mol %), (For these metal catalysts, the amount
of hydrogen was measured up to 120 min.) or no catalyst.
The “BO2-” product shown in eqs 9 and 10 is only a
hypothetical formulation, with the actual borate products of these
reactions depending upon the final solution concentration and
pH. The 11B NMR spectrum in Figure 15a of a solution obtained
after the metal-catalyzed hydrolysis of a 0.45 wt % solution of
1, exhibits a signal in the 15-17 ppm range resulting from a
-
The rhodium-based systems, Rh(0) supported on alumina and
RhCl3, proved to be the most active. For example, 5 wt % Rh(0)
supported on alumina gave over 7.5 H2 equiv in 15 min, while
RhCl3 yielded ∼7 H2 equiv in only 1.5 min. Although they were
unchanged upon initially dissolving in water, both [Rh(COD)(µ-
Cl)]2 and RhCl3 appeared to undergo reduction upon addition
of 1 suggesting that Rh clusters and/or colloids may be the active
catalytic species in these systems.18 In contrast, neither 5 wt %
Rh supported on carbon nor 5 wt % Ru supported on alumina
fast-exchanging equilibrium mixture of B(OH)3/B(OH)4 , along
with a weak signal at 12.4 ppm from B3O3(OH)4 .19 With more
-
concentrated solutions of 1 (Figure 15b (4.9 wt %) and c (22.7
wt %)), the signals for the condensed borates B3O3(OH)4- and
19
-
B5O6(OH)4 were dominant. Thus, although the 5 wt % Rh/
Al2O3 showed extended catalytic activity for the hydrolysis of
the diluted solution of 1, more concentrated solutions will
probably not exhibit such extended activity because the borates
will precipitate from the solutions and coat the catalyst surface.
(18) (a) Chen, Y.; Fulton, J.; Linehan, J.; Autrey, T. Prepr. Sym., DiV.
Fuel Chem. 2004, 49, 972–973. (b) Schulz, J.; Roucoux, A.; Patin,
H. Chem. Commun. 1999, 535–536.
(19) (a) Momii, R. K.; Nachtrieb, N. H. Inorg. Chem. 1967, 6, 1189–1192.
(b) Smith, H. D., Jr.; Wiersema, R. J. Inorg. Chem. 1972, 11, 1152–
1154. (c) Salentine, C. G. Inorg. Chem. 1983, 22, 3920–3924.
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 863

A R T I C L E S
Yoon et al.
Figure 15. 11B NMR spectra of aqueous 1 solutions (a) after hydrolysis
of 0.45 wt %, (b) after hydrolysis of 4.9 wt %, and (c) after hydrolysis of
22.7 wt % aqueous solution; b, an equilibrium mixture of B(OH)3 and
-
B(OH)4-; [, B3O3(OH)4-; f, B5O6(OH)4
.
Similar to the results previously observed for the catalyzed
NaBH41a,c,d and NH3BH32i,j hydrolysis reactions, the H2-release
rate appeared to be independent of the concentration of 1 with,
for example, identical rates observed for 4.9 and 22.7 wt %
solutions of 1 with 5 wt % Rh/Al2O3 (1.1 mol % Rh). However,
the H2-release rates of the 5 wt % Rh/Al2O3 and RhCl3 could
be effectively controlled and tuned by adjusting the catalyst
loadings and reaction temperatures. As illustrated in Figure 16a
for 5 wt % Rh/Al2O3-catalyzed reactions, while a 1.4 mol %
Rh loading gave 5.4 H2 equiv in 3 h, 7.0 mol % of Rh gave 7.6
H2 equiv in only 17 min. For a RhCl3-catalyzed reaction, 1.4
mol % Rh gave 5.6 H2 equiv in 40 min, while 6.9 mol % Rh
gave 7.0 H2 equiv in only 2 min.
As shown in Figure 16b, at 50 °C over 7 H2 equiv were
produced in 25 min from a 4.9 wt % aqueous solution of 1
catalyzed by 5 wt % Rh/Al2O3 (1.1 mol % Rh). Decreased rates
were observed at lower temperatures, but even at 0 °C, 6 H2
equiv were released in 17 h. An Arrhenius plot of the H2-release
rate data from a 5 wt % Rh/Al2O3 (1.1 mol % Rh) catalyzed
reaction of a 4.9 wt % 1 solution at different temperature yielded
an activation energy (13.4 kcal/mol) in the range found for
metal-catalyzed NaBH4 hydrolysis (∼9-18 kcal/mol, depending
on the catalyst1a-f).
Figure 16. (a) Effect of catalyst loadings on H2-release rates from a 0.45
wt % aqueous solution of 1 with different amounts of 5 wt % Rh/Al2O3
(9, 7.0; b, 2.9; 2, 1.4 mol %) and (b) the effect of temperature on H2
release from a 4.9 wt % aqueous solution of 1 catalyzed by 5 wt % Rh/
Al2O3 (1.1 mol% Rh). (1, 50; 2, 35; b, 25; 9, 0 °C).
material weight, this mixture corresponds to a 6.1 materials-wt
% H2 [i.e., mat-wt % H2 ) H2 wt/(NH3B3H7 + H2O + Rh/
Al2O3-wts)]. Therefore, the longer-term DOE total-system H2-
storage targets (2015, 9.0 wt %)21 for transportation will not
be attainable with this, or any other, hydrolytic-based system.
Nevertheless, an ammonia triborane-based hydrolytic system
can provide a means of safe hydrogen generation that should
be competitive with both the NH3BH3- and NaBH4-based
hydrolysis systems for other applications having less stringent
weight requirements. As will be discussed in future publications,
the efficient synthesis of 1 that was reported herein is now
enabling our ongoing explorations of these and other potential
applications, as well as our systematic investigation of the
chemistry of this unique compound.
Calculations of the standard heats of hydrolysis from the standard
enthalpies of formation, using Dixon’s value for the heat of
formation of solid NH3B3H7 (-24.2 kcal/mol),5d indicate that H2
release from NH3B3H7 is slightly more exothermic (18.9 kcal/mol
H2) than from either NaBH4 (14.9 kcal/mol H2) or NH3BH3 (12.7
kcal/mol H2) but is much less exothermic than the hydrolytic
reactions of metal hydrides (LiAlH4 (29.0 kcal/mol H2), MgH2 (33.5
kcal/mol H2), and LiH (31.6 kcal/mol H2).20
The hydrolysis reaction of a highly concentrated 22.7 wt %
sample containing 0.30 g of H2O, 0.10 g of 1 (1.8 mmol) and
0.04 g of 5 wt % Rh/Al2O3 (0.02 mmol of Rh) produced 0.027
g (13.5 mmol, 7.5 equiv) of H2 over 3 h at 21 °C. Since the
excess water solvent and catalyst must also be included as a
Acknowledgment. We thank the U.S. Department of Energy
for grants from the Division of Energy Efficiency and Renewable
Energy through the Center of Excellence for Chemical Hydrogen
Storage and from the Division of Basic Energy Sciences for the
support of this research. We also thank Dr. Goji Kodama for his
helpful comments.
Supporting Information Available: Tables of experimental
decomposition and hydrolytic H2-release data for ammonia
triborane and tables listing the Cartesian coordinates for all DFT-
optimized geometries; complete ref 10a. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA808045P
(20) (a) Messer, C. E.; Fasolino, L. G.; Thalmayer, C. E. J. Am. Chem.
Soc. 1955, 77, 4524–4526. (b) Davis, W. D.; Mason, L. S.; Stegeman,
G. J. Am. Chem. Soc. 1949, 71, 2775–2781.
(21) http://www.eere.energy.gov/hydrogenandfuelcells/storage/pdfs/targets_
onboard_hydro_storage.pdf.
9
864 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009