Acid initiation of ammonia-borane dehydrogenation for hydrogen storage

Article abstract of DOI:10.1002/anie.200603285

(Formula Presented) Liberating hydrogen: Strong Lewis or Bronsted acids react with ammonia-borane (H₃NBH₃) to form an in situ boronium cation, resulting in the formation of a mixture of cyclic and acyclic BNH_x oligomers and the liberation of H₂ (see scheme). A proposed mechanism is supported by an examination of the reaction thermodynamics using density functional theory.

Full text of DOI:10.1002/anie.200603285

Communications
DOI: 10.1002/anie.200603285
Hydrogen Storage
Acid Initiation of Ammonia–Borane Dehydrogenation for Hydrogen
Storage**
Frances H. Stephens, R. Tom Baker,* Myrna H. Matus, Daniel J. Grant, and David A. Dixon*
On-board storage is a major technical barrier to the use of
hydrogen as a transportation fuel.[1,2] Ammonia–borane
(H3NBH3, 1) has a hydrogen capacity of 19.6 wt%, exceeding
that of gasoline and making it an attractive candidate for
chemical hydrogen-storage applications.[3] At temperatures
up to 1008C, solid 1 affords a single equivalent of H2 and
aminoborane oligomers, [H2NBH2]n.[4] While both rate and
extent of dehydrogenation are increased in nanoscaffolds[5]
and concentrated ionic-liquid solutions of 1,[6] catalysts will be
required to meet the demands of transportation-based hydro-
gen storage.[7] Transition-metal catalysts have shown prom-
ise,[8–11] but the presence of a metal in the BNHx “spent fuel”
may complicate the required chemical regeneration of 1.
Herein we present a unique catalytic process for releasing H2
from ammonia–borane, wherein the dehydrocoupling of 1 is
initiated by small amounts of strong Lewis or Brønsted acids.
Denis et al. demonstrated that dehydrocoupling of phos-
phine–boranes H2RP·BH3 can be catalyzed by addition of the
Lewis acid tris(pentafluorophenyl)borane (B(C6F5)3).[12] They
proposed formation of the reactive intermediate H2PhP·
B(C6F5)3 by ligand exchange. Because B(C6F5)3 is more Lewis
at d = ꢀ25 ppm, a doublet (JB,H = 90 Hz) characteristic of the
anion [HB(C6F5)3]ꢀ resulting from hydride abstraction
[Eq. (1)].[14] Addition of 25 mol% B(C6F5)3 to 1 resulted in
H3NBH3 þ BðC6F5Þ3 ! ½BH2ðNH3Þꢁ½HBðC6F5Þ3ꢁ
ð1Þ
consumption of the starting material after about 24 h at 608C
with concomitant formation of an insoluble, colorless solid[4]
and 0.6 equivalents of H2.[15] Similar products were obtained
when the strong Brønsted acid trifluoromethanesulfonic acid
(HOSO2CF3)[16] was utilized. Protonolysis of the B Hbond
ꢀ
occurred, resulting in liberation of 0.25 equivalents of H2 (an
amount equal to the catalyst loading) and generating the
boronium cation [BH2(NH3)(L)]+ (2; L = solvent)
[Eq. (2)].[17–19] Subsequent dehydropolymerization liberated
an additional 0.6 equivalents of H2.[15]
H3NBH3 þ HOSO2CF3 ! ½BH2ðNH3Þꢁ½OSO2CF3ꢁ þ H2
ð2Þ
Stoichiometric, slow addition of a dilute solution of 1
in diethyl ether to
a
solution of [H(OEt2)2][B-
{3,5-(CF3)2C6H3}4][20] in diethyl ether resulted in formation
of the boronium salt [BH2(NH3)(OEt2)][B{3,5-(CF3)2C6H3}4]
(2a) in 84% yield. The electron-deficient boron center is
stabilized by one solvent molecule (L = Et2O). The 11B NMR
acidic than BH3,[13] the protic character of the P Hbond is
ꢀ
ꢀ
increased, enhancing its reactivity with the hydridic B H
bond of H2PhP·BH3.
Although Manners and co-workers reported that B(C6F5)3
does not catalyze the dehydrocoupling of dimethylamine–
borane,[8b] we observed reaction upon mixing of this Lewis
acid with 1 with concomitant H2 release. The 11B NMR
spectrum of the reaction mixture displayed a prominent peak
chemical shift (d = 0.21 ppm) and coupling constant (JB,H
=
125 Hz) are consistent with a four-coordinate boronium
cation in solution.[21] Isolation of this initiator will be
invaluable for future studies of reaction kinetics and isotope
effects.
The initial step of the reaction pathway is likely the
interaction of boronium 2 with one equivalent of 1 to form
intermediate 3 (Scheme 1). This configuration appears well-
[*] Dr. F. H. Stephens, Dr. R. T. Baker
Chemistry Division
Los Alamos National Laboratory
Mailstop J514, Los Alamos, NM 87507 (USA)
Fax: (+1)505-667-3314
E-mail: bakertom@lanl.gov
Dr. M. H. Matus, D. J. Grant, Prof. D. A. Dixon
Department of Chemistry
University of Alabama
Tuscaloosa, AL 35487 (USA)
Fax: (+1)205-348-9104
E-mail: dadixon@bama.ua.edu
[**] We thank the U.S. Department of Energy, Office of Energy Efficiency
and Renewable Energy for funding. F.H.S. thanks Los Alamos
National Laboratory for a postdoctoral fellowship. D.A.D. thanks the
Robert Ramsay Fund at the University of Alabama for partial
support. All calculations were completed at the Alabama Super-
computer Center. Martin Bluhm, Mark Bradley, and Larry Sneddon
are thanked for their assistance with hydrogen quantification and
helpful discussions.
Scheme 1. Reaction of 1 with a Lewis or Brønsted acid (A) results in
formation of borenium cation 2. Subsequent reaction with another
equivalent of 1 results in formation of 3 with subsequent expulsion of
H2 and concomitant formation of 4. Aproposed transition state for
the conversion of 3 into 4 is shown. Solvent and anion interactions
have been removed for clarity.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
746
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 746 –749

Angewandte
Chemie
ꢀ
positioned for loss of H2 with concomitant B N bond
species observed here and in ionic liquids are a mixture of
formation. We investigated the reaction pathway in detail
using density functional theory (DFT; gas phase) at the
B3LYP/DZVP2 level[22–24] utilizing the program Gaus-
sian03.[25] Calculations indicate that formation of intermedi-
ate 3 is strongly exothermic (ꢀ191 kJmolꢀ1), consistent with
stabilization of a cation by interaction with a neutral species
(Scheme 2). In contrast, loss of hydrogen from 3 to form 4 is
approximately thermoneutral (DH(298) = 5.2 kJmolꢀ1), and
is driven by entropy (DG(298) = ꢀ23.0 kJmolꢀ1). Subsequent
additions of 1 to the growing chain afford amino-bridged
species for which four-coordination at boron is maintained
after loss of H2. The less-substituted borenium cation 5b is
favored thermodynamically (DDH = ꢀ58 kJmolꢀ1) over the
more-substituted borenium cation 5a. The chain structure
with the fewest branches (6d) is the most stable, although the
branched structure investigated is similar in energy (6c:
ꢀ291 kJmolꢀ1; 6d: ꢀ309 kJmolꢀ1).[26]
linear and branched acyclic aminoborane oligomers. Obser-
vation of these neutral oligomers from acid-initiated dehy-
drogenation of 1 is indicative of rapid chain termination by
hydride abstraction from 1 (or other neutral oligomers;
Scheme 3).
Scheme 3. Chain termination can occur by hydride abstraction or by base
transfer. Chain transfer results in termination when the cation 4, for example,
abstracts Hꢀ from 1 or neutral oligomers. Boronium cation 2 can abstract NH3
from neutral species to form m-aminodiborane 7 without H2 evolution.
As dehydrocoupling of 1 progresses utilizing 10 mol% or
more acid, a broad triplet is observed in the 11B NMR
spectrum at d = ꢀ24 ppm. On the basis of variable-temper-
ature 11B NMR spectroscopy, this triplet is consistent with the
parent m-aminodiborane [B2H5(m-NH2)] (7),[27] not (primarily)
species 4 or 6. Formation of 7 is consistent with base (NH3)
transfer from neutral oligomers to 2 due to a rate mismatch of
The 11B NMR spectra of the product mixtures are similar
to those obtained in ionic liquids where no catalyst was
added.[6] The boronium borohydride [BH2(NH3)2][BH4], an
isomer of 1 that is chemically similar to 2, has been implicated
as the reaction initiator in the thermolysis of 1 in ionic liquids.
As validated by 11B NMR chemical shift calculations, the
Scheme 2. a) Possible reaction pathways investigated using DFT calculations. b) Calculated gas-phase energies DH(298 K) (and free energies
DG(298) in parentheses) of possible reaction intermediates and products. Energies (in kJmolꢀ1) are relative to that for 2 + [HB(C6F5)3]ꢀ at
0 kJmolꢀ1 for the following ions: 1 + B(C6F5)3, ꢀ407 (ꢀ377); 3, ꢀ195 (ꢀ146); 4, ꢀ190 (ꢀ218); 5a, ꢀ230 (ꢀ185); 5b, ꢀ288 (ꢀ249); 6a, ꢀ205
(ꢀ235); 6b, ꢀ250 (ꢀ275); 6c, ꢀ291 (ꢀ318); 6d, ꢀ309 (ꢀ377).
Angew. Chem. Int. Ed. 2007, 46, 746 –749
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
747

Communications
the initiation and propagation steps. As in the chain-transfer
mechanism, the reaction depicted in Scheme 3 results in
consumption of 1 without formation of additional H2 [Eq. (3),
solid-supported acids will initiate dehydropolymerization of
1, this chemistry may be promising for automotive applica-
tions if energy-efficient regeneration processes can be devel-
oped.
3 H3NBH3 þ acid ! H2 þ ½BH2ðNH3Þ2ꢁþ þ B2H5ðm-NH2Þ
ð3Þ
Experimental Section
General method for the reaction of 1 with acids: A solution of an acid
was prepared in diglyme and placed in a Schlenk flask. With the side
arm of the Schlenk flask attached to the gas burette, solid 1 was
rapidly added using a Merlic solid addition funnel. The reaction vessel
was then heated using an oil bath. A detailed synthetic protocol and
analytical information for 2a can be found in the Supporting
Information along with variable-temperature 11B NMR data support-
ing the assignment of compound 7.
a simplification of Scheme 1 and 2]. By reducing the ratio of
acid to 1, then, we should reduce the formation of 7 and
increase the amount of hydrogen evolved. Indeed, use of
0.5 mol% B(C6F5)3 resulted in significantly less 1 being
diverted to 7 and a larger amount (ca. 1.1 equiv) of H2
generated (Table 1).
Computational details: Geometries were optimized and frequen-
cies were calculated to ensure minima using density functional theory
(DFT) with the B3LYP exchange-correlation functional[22, 23] and the
polarized double-z/DZVP2 basis set.[24] The NMR chemical shifts
were calculated at the DFT level in the gauge-invariant atomic orbital
(GIAO) formalism[33] with a polarized triple-z basis set[34] and the
B3LYP functional. All calculations were done with the program
Gaussian03[25] on a Cray XD1 computer. Estimates of the salt-
formation energies for [2][X] (Xꢀ = CF3SO2Oꢀ, [HB(C6F5)3]ꢀ) were
made on the basis of the relationship given by Jenkins et al.[35, 36]
Calculated geometries of 1–6 can be found in the Supporting
Information.
Table 1: Reactivity of 1 with acids.
Acid
Loading
[mol%]
Conc. of
1 [m]
T [8C]
t [h]
H2
[equiv]
B(C6F5)3
B(C6F5)3
HOSO2CF3
HOSO2CF3
HCl
25
0.5
25
0.5
0.5
0.14[a]
2.6[b]
0.13[a]
6.2[c]
60
60
60
60
60
24
20
18
18
20
0.6
1.1
0.8
1.3
1.2
2.9[c]
[a] Reaction in glyme. [b] Reaction in tetraglyme. [c] Reaction in diglyme.
Received: August 11, 2006
Published online: November 28, 2006
Cyclic products are generated in these reactions, whereas
preliminary calculations (Scheme 2) indicated linear and
branched acyclic products. To probe this apparent dichotomy,
synthesis of [H2NBH2]n using the method of Wolf and co-
workers[4b] was undertaken followed by solution thermolysis
of this species. The 11B NMR spectra of [H2NBH2]n indicated
limited solubility in diglyme, but revealed cyclic products
after as little as 2.5 h at 608C. This differs from solid-state
thermolysis, whereby no decomposition of [H2NBH2]n is
observed at temperatures below about 1208C.[4c] An exami-
nation of the linear species BH3NH2BH2NH2BH2NH3 (8) at
the DFT and molecular orbital theory levels indicates that
cyclization can be initiated by coiling.[28] Formation of a short
Keywords: Brønsted acids · cations ·
.
density functional calculations · hydrogen · Lewis acids
[1] “National Research Council and National Academy of Engi-
neering, Committee on Alternatives and Strategies for Future
Hydrogen Production and Use”: The Hydrogen Economy:
Opportunities, Costs, Barriers, and R&D Needs, National Aca-
demies Press, Washington, DC, 2004.
[2] Basic Research Needs for the Hydrogen Economy (Eds.: M.
Dresselhaus, G. Crabtree, M. Buchanan), Basic Energy Sciences,
Office of Science, US Department of Energy, Washington, DC,
2003.
B H···H N interaction,[29,30] oriented to release H2, stabilizes
8 by 97.1 kJmolꢀ1. Therefore, in solution dehydrocyclization
is facile, whereas the same process is precluded in the solid
state as a result of hindered molecular motion.
ꢀ
ꢀ
[3] E. Fakioglu, Y. Yurum, T. Nejat Veziroglu, Int. J. Hydrogen
Energy 2004, 29, 1371 – 1376.
[4] a) G. Wolf, J. Baumann, F. Baitalow, F. P. Hoffmann, Thermo-
chim. Acta 2000, 343, 19 – 25; b) F. Baitalow, J. Baumann, G.
Wolf, K. Jaenicke-Rossler, G. Leitner, Thermochim. Acta 2002,
391, 159 – 168; c) J. Baumann, F. Baitalow, G. Wolf, Thermochim.
Acta 2005, 430, 9 – 14.
[5] A. Gutowska, L. Li, Y. Shin, C. M. Wang, X. S. Li, J. C. Linehan,
R. S. Smith, B. D. Kay, B. Schmid, W. Shaw, M. Gutowski, T.
Autrey, Angew. Chem. 2005, 117, 3644 – 3648; Angew. Chem. Int.
Ed. 2005, 44, 3578 – 3582.
In summary, we have shown that strong Lewis and
Brønsted acids initiate dehydrocoupling of 1 under mild
conditions. The novel hydride-abstraction pathway differs
from that proposed by Denis et al. for dehydropolymerization
of phosphine–boranes proceeding via a H2RP·B(C6F5)3 inter-
mediate.[12] This difference likely stems from the stronger B
ꢀ
[31,32]
ꢀ
N versus B P dative bonds.
The formation of acyclic
aminoborane oligomers by chain transfer is followed by facile
dehydrocyclization above 608C to afford borazine and addi-
tional H2. The boronium borohydride [BH2(NH3)2][BH4]
implicated in thermolyses of 1 in ionic liquids is structurally
similar to the boronium cations discussed here, and isolation
of boronium salt 2a will permit further studies. As we have
been able to effect these reactions using the low-molecular-
weight acid HCl in nonvolatile polyethers, and we postulate
that use of nonvolatile acids such as polyphosphoric acid or
[6] M. E. Bluhm, M. G. Bradley, R. Butterick III, U. Kusari, L. G.
Sneddon, J. Am. Chem. Soc. 2006, 128, 7748 – 7749.
[7] Uncatalyzed thermolysis of dilute ethereal solutions of ammo-
nia–borane is slow: J. S. Wang, R. A. Geanangel, Inorg. Chim.
Acta 1988, 148, 185 – 190.
[8] a) C. A. Jaska, K. Temple, A. J. Lough, I. Manners, Chem.
Commun. 2001, 962 – 963; b) C. A. Jaska, K. Temple, A. J.
Lough, I. Manners, J. Am. Chem. Soc. 2003, 125, 9424 – 9434.
[9] Y. Chen, J. L. Fulton, J. C. Linehan, T. Autrey, J. Am. Chem. Soc.
2005, 127, 3254 – 3255.
748
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 746 –749

Angewandte
Chemie
[10] M. C. Denney, V. Pons, T. J. Hebden, D. M. Heinekey, K. I.
Goldberg, J. Am. Chem. Soc. 2006, 128, 12048 – 12049.
[11] T. J. Clark, C. A. Russell, I. Manners, J. Am. Chem. Soc. 2006,
128, 9582 – 9583.
[24] N. Godbout, D. R. Salahub, J. Andzelm, E. Wimmer, Can. J.
Chem. 1992, 70, 560 – 571.
[25] Gaussian03 (RevisionC.02): M. J. Frisch et al., see Supporting
Information.
[12] J.-M. Denis, H. Forintos, H. Szelke, L. Toupet, T.-N. Pham, P.-J.
Madec, A.-C. Gaumont, Chem. Commun. 2003, 54 – 55.
[13] M. A. Beckett, D. S. Brassington, S. J. Coles, M. B. Hursthouse,
Inorg. Chem. Commun. 2000, 3, 530 – 533.
[14] J. M. Blackwell, E. R. Sonmor, T. Scoccitti, W. E. Piers, Org.
Lett. 2000, 2, 3921 – 3923.
[26] There is little temperature dependence of DH or DG. When T is
increased to 398 K, DH changes by no more than 8 kJmolꢀ1 and
DG changes by no more than 12 kJmolꢀ1
.
[27] L. D. Schwartz, P. C. Keller, J. Am. Chem. Soc. 1972, 94, 3015 –
3018.
[28] J. Li, S. M. Kathmann, G. K. Schenter, J. C. Linehan, T. Autrey,
M. Gutowski, Prepr. Am. Chem. Soc. Div. Fuel Chem. 2006, 51,
577 – 581.
[15] Under the same reaction conditions without B(C6F5)3, 1 released
only 0.05 equivalents of H2.
[16] F. G. Bordwell, Acc. Chem. Res. 1988, 21, 456 – 463.
[17] The reaction in Equation (1), in which the product is a solid salt
with a close anion–cation contact, is predicted to be near
thermoneutral (6.7 kJmolꢀ1). The reaction in Equation (2), in
which CF3SO2Oꢀ is coordinated to 2, is predicted to be
exothermic (ꢀ40.2 kJmolꢀ1 at 298 K). The complexation
energy of glyme with 2 is predicted to be ꢀ187.9 kJmolꢀ1
(298 K, gas phase).
[18] W. F. Schneider, C. K. Narula, H. Nꢀth, B. E. Bursten, Inorg.
Chem. 1991, 30, 3919 – 3927.
[19] W. E. Piers, S. C. Bourke, K. D. Conroy, Angew. Chem. 2005, 117,
5142 – 5163; Angew. Chem. Int. Ed. 2005, 44, 5016 – 5036.
[20] M. Brookhart, B. Grant, A. F. Volpe, Jr., Organometallics 1992,
11, 3920 – 3922.
[29] W. T. Klooster, T. F. Koetzle, P. E. M. Siegbahn, T. B. Richard-
son, R. H. Crabtree, J. Am. Chem. Soc. 1999, 121, 6337.
[30] a) R. Custelcean, Z. A. Dreger, J. Phys. Chem. B 2003, 107,
9231 – 9235; b) S. Trudel, D. F. R. Gilson, Inorg. Chem. 2003, 42,
2814 – 2816.
[31] D. J. Grant, D. A. Dixon, J. Phys. Chem. A 2005, 109, 10138 –
10147.
[32] D. A. Dixon, M. Gutowski, J. Phys. Chem. A 2005, 109, 5129 –
5135.
[33] K. Wolinski, J. F. Hinton, P. Pulay, J. Am. Chem. Soc. 1990, 112,
8251 – 8260.
[34] A. Schaefer, H. Horn, R. Ahlrichs, J. Chem. Phys. 1992, 97,
2571 – 2577.
[35] H. D. B. Jenkins, H. K. Roobottom, J. Passmore, L. Glasser,
Inorg. Chem. 1999, 38, 3609 – 3620.
[36] H. D. B. Jenkins, D. Tudela, L. Glasser, Inorg. Chem. 2002, 41,
2364 – 2367.
[21] N. E. Miller, E. L. Muetterties, J. Am. Chem. Soc. 1964, 86,
1033 – 1038.
[22] A. D. Becke, J. Chem. Phys. 1993, 98, 1372 – 1377.
[23] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785 – 789.
Angew. Chem. Int. Ed. 2007, 46, 746 –749
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
749