Indirect, reversible high-density hydrogen storage in compact metal ammine salts

Article abstract of DOI:10.1021/ja076762c

The indirect hydrogen storage capabilities of Mg(NH₃) ₆Cl₂, Ca(NH₃)₈Cl₂, Mn(NH₃)₆Cl₂, and Ni(NH₃) ₆Cl₂ are investigated. All four metal ammine chlorides can be compacted to solid tablets with densities of at least 95% of the crystal density. This gives very high indirect hydrogen densities both gravimetrically and volumetrically. Upon heating, NH₃ is released from the salts, and by employing an appropriate catalyst, H₂ can be released corresponding to up to 9.78 wt % H and 0.116 kg H/L for the Ca(NH ₃)₈Cl₂ salt. The NH₃ release from all four salts is investigated using temperature-programmed desorption employing different heating rates. The desorption is found mainly to be limited by heat transfer, indicating that the desorption kinetics are extremely fast for all steps. During desorption from solid tablets of Mg(NH₃) ₆Cl₂, Mn(NH₃)₆Cl₂, and Ni(NH₃)₆Cl₂, nanoporous structures develop, which facilitates desorption from the interior of large, compact tablets. Density functional theory calculations reproduce trends in desorption enthalpies for the systems studied, and a mechanism in which individual chains of the ammines are released from the surface of the crystal is proposed to explain the fast absorption/desorption processes.

Full text of DOI:10.1021/ja076762c

Published on Web 06/13/2008
Indirect, Reversible High-Density Hydrogen Storage in
Compact Metal Ammine Salts
†
‡,§
†
Rasmus Z. Sørensen, Jens S. Hummelshøj, Asbjørn Klerke,
†
§
‡
,†
Jacob Birke Reves, Tejs Vegge, Jens K. Nørskov, and Claus H. Christensen*
Center for Sustainable and Green Chemistry, Department of Chemistry, Building 206, Technical
UniVersity of Denmark, DK-2800 Kgs. Lyngby, Denmark, Center for Atomic-scale Materials
Design, Department of Physics, Building 310, Technical UniVersity of Denmark, DK-2800 Kgs.
Lyngby, Denmark, and Materials Research Department, Risø National Laboratory for Sustainable
Energy, NanoDTU, Building 228, Technical UniVersity of Denmark, DK-4000 Roskilde, Denmark
Received September 14, 2007; E-mail: chc@kemi.dtu.dk
ࠗ
w
This paper contains enhanced objects available on the Internet at http://pubs.acs.org/jacs.
Abstract: The indirect hydrogen storage capabilities of Mg(NH
Ni(NH Cl are investigated. All four metal ammine chlorides can be compacted to solid tablets with densities
of at least 95% of the crystal density. This gives very high indirect hydrogen densities both gravimetrically
and volumetrically. Upon heating, NH is released from the salts, and by employing an appropriate catalyst,
can be released corresponding to up to 9.78 wt % H and 0.116 kg H/L for the Ca(NH Cl salt. The
NH release from all four salts is investigated using temperature-programmed desorption employing different
heating rates. The desorption is found mainly to be limited by heat transfer, indicating that the desorption
kinetics are extremely fast for all steps. During desorption from solid tablets of Mg(NH Cl , Mn(NH Cl
and Ni(NH Cl , nanoporous structures develop, which facilitates desorption from the interior of large,
3 6 2 3 8 2 3 6 2
) Cl , Ca(NH ) Cl , Mn(NH ) Cl , and
)
3 6
2
3
H
2
3
)
8
2
3
3
)
6
2
3
)
6
2
,
3
)
6
2
compact tablets. Density functional theory calculations reproduce trends in desorption enthalpies for the
systems studied, and a mechanism in which individual chains of the ammines are released from the surface
of the crystal is proposed to explain the fast absorption/desorption processes.
8
–14
Introduction
direct storage of hydrogen have been proposed.
Clearly,
direct hydrogen storage methods will involve the lowest possible
number of chemical transformations, and thereby minimize the
inherent losses associated with multistep chemical reactions, but
in practice there are still a large number of challenges to be
dealt with before such large-scale direct hydrogen storage is
technically and economically feasible. This has led to increased
interest in indirect storage of hydrogen, e.g., in the form of
The rising concern over dwindling resources and the envi-
ronmental impact of burning fossil fuels has generated interest
in using hydrogen as an alternative energy carrier. Significant
challenges remain in the development of economically viable
solutions for production, storage, and use of hydrogen as an
1
–6
energy carrier.
Widespread use of hydrogen as a fuel is limited by the lack
7
15–18
of safe and efficient systems for its storage. The large deviation
methanol, ethanol, ammonia, urea, or guanidine.
from ideality observed upon compression of hydrogen to high
pressures, combined with the low condensing point and low
density even in the liquid state, has limited the use of
conventional storage systems, and a plethora of strategies for
(8) Z u¨ tel, A. Naturwissenschaften 2004, 91, 157–172.
(
9) Cheng, P.; Xiong, Z.; Lou, J.; Lin, J.; Tan, K. L. Nature 2002, 420,
3
02–304.
(
10) Bououdina, M.; Grant, D.; Walker, G. Int. J. Hydrogen Energy 2006,
31, 177–182.
†
Center for Sustainable and Green Chemistry, Department of Chemistry.
Center for Atomic-scale Materials Design, Department of Physics.
Materials Research Department, Risø National Laboratory for Sustain-
‡
(11) Stephens, F. H.; Baker, R. T.; Matus, M. H.; Grant, D. J.; Dixon,
D. A. Angew. Chem., Int. Ed. 2006, 46, 746–749.
§
able Energy.
(12) Bogdanoviæ, B.; Felderhoff, M.; Pommerin, A.; Sch u¨ the, F.; Spielka-
mp, N. AdV. Mater. 2006, 18, 1198–1201.
(
1) Dresselhaus, M.; Crabtree, G.; Buchanan, M. Basic Research Needs
for the Hydrogen Economy; U.S. Department of Energy: Washington,
DC, 2003. (available online at http://www.sc.doe.gov/bes/hydrogen.
pdf).
(13) Latroche, M.; Surbl e´ , S.; Serre, C.; Mellot-Draznieks, C.; Llewellyn,
P. L.; Lee, J.-H.; Chang, J.-S.; Jhung, S. H.; F e´ rey, G. Angew. Chem.,
Int. Ed. 2006, 45, 8227–8231.
(
2) Crabtree, G.; Dresselhaus, M.; Buchanan, M. Phys. Today 2004, 57,
(14) Welch, G. C.; Juan, R. R. S.; Masuda, J. D.; Stephan, D. W. Science
2006, 314, 1124–1126.
1
2, 39–44.
(
(
(
3) Kennedy, D. Science 2004, 305, 917.
(15) Metkemeijer, R.; Achard, P. Int. J. Hydrogen Energy 1994, 19, 535–
542.
4) Turner, J. A. Science 2004, 305, 972–974.
5) Wu, W.; Kawamoto, K.; Kuramochi, H. J. Mater. Cycles Waste
Manag. 2006, 8, 70–77.
(16) Thomas, G.; Parks, G. Potential Roles of Ammonia in a Hydrogen
Economy; U.S. Department of Energy: Washington, DC, 2006.
(available online at http://hydrogen.energy.gov/pdfs/nh3_paper.pdf).
(17) Christensen, C. H.; Johannessen, T.; Sørensen, R. Z.; Nørskov, J. K.
Catal. Today 2006, 111, 140–144.
¨
6) Avci, A. K.; Onsan, Z. I.; Trimm, D. L. Top. Catal. 2003, 22, 359–
(
(
3
67.
7) Takimoto, M.; Hou, Z. Nature 2006, 443, 400–401.
8
660 9 J. AM. CHEM. SOC. 2008, 130, 8660–8668
10.1021/ja076762c CCC: $40.75  2008 American Chemical Society

Hydrogen Storage in Compact Metal Ammine Salts
A R T I C L E S
Table 1. Mass and Volume of 10 kg of Hydrogen Stored
a
Reversibly by Six Different Methods
3 6
Mg(NH ) Cl
2
H
2
(liquid)
Mg NiH
2
4
LaNi H
5 6
NaAlH
4
H
2
(g, 200 bar)
9
1
1.4 L
09 kg
141.2 L
10 kg
252.6 L
392 kg
276.9 L
730 kg
380.9 L
286 kg
714.2 L
10 kg
a
All data are based on the best-obtained reversible densities reported
in the literature without considering the space occupied by the
2
6–32
container.
Molecular ammonia contains 17.8 wt % hydrogen and can
1
9
be used directly as a fuel in internal combustion engines or
2
0
in solid oxide fuel cells, but it can also be catalytically
converted to hydrogen and nitrogen at temperatures below 650
Figure 1. Principle of hydrogen storage in metal ammine salts of the general
2
1–23
K
in high yields. The production of ammonia from
formula M(NH ) X .
3
n
m
hydrogen and nitrogen using natural gas as the feedstock has
24
been optimized in the Haber-Bosch process, which is among
25
the largest catalytic processes currently in industrial operation.
Commercially, ammonia is transported as a liquid at a pressure
around 16 bar, achieved by addition of nitrogen, but for
distribution to consumers the safety of this solution is a
2
6
concern.
In Table 1, the practically obtainable hydrogen storage density
of a selection of the more promising direct hydrogen storage
approaches is compared to the results of a recently published
method of hydrogen storage in the ammine complex
Mg(NH3)6Cl2. From Table 1, it is evident that reversible storage
of 10 kg of H in a compact and convenient way is not a trivial
task. Since the comparison indicates that the hexaammine
magnesium salt is a promising candidate for hydrogen storage,
it is interesting to explore the hydrogen storage potential for a
broader range of metal ammine salts. Recently, metal ammines
have also been proposed for ammonia storage in connection
Figure 2. Van’t Hoff plot showing the equilibrium pressure of ammonia
over 90 metal ammine halides of the general formula M(NH3)nX2. The
complete list of complexes is given as Supporting Information.
3
4
ammonia storage materials to demonstrate that the family of
metal ammine salts, in fact, represents a large number of
compounds that can potentially be used for reversible, indirect
hydrogen storage. Furthermore, the investigated metal ammine
complexes are found to exhibit facile ammonia release kinetics.
The general principle of storing hydrogen in metal ammine
complexes is shown in Figure 1. The absorption of ammonia is
exothermic (negative absorption enthalpy) for all salts. Thus,
the release is necessarily endothermic and must require the same
amount of energy as is released during the absorption. The
equilibrium vapor pressures can be determined by the van’t Hoff
relationship given as eq 1, from which it is seen that the safer
the storage material (low vapor pressure at ambient temperature),
the more energy required for releasing NH3 at a given pressure.
3
3
with vehicular DeNOx systems, giving an extra incentive for
studying these materials in more detail.
Here, Ca(NH3)8Cl2, Mn(NH3)6Cl2, and Ni(NH3)6Cl2 are
investigated and compared with Mg(NH3)6Cl2 as hydrogen/
(
18) Van Vechten, J. A. Presented at the 2007 APS March Meeting, Denver,
CO;; American Physical Society: College Park, MD, 2007; Abstract
S39.00010.
(
20) Fournier, G. G. M.; Cumming, I. W.; Hellgardt, K. J. Power Sources
2
006, 162, 198–206.
(
(
19) Steele, R. B. Chemtech 1999, (August), 28.
21) Rar o´ g-Pilecka, W.; Szmigiel, D.; Kowalczyk, Z.; Jodzis, S.; Zielinski,
J. J. Catal. 2003, 218, 465–469.
(
(
22) Li, X.-K.; Ji, W.-J.; Zhao, J.; Wang, S.-J.; Au, C.-T. J. Catal. 2005,
2
36, 181–189.
23) Boisen, A.; Dahl, S.; Nørskov, J. K.; Christensen, C. H. J. Catal. 2005,
2
30, 309–312.
∆
Hdes ∆Sdes
(
(
24) Schl o¨ gl, R. Angew. Chem., Int. Ed. 2003, 42, 2004–2008.
25) Eggmann, T. Kirk-Othmer Encyclopedia of Chemical Technology:
Ammonia; John Wiley & Sons, Inc.: New York, 2001.
ln(p) ) -
+
(1)
RT
R
The possibility for choosing a metal ammine salt with the
desired thermodynamic properties is illustrated in Figure 2.
Compared to hydrides and alanates with comparable storage
pressures, the pressure change with temperature is larger for
metal ammines. This is because the ammine salts generally have
higher entropies of desorption than those of metal hydrides.
Thus, it is evident from the van’t Hoff relationship that metal
complexes also have higher enthalpies of desorption than
hydrides with similar vapor pressures at a given temperature.
The formation of metal ammine complexes and the thermo-
dynamics of ammonia desorption from these have been thor-
oughly described in the early chemical literature. From these
literature data, it is possible to find general trends for the NH3
(
26) Appl, M. Ullmann’s Encyclopedia of Industrial Chemistry: Ammonia;
Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, 2007.
27) Mosher, D.; Tang, X.; Arsenault, S. High Density Hydrogen Storage
System Demonstration Using NaAlH4 Based Complex Hydrides. FY
(
2006 Annual Progress Report, DOE Hydrogen Program; U.S. Depart-
ment of Energy: Washington, DC, 2006; pp 281-284. (available online
at http://www.hydrogen.energy.gov/pdfs/progress06/iv_a_1_mosher.pdf).
28) Liu, X.; Zhu, Y.; Li, L. Int. J. Hydrogen Energy 2007, 32, 2455–
(
(
(
(
(
(
2
460.
29) Nomura, K.; Fujiwara, S.; Hayakawa, H.; Akiba, E.; Ishido, Y.; Ono,
S. Journal of the Less-Common Metals 1991, 169, 9–17.
30) Suissa, E.; Jacob, I.; Hadari, Z. J. Less-Common Metals 1984, 104,
2
87–295.
31) El-Osairy, M. A.; el-Osery, I. A.; Metwally, A. M.; Hassan, M. A.
Int. J. Hydrogen Energy 1993, 18, 517–524.
32) Laidler, K. J.; Meiser, J. H. Physical Chemistry, 3rd ed.; Houghton
Mifflin Co.: New York, 1999.
33) Elmøe, T. D.; Sørensen, R. Z.; Quaade, U.; Christensen, C. H.;
Nørskov, J. K.; Johannessen, T. Chem. Eng. Sci. 2006, 61, 2618–
(34) Biltz, W.; Messerknecht, C. Z. Anorg. Allg. Chem. 1923, 129, 161–
2
625.
175.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008 8661

A R T I C L E S
Sørensen et al.
3
5–37
desorption properties of metal ammine halides.
The most
Table 2. Indirect Hydrogen Storage Capacity of Four Metal
Ammine Salts
general trend in these original data is that the desorption enthalpy
of ammonia increases from chloride through bromide to iodide.
The data available for fluorides indicate that fluoride does indeed
follow the same trend, but, in fact, only a few metal ammine
a
3
F, g/cm
gravimetric H, wt % H
volumetric H, kg H/L
Mg(NH3)6Cl2
Ca(NH3)8Cl2
Mn(NH3)6Cl2
Ni(NH3)6Cl2
1.25
1.19
1.41
1.53
9.19
9.78
7.96
7.83
0.115
0.116
0.112
0.119
3
5,37–39
fluorides have been investigated in detail.
The effect of
the metal cation is not as clear, but some trends can be seen.
For the alkali and alkaline earth metals, the enthalpy of
a
42
Crystal densities.
3
7,38
desorption decreases down through the groups,
and for the
Table 3. Ammonia Desorption Enthalpies for Each Desorption
transition metals, the enthalpy of desorption increases slightly
when moving from left to right in the Periodic Table, i.e., from
43
Step for Four Different Metal Ammine Complexes
3
7
manganese to nickel in oxidation state 2+.
3 n
Mg(NH ) Cl
2
3 n
Ca(NH ) Cl
2
3 n
Mn(NH ) Cl
2
3 n 2
Ni(NH ) Cl
For a metal ammine complex to be considered useful as an
indirect hydrogen storage material, it needs to desorb ammonia
in a relatively narrow temperature range around or above
ambient temperature. For the storage to be safe, the ammonia
vapor pressure should preferably be below 1 bar at ambient
temperature. Possibly, somewhat higher pressures could be
handled appropriately in practical systems. However, with such
materials, leaks would represent a significant hazard. At the same
time, 1 bar of ammonia pressure should preferably be reached
below 650 K for all desorption steps to avoid desorption of
ammonia becoming too energy intensive. This is so because
the ammonia decomposition reaction is best conducted above
∆Hdes
,
∆Hdes
,
∆Hdes
,
∆Hdes,
kJ/mol
n
kJ/mol
n
kJ/mol
n
kJ/mol
n
6f2
f1
f0
55.7
74.9
87.0
8f4
4f2
2f1
41.0
42.3
63.2
69.1
6f2
2f1
1f0
47.4
71.0
84.2
6f2
2f1
1f0
59.2
79.5
89.8
2
1
1
f0
which the equilibrium vapor pressure is 1 bar for the first
desorption step in the four different complexes are 305
Ca(NH3)8Cl2), 358 (Mn(NH3)6Cl2), 413 (Mg(NH3)6Cl2), and
(
43
4
49 K (Ni(NH3)6Cl2). The theoretical storage capacities of
the metal ammine complexes are given in Table 2, and
desorption enthalpies for the individual desorption steps are
reported separately in Table 3.
6
50 K, where a sufficiently high rate can be achieved and
simultaneously a sufficiently low equilibrium ammonia con-
centration is reached. Moreover, the gravimetric and volumetric
hydrogen density of the chosen metal ammine salt(s) should
clearly be as high as possible. This criterion obviously favors
light cations and anions such as alkali metals and fluorides.
These are, however, impractical because the alkali metals do
not bind ammonia sufficiently well at ambient temperature
according to the above criteria, and the fluorides are usually
toxic and can form hydrofluoric acid when they are heated to
In utilizing the present approach for indirect hydrogen storage
above the gram scale, it is important that the complexes can be
compacted into tablets or other shaped bodies with as little void
space as possible. This was previously reported to be possible
for Mg(NH3)6Cl2, and it was found that, during desorption of
ammonia from tablets of this salt, a sponge-like structure
maintaining the shape of the original tablet was formed featuring
a nanopore system, which facilitates desorption of ammonia
3
9
desorb the ammonia.
44,45
from the interior of the compact material.
So far, the only metal ammine complex which has been
investigated in any detail as an indirect hydrogen storage
The compactability, the ability to form of nanopores, and the
kinetics of ammonia desorption are investigated in this study
for Ca(NH3)8Cl2, Mn(NH3)6Cl2, and Ni(NH3)6Cl2, and the results
are reported in the following sections.
4
0
material is Mg(NH3)6Cl2. Mg(NH3)6Cl2 was chosen initially
because it has a vapor pressure of only 2.2 mbar at 300 K, and
additionally MgCl2 is both widely available and nontoxic.
However, other salts can similarly bind ammonia to form
interesting indirect hydrogen storage materials. For one, CaCl2
binds eight ammonia molecules to form Ca(NH3)8Cl2 at 300 K
and 1 bar of NH3. This gives an even higher indirect hydrogen
storage density than that achieved in Mg(NH3)6Cl2 on both a
mass and volume basis, but it also results in an equilibrium
ammonia pressure of 0.77 bar at 300 K. In Ca(NH3)8Cl2, only
six of the NH3 molecules are coordinated directly to calcium.
Experimental Methods
Commercial anhydrous salts (CaCl
Aldrich, 98%; MnCl , Aldrich, 98%) were transferred to the reaction
vessel in a glovebox containing dry air (6-8 ppm H O) and dried
at 400-500 °C in a stream of N before use. The vessel was purged
with NH gas (Hede Nielsen, N45) and left overnight under a
pressure of NH slightly above 4 bar. NH uptakes were determined
2 2
, Alfa Aesar, 97%; NiCl ,
2
2
2
3
3
3
gravimetrically. Each metal ammine halide was pressed into tablets
to determine the maximal bulk density of the storage material.
Tablets of the material were also subjected to measurements of
pore size distributions. These were performed using nitrogen
absorption and desorption measurements on a Micrometrics ASAP
2020N, with pretreatment of the samples at temperatures and
41
The last two are more freely bound in the crystalline structure.
MnCl2 and NiCl2 coordinate ammonia to form Mn(NH3)6Cl2
and Ni(NH3)6Cl2, respectively. Both of these have higher
molar masses than Mg(NH3)6Cl2, but as their crystal densities
4
2
are also higher, the volumetric hydrogen contents are
essentially the same as in Mg(NH)6Cl2. The temperatures at
3
pressures chosen to give the desired levels of NH desorption.
(
42) Gmelin Data: 2000-2006, Gesellschaft Deutscher Chemiker licensed
(
(
(
(
(
(
35) Ephraim, F. Chem. Ber. 1912, 45, 1322–1331.
to MDL Information Systems GmbH; 1988-1999, Gmelin Institut
f u¨ r Anorganische Chemie und Grenzgebiete der Max-Planck-Gesell-
schaft zur Foerderung der Wissenschaften.
36) Biltz, W.; H u¨ ttig, G. F. Z. Anorg. Allg. Chem. 1919, 109, 88–110.
37) Biltz, W. Z. Anorg. Allg. Chem. 1923, 130, 93–139.
38) Biltz, W.; Hansen, W. Z. Anorg. Allg. Chem. 1923, 127, 1–33.
39) Patil, K. C.; Secco, E. A. Can. J. Chem. 1972, 50, 567–573.
40) Christensen, C. H.; Sørensen, R. Z.; Johannessen, T.; Quaade, U.;
Honkala, K.; Elmøe, T. D.; Køhler, R.; Nørskov, J. K. J. Mater. Chem.
(43) Lepinasse, E.; Spinner, B. ReV. Int. Froid. 1994, 17, 309–321.
(44) Hummelshøj, J. S.; Sørensen, R. Z.; Kustova, M. Y.; Johannesen, T.;
Nørskov, J. K.; Christensen, C. H. J. Am. Chem. Soc. 2006, 128, 16–
17.
2
005, 15, 4106–4108.
(45) Jacobsen, H. S.; Hansen, H. A.; Andreasen, J. W.; Shi, Q.; Andreasen,
A.; Feidenhans’l, R.; Nielsen, M. M.; Ståhl, K.; Vegge, T. Chem. Phys.
Lett. 2007, 441, 255–260.
(
41) Westman, S.; Werner, P.-E.; Schuler, T.; Raldow, W. Acta Chem.
Scand. 1981, 35, 467–472.
8
662 J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008

Hydrogen Storage in Compact Metal Ammine Salts
A R T I C L E S
Table 4. Tablet Densities
3
F
tablet, g/cm
F
crystal, %
volumetric H, kg H/L
3
3
Mg(NH3)6Cl2
Ca(NH3)8Cl2
Mn(NH3)6Cl2
Ni(NH3)6Cl2
1.19
1.18
1.34
1.41
95
99
95
95
0.11
0.12
0.11
0.11
Desorption characteristics were determined by temperature-
programmed desorption (TPD). Samples of 0.5 g were transferred
to a closed test tube under an NH
a desired temperature ramp. NH
3
atmosphere and heated following
was released into a carrier stream
3
of Ar through a T-joint with a thin connection tube to maintain the
NH atmosphere over the sample. This procedure gave an NH
pressure slightly above atmospheric pressure over the ammine
sample. For TPDs of Mg(NH Cl , a sample obtained from
3
3
3
)
6
2
Amminex A/S was used as received. The ammonia content in the
carrier stream was determined using a Fischer-Rosemount NGA
2
3
000 equipped with an MLT analyzer calibrated to NH concentra-
tions from 0.03 to 30%. Desorption rates were calculated from the
ammonia content in the carrier stream and the flow of Ar, which
was kept constant at 213 mL of N/min using a calibrated Brooks
5
850 TR mass flow controller. For all four metal ammine salts,
TPDs were obtained with heating rates of both 1 and 5 K/min.
Results and Discussion
The behavior of the four different metal ammine chlorides
during TPD is shown in Figure 3. In accordance with literature
data, the trend in temperature for the first desorption peak is
Ca < Mn < Mg < Ni. During TPD, Ca(NH3)nCl2 has stable
compositions with n ) 8, 4, 2, and 1, and essentially all
ammonia desorbs in the temperature range of 300-550 K.
Mn(NH3)nCl2 has stable compositions with n ) 6, 2, and 1 and
two compositions with n < 1. Stable structures with less than
one ammonia molecule per metal atom are well known for a
range of transition metal ammine salts and are therefore not
unlikely, even though they have not been previously reported
for MnCl2. For this compound, essentially all ammonia desorbs
in the temperature range of 350-675 K. TPDs of Mg(NH3)nCl2
were reported previously, and there are stable compositions with
n ) 6, 2, and 1. Here, essentially all ammonia desorbs in the
temperature range of 410-700 K. During TPD, Ni(NH3)nCl2
shows stable compositions with n ) 6, 2, and 1, and all ammonia
desorbs in the temperature range of 440-700 K. Independently
of the temperature ramp, all the TPD peaks start within a few
degrees of the temperature for 1 bar equilibrium pressure
43
calculated from literature data. This and the nearly exponential
rise in desorption rate with temperature indicate that desorption
is equilibrium-limited, even at high desorption rates. During fast
desorption, the endothermic reaction causes the sample to cool,
which is evident as a small deviation from the predefined
temperature ramp when the desorption peaks (data not shown).
This cooling will, in effect, decrease the equilibrium vapor
pressure and thereby alter the desorption rate. Hence, for large
samples, heat transport to the reaction zone seems to be the
main limitation on the desorption rate for all four metal
ammine salts studied. This is also supported experimentally, as
the shape of TPD peaks depends on the sample size (data not
shown). When the sample is large, the peak is broadened and
the slopes (both increasing and decreasing) are less steep. This
corresponds to a larger temperature distribution inside the
sample. In an earlier work, computational modeling showed heat
transport to be the main limiting factor for desorption from
Figure 3. TPDs of (a) Ca(NH3)8Cl2, (b) Mn(NH3)6Cl2, (c) Mg(NH3)6Cl2,
and (d) Ni(NH3)6Cl2, with a temperature ramp of 5 K/min (red) and 1 K/min
(
blue). The solid lines show the desorption rate, and the dashed lines show
the total amount of desorbed ammonia.
diffusion kinetics or large activation energies, but only by
thermodynamic equilibrium, and hence by heat transport to the
reaction zone.
The desorption of ammonia from Ni(NH3)6Cl2 was also
examined by thermogravimetric analysis coupled with a mass
spectrometer. This confirmed that the sample was maintained
3
3
Mg(NH3)6Cl2. The present results indicate that it is a general
feature for metal ammine salts that desorption is not limited by
J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008 8663

A R T I C L E S
Sørensen et al.
Figure 4. Photos of (a) Mn(NH3)6Cl2, (b) Ca(NH3)8Cl2, and (c)
Ni(NH3)6Cl2 as powder and tablets.
Figure 6. Pore structures developed after different levels of ammonia
desorption from a Ni(NH3)6Cl2 tablet.
Table 5. Pore Characteristics for Mn(NH3)6Cl2
2
3
3
NH desorbed, % BET area, m /g average pore size, nm total pore volume, cm /g
0
3
5
7
8
.0
0.07
9.7
9.5
9
89.6
84.0
91.7
43.3
10.9
15.6
20.3
35.7
0.308
0.421
0.560
0.507
2.7
Table 6. Pore Characteristics for Ni(NH3)6Cl2
2
desorbed, % BET area, m /g average pore size, nm total pore volume, cm /g
3
NH
3
0
6
9
.0
8.2
8.4
12.0
53.9
120.5
6.8
25.5
21.1
0.051
0.432
0.687
Figure 5. Pore structures developed after different levels of ammonia
desorption from a Mn(NH3)6Cl2 tablet.
at complete ammonia saturation during handling and that the
only gas which desorbed from the sample was ammonia.
The densities of the metal ammine chloride tablets are
reported and compared to the crystal densities in Table 4. In
all cases, the metal ammine halides exhibit tablet densities of
at least 95% of their crystal densities. Photos of powders and
tablets are given in Figure 4.
different happens. The pores of 20-30 nm diameter disappear
without the number of pores in the size range 30-50 nm
increasing. For Ni(NH3)6Cl2, the data are not as detailed, but it
is evident from Figure 6 that, from a relatively small pore
volume with pore sizes distributed from 3 to 20 nm in the initial
tablet, a larger volume of pores in the size range 10-100 nm
develops as NH3 desorbs. As a consequence of this development
in pore structure, the average pore sizes and total pore volumes
in both salts generally increase during the desorption. Table 5
lists the Brunauer-Emmett-Teller (BET) areas measured at
various stages during the desorption of NH3 from Mn(NH3)6Cl2.
They are almost constant until the last measurement, where a
decrease is seen. For Ni(NH3)6Cl2, however, the BET area
continues to increase during the desorption process (see Table
For desorption of ammonia to be facile from dense bodies
of metal ammine complexes such as the present tablets, it is
important that a pore structure develops through which ammonia
can leave the tablet without a long diffusion path in the solid
state. The development of nanopores during ammonia desorption
has previously been demonstrated experimentally for Mg-
4
4,45
(
NH3)6Cl2.
Here, pore size distributions for both Mn-
(
NH3)6Cl2 and Ni(NH3)6Cl2 were measured after desorption of
6). As the increase in surface area makes more surface available
part of the ammonia from dense tablets of these materials.
Before the measurements, the samples must be evacuated at
ambient temperature or higher. Therefore, some NH3 necessarily
desorbs before the first measurement. For Ni(NH3)6Cl2, this loss
is negligible since the vapor pressure at room temperature is
very low. Hence, the porosity in the initial tablet could be
characterized carefully. For Mn(NH3)6Cl2, the initial porosity
can be calculated from the tablet density in Table 4 and the
for desorption of ammonia, this facilitates desorption of am-
monia from the interior of dense bodies of the material.
Modeling of Ammonia Desorption. A series of periodic density
functional theory (DFT) calculations, using the Dacapo planewave
46–48
pseudopotential implementation,
were performed to determine
this method’s ability to accurately calculate ammonia desorption
enthalpies for metal ammine salts, specifically Mg(NH3)6Cl2,
Ca(NH3)8Cl2, Mn(NH3)6Cl2, and Ni(NH3)6Cl2. The calcula-
tions were also used to provide additional understanding of
3
crystal density in Table 2 to be 0.07 cm /g. In Figure 5, it is
seen that, in the first part of desorption of NH3 from
Mn(NH3)6Cl2, pores of 2-3 nm are formed along with some
pores around 30-50 nm. As more and more NH3 is desorbed,
the smaller pores disappear while the number of larger pores
increases. From 79.0% desorbed to 82.7% desorbed, something
(
(
(
46) Kresse, J. Comput. Mater. Sci. 1996, 6, 15–50.
47) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892–7895.
48) Hammer, B.; Hansen, L. B.; Norskov, J. K. Phys. ReV. B 1999, 59,
7413–7421.
8
664 J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008

Hydrogen Storage in Compact Metal Ammine Salts
A R T I C L E S
Figure 7. Optimized structures found using DFT for (a) MgCl2, (b) Mg(NH3)Cl2, (c) Mg(NH3)2Cl2, and (d) Mg(NH3)6Cl2, viewed along the chains that
continue infinitely in and out of the picture. Four formula units are shown from the side, corresponding to the highlighted fragments in the end view.
5
1
the mechanistic details underlying the fast ab/desorption
processes observed for these materials as described above,
and finally also to predict potential electronic trends, which
can be utilized in the design of novel metal ammines with
optimized properties in hydrogen/ammonia storage.
Calculation of Desorption Enthalpies. We determine the
desorption enthalpy of going from one phase to another as the
difference in total energy between the most stable structures found
for the two phases plus the gas-phase energy of the released
ammonia molecules. The energy of an ammonia molecule in the
gas phase is calculated by placing one molecule in a vacuum cube
of side length 10 Å; zero-point energies are not included.
mental structure found for Ni(NH3)Cl2, which is believed to
be isostructural with the Mg(NH3)Cl2 equivalent, the double
octahedral chains have two orientations in the I2/m space group.
This structure was reproduced in our calculations; however, a
structure very similar to this one with respect to coordination
of the H atoms, but with only one orientation of the chains,
was found to have the same energy. For practical computational
reasons, the simple structure as shown in Figure 7b was chosen
for further calculations.
For n ) 0, the octahedra of chlorine atoms with central
magnesium atoms share half of their edges, resulting in layers
j
of MgCl2 of the CdCl2 type. The space group is R3m, with a )
3 n 2
Structures of Mg(NH ) Cl . To determine the most stable
structures for the different ammine phases, a rigorous search
3
1
.78 Å and c ) 18.52 Å, compared to a ) 3.64 Å and c )
52 52
7.67 Å in the experiment. As found in experiments, the
was first performed on the Mg(NH3)nCl2 (n ) 6, 2, 1, 0) systems
octahedra that are occupied by Mg are flattened, and the empty
octahedra between the layers are elongated. The ratio of the
short and the long edges of the occupied octahedra are r ) 0.92,
close to the value of r ) 0.934 from the experiment. In general,
the lattice is a little expanded in the calculations due to the
4
9–52
based on the experimentally reported structures.
In all the
structures, a central magnesium atom is coordinated octahedrally
to six ligands, i.e., either chlorine atoms for n ) 0, ammonia
molecules for n ) 6, or a combination of the two for n ) 2 or
52
1
, as shown in Figure 7.
For n ) 6, the structure is of the K2PtCl6 type and is
48
RPBE functional used. The RPBE functional is known to
4
9
overestimate lattice constants slightly. For a weakly bonded
system like the one treated here, we expect to see this problem
even more clearlysthis is a general feature of GGA-type
exchange correlation functionals. We will show in the following
that this does not affect the ability of the RPBE functional to
describe trends in stability of the ammines.
traditionally described as octahedral Mg(NH3)6 complexes
contained in a cubic lattice of chlorine atoms, with Mg body-
centered and NH3 face-centered in every second cube of chlorine
atoms, but the structure can also be described as chains of
Mg(NH3)6Cl2, as seen in Figure 7d, running along the face
diagonals of the cubes. In the optimized DFT structure, the
experimentally observed K2PtCl6 structure is slightly distorted.
For n ) 2, each chlorine atom is shared by two neighboring
magnesium atoms in edge-sharing octahedral chains. The space
group is Cmmm, with cell parameters a ) 8.73 Å, b ) 8.82 Å,
and c ) 4.17 Å somewhat larger than the experimental ones of
The interaction between individual Mg(NH3)nCl2 chains in the
respective structures can be examined quantitatively by comparing
the total energy of a Mg(NH3)nCl2 structure with the total energy
of a Mg(NH3)nCl2 chain in vacuum. This shows that the energy
per Mg atom needed to move a chain of Mg(NH3)nCl2 from the
bulk to vacuum is 26, 48, and 76 kJ/mol for n ) 1, 2, and 6,
respectively. For n ) 0, the corresponding value is 15 kJ/mol,
calculated as the binding between fragments (highlighted in Figure
5
0
a )8.18 Å, b ) 8.21 Å, and c ) 3.76 Å.
For n ) 1, a chlorine atom is shared by three magnesium
atoms in edge-sharing double octahedral chains. In the experi-
7
a) of the MgCl2 layers. The layers of MgCl2 are actually slightly
repulsive in this description.
(
(
49) Olovsson, I. Acta Crystallogr. 1965, 18, 889–893.
50) Leineweber, A.; Friedriszik, M. W.; Jacobs, H. J. Solid State Chem.
To understand the nature of the interaction between the chains
and layers of Mg(NH3)nCl2, we use density difference plots,
where the ground-state electron density of a given structure is
subtracted from that of the individual atomic species comprising
1
999, 147, 229–234.
51) Leineweber, A.; Jacobs, H.; Ehrenberg, H. Z. Anorg. Allg. Chem. 2000,
26, 2146–2152.
52) Partin, M. J. Solid State Chem. 1991, 95, 176–183.
(
(
6
J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008 8665

A R T I C L E S
Sørensen et al.
Figure 8. Density difference plots of (a) a NH3 molecule, (b) MgCl2, and (c) Mg and Cl in Mg(NH3)2Cl2 (yellow, Mg; green, Cl; blue, N; and white, H).
Blue areas correspond to an excess of electrons or negative charge, and red indicates depletion of electrons or positive charge. In (b) and (c), four formula
units are shown from the side, corresponding to the highlighted fragments in the end view.
4
9,53,54
the structure treated alone and in the same positions. This
effectively shows how the electron density is redistributed, and
by plotting isosurfaces of density increments and decrements,
negatively and positively charged areas become visible.
Some density difference plots are shown in Figure 8, where
one can see (i) the lone pair of ammonia and the positively
charged hydrogen atoms (Figure 8a); (ii) how the electrons
concentrate around the chlorine atoms in a MgCl2 layer (Figure
for nickel with n ) 2 as well.
The Ca(NH3)nCl2 structures,
on the other hand, are different and, moreover, go through n ) 8,
4, 2, 1, and 0. However, since the experimental enthalpies for the
8f4 and 4f2 transitions are very close in energy (41 and 42 kJ/
mol, respectively; see Table 3), we rely on the 6f2 reaction
enthalpy of the model to provide a reasonable approximation for
both of these; here, an enthalpy of 41 kJ/mol was chosen for
comparison with the 6f2 reaction enthalpy of the model.
Results I: Desorption Enthalpies from a Stable Structure
Comparison. The calculated desorption enthalpies are compared
with experimental values (see Figure 9), and it is seen that the
DFT model is able to quite accurately describe the trends in
desorption enthalpies of the metal ammine salts studied here.
A straight line is obtained in Figure 9, but the slope is less
than 1; i.e., the calculated enthalpies of desorption are too low
by a certain factor. The inclusion of zero-point energies or the
use of other exchange-correlation functionals only amounts to
a shift of the line and cannot account for the slope being less
than 1. Although the absolute numbers are not accurate, the
trend is very clear, and the precision of Figure 9 is usually
8
b), with the negatively charged areas protruding the layer,
rendering them slightly repulsive; and (iii) the charge redistribu-
tion caused by Mg and Cl alone in the Mg(NH3)2Cl2 structure
(
Figure 8c), showing how the ammonia molecules fit nicely. In
general, the lone pair of ammonia connects to the positively
charged areas surrounding the magnesium atoms, and the
positively charged hydrogen atoms connect to the negatively
charged areas near the chlorine atoms.
Thus, the chains for n ) 6, 2, and 1 are interconnected by
the electrostatic attraction between the positively charged H
atoms of an ammonia molecule and the negatively charged Cl
atoms (Figure 8a,c). For n ) 0, the layers are slightly repulsive,
and the real structure is therefore mainly held together by van
der Waals forces, which are not accurately described in these
DFT calculations.
48
considered more than acceptable in a DFT study like this.
Fast Ab/Desorption Processes. To understand the fast ab/
desorption of ammonia in the metal ammine salts, we considered
the structures again, but this time from the perspective of how
ammonia can get in and out of these materials. We restrict the
discussion to involve only the isostructural Mg, Mn, and Ni
salts and focus on the magnesium salt as a representative.
As mentioned previously, clean MgCl2 has a layered structure,
while the structures of Mg(NH3)nCl2 for n ) 1, 2, and 6 consist
Structures of the Ca, Mn, and Ni Salts. The structures found
for the Mg salt were also used for the three other salts (Ca,
Mn, Ni) by simply exchanging the metal atoms and letting the
structures relax while scaling the unit cell size linearly until a
minimum in total energy was reached.
For Mn(NH3)nCl2 and Ni(NH3)nCl2, this approximation is
expected to be rather precise, since these salts are known to be
isostructural with the magnesium analogue for n ) 6 and 0 and
(
53) Ferrari, A.; Bigliard, G.; Braibant, A. Acta Crystallogr. 1963, 16, 846–
47.
(54) Leineweber, H. J. Solid State Chem. 2000, 152, 381–387.
8
8
666 J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008

Hydrogen Storage in Compact Metal Ammine Salts
A R T I C L E S
Three principles underlie the proposed mechanism for
understanding the dynamics of ammonia absorption and de-
sorption: (a) it is energetically preferred for Mg to retain a six-
fold coordination to the ligands (either Cl or NH3), (b) chains
are released from the surface to facilitate desorption or absorp-
tion, and (c) reactions involving a minimal number of chains
are expected to be faster.
When a chain is released from the hexammine (Figure 10a), it
releases four of its six NH3 molecules per Mg and transforms into
a diammine chain, thus maintaining the energetically preferred six-
fold coordination of the Mg atoms (four Cl’s substitute the released
NH3). If the hexammine chain were to release more than four NH3
molecules directly, it would require the expectedly slower com-
bination of two or more chains to keep the Mg atoms fully
coordinated. The formed diammine chains will subsequently
arrange into the bulk structure of Mg(NH3)2Cl2.
As the temperature is further increased, chains break off from
the diammine structure (Figure 10b). For this desorption
reaction, a minimum of two chains must combine to release
Figure 9. Calculated versus experimental desorption enthalpies for the
different desorption steps, 6f2, 2f1, and 1f0, of Mg(NH3)6Cl2 (triangles),
Ca(NH3)8Cl2 (squares), Mn(NH3)6Cl2 (circles), and Ni(NH3)6Cl2 (diamonds)
and in a comparison of stable structures. The calcium salt is approximated
by model hexaammine salt Ca(NH3)6Cl2, which follows the same desorption
route as the three other salts. In this approximation, the first desorption
step is representative of the first two desorption steps of the real octaammine
salt, Ca(NH3)8Cl2.
(
half their) NH3 molecules, in order to maintain the six-fold
Mg coordination. The resulting monoammine chains later
rearrange into the bulk structure of Mg(NH3)Cl2.
Similarly, for the monoammine structure (Figure 10c), chains
combine in pairs to give Mg(NH3)0.5Cl2 chains. This combina-
tion of chains to release NH3 and form chains of lower ammonia
content can, in principle, continue until all the ammonia is
desorbed and the layered structure of MgCl2 is formed. The
intermediate n < 1 chains are, in general, not expected not to
be stable, and therefore not detected in the experiments, except
for Mn(NH3)nCl2, which finds a stable arrangement of the
Mn(NH3)0.5Cl2 chains (see Figure 3b).
of chains (see Figure 7). Going from one phase to another, the
chains/layers of the initial structure are either cleaved to absorb
ammonia or recombined to desorb ammonia. For the 6f2/2f6
reactions, the chains are merely stretched/compressed (see Figure
7
). In all cases, the resulting chains or layers are rearranged to
produce the final structure.
The details of this cleavage and recombination and the
detailed mechanism for transport of ammonia into and out of
the materials are unknown, but both desorption and absorption
For absorption the picture is reversed, and the chains/layers
are cleaved instead of combined (Figure 10d-f).
4
0
are known from experiments to be facile. We shall first
consider the situation where bulk diffusion of ammonia is the
dominant transport process.
Results II: A New Scheme for Calculating Desorption
Enthalpies. We have presented a model for the detailed mechanism
behind desorption and absorption processes in metal amine salts,
and we now return to the modeling of desorption enthalpies. In
the alternative mechanism, the system does not transform directly
from one stable structure to the next during desorption. Instead, it
goes through an intermediate state where chains of the final phase
are free and uncoordinated. These free chains could combine
sufficiently fast into the final structure to make it reasonable to
compare the energies of the stable structures, as done in Figure 8.
However, it is also interesting to model the desorption enthalpies
under the assumption that the free chains are sufficiently long-
lived to be detected experimentally. In that case, the relevant energy
to calculate is the difference between the initial ammine in its stable
bulk form and the final ammine as a chain in vacuum. For the
1f0 transition, the relevant free chains are the Mg(NH3)0.5Cl2
chains, as shown in Figure 10c.
This can easily be done in the DFT calculations, and the
results of this scheme applied to all four salts are again compared
with apparent experimental desorption enthalpies and shown
in Figure 11.
Although the straight line in Figure 11 has a slope of 1, it
does not confirm the mechanism presented in Figure 10, nor
does it prove that desorption in the TPD experiments is not
equilibrated. The slope of less than 1 in Figure 9 could very
well be due to limitations in the calculations. However, it does
provide a valid way to predict desorption enthalpies using DFT,
which appears to be more precise than comparing the energies
of the stable structures. More importantly, the mechanism which,
as mentioned, is also compatible with the equilibrium situation
Bulk Diffusion in Metal Ammines. The transport of ammonia
out of the systems during desorption could be dominated by
bulk diffusion of ammonia. Indeed, from Figure 7, it seems
likely that, if ammonia desorbs by diffusing along the chains
of Mg(NH3)nCl2, the system can transform from one bulk phase
directly to another in a continuous way. However, the barriers
for diffusion need to be sufficiently small that no competing
mechanisms become dominant.
The barriers for diffusion along the chains of Mg(NH3)nCl2
have been calculated to be 58, 135, and 121 kJ/mol for n ) 6,
2
, and 1, respectively; this is clearly too high for n ) 2 and 1
when compared with the experimental desorption enthalpy. If
bulk diffusion of ammonia were indeed rate-limiting, one would
expect the last two peaks in a TPD experiment to be broader
than the first peak, which is not the case, as seen in Figure 3.
Although the bulk diffusion picture cannot be ruled out, at
least for the n ) 6 case, an alternative mechanism would be
desirable. In the following, such a mechanism, which explains
the fast kinetics for both the absorption and desorption processes,
is proposed.
Restructuring at the Surface. We now consider the rather
weak binding between the individual chains in Mg(NH3)nCl2
(
15, 26, 48, and 76 kJ/mol for n ) 0, 1, 2, and 6, respectively),
which allows for an alternative desorption process in which the
individual chains are released from the surface to facilitate
desorption or absorption of ammonia from or to the chains (see
Figure 10), as investigated in the following.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008 8667

A R T I C L E S
Sørensen et al.
Figure 10. Proposed mechanism for desorption and absorption of ammonia in Mg(NH3)nCl2, Mn(NH3)nCl2, and Ni(NH3)nCl2.
ࠗ
w
Animations illustrating front and side views of the desorption of ammonia from modeling studies are available.
namic equilibrium and heat transfer, as previously shown for
Mg(NH3)6Cl2. It is shown that the metal ammine complexes
can be compacted to tablets with densities very close to the
crystal densities and still maintain the same desorption proper-
ties. This is due to the development of nanopores in the tablets
during desorption of ammonia. The nanopores facilitate diffusion
of ammonia through the dense tablets.
DFT calculations are shown to reproduce accurately the trends
in desorption enthalpies for all the systems studied. On the basis
of extensive DFT calculations, we propose a mechanism for
absorption and desorption of ammonia from metal ammine salts
in which chains of the ammines are released from the surface
of the crystal to explain the fast ab/desorption processes
observed experimentally.
Acknowledgment. The authors thank Marina Y. Kustova and
Mo Hongling Sønnichsen for experimental assistance and Amminex
3 6 2
A/S for providing Mg(NH ) Cl . The Danish National Research
Figure 11. Calculated versus apparent experimental desorption enthalpies
for the different desorption steps of Mg(NH3)6Cl2 (triangles), Ca(NH3)8Cl2
Foundation sponsors the Center for Sustainable and Green Chem-
istry. Lundbeck Foundation sponsors the Center for Atomic-scale
Materials Design. The project is also supported by The Danish
Council for Strategic Research (project no. 2104-05-0016) and the
European Commission DG Research (contracts SES6-2006-51827/
NESSHy and MRTN-CT-2006-032474/HYDROGEN).
(
squares), Mn(NH3)6Cl2 (circles), and Ni(NH3)6Cl2 (diamonds) for the
proposed mechanism in Figure 10.
explains why desorption and absorption can happen fast in all
the steps of the ab/desorption processes.
Conclusion
Supporting Information Available: Table of thermodynamic
properties of 90 metal ammine complexes. This information is
available free of charge via the Internet at http://pubs.acs.org.
It is demonstrated that metal ammine complexes can be
utilized as reversible, indirect hydrogen storage materials. The
desorption of ammonia from Ca(NH3)8Cl2, Mn(NH3)6Cl2, and
Ni(NH3)6Cl2 is shown to be controlled mainly by thermody-
JA076762C
8
668 J. AM. CHEM. SOC. 9 VOL. 130, NO. 27, 2008