Scripta Materialia 56 (2007) 817–822
www.actamat-journals.com
Viewpoint Paper
Metal–N–H systems for the hydrogen storage
Ping Chen,a,b,* Zhitao Xiong,a Guotao Wu,a Yongfeng Liu,a
Jianjiang Hua and Weifang Luoc
a
Department of Physics, Faculty of Science, National University of Singapore, Singapore 117542, Singapore
Department of Chemistry, Faculty of Science, National University of Singapore, Singapore 117543, Singapore
c
Sandia National Laboratories, Livermore, CA, USA
b
Received 31 July 2006; revised 26 September 2006; accepted 2 January 2007
Available online 14 February 2007
Abstract—The hydrogen storage in metal–N–H systems is reviewed. Exemplary systems including Li–N–H, Mg–N–H, Li–Mg–N–H
and Li–Al–N–H are highlighted. Analyses and discussions are focused on the thermodynamics and kinetics of the respective
systems.
Ó 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Amides; Imides; Nitrides; Hydrogen storage
1. Introduction
The demand for highly efficient solid-state hydrogen
storage materials for the coming hydrogen economy
has encouraged tremendous efforts worldwide in the research and development of novel storage systems [1–8].
Hydrogen can be stored in solid media either through
physical or chemical processes. For chemical storage,
the process can be either reversible or irreversible,
depending on the thermodynamic nature of the corresponding reactions.
A variety of promising storage systems are currently
being intensively investigated. Complex hydrides [2,3],
chemical hydrides [4,5], carbonaceous materials [6] and
nanocomposites [7] are among the systems being focused
on. Material with a high H content is attractive because
it allows more space for subsequent optimization.
Studies on the hydrogen storage in metal–N–H
systems were initiated when researchers noticed accidentally that the mixture of metallic lithium and carbon
nanotubes pretreated in a purified N2 atmosphere could
absorb a large amount of hydrogen at temperatures
above 150 °C [8]. The hydrogenated solid-state sample
was identified as containing LiNH2, LiH and unreacted
* Corresponding author. Address: Department of Physics, Faculty of
Science, National University of Singapore, Singapore 117542,
Singapore. Tel.: +65 65165100; fax: +65 67776126; e-mail:
[email protected]
carbon nanotubes. Further investigations revealed that
the N2 treated Li–C mixture was actually composed
of Li3N and carbon nanotubes. The hydrogen storage
in Li3N follows the B-1 reaction (see Table 1). The subsequent dehydrogenation is via the chemical reaction
between LiNH2 and LiH, indicating that there must
be a driving force for the reaction to be taken place.
N, like O, can bond with H and form –NH, –NH2
and NH3. H bonded with N normally exhibits a
positive oxidation state (Hd+) [9]. N can also bond with
metals and metalloids. When both H and metal bond
to N, imides (M(NH)n) and amides (M(NH2)n) may
form. Amides are analogous to NH3 but with considerable changes in the structure and electronic properties
[10]. The H content of amides is relatively high, especially when M is Li and Mg. However, the direct
decomposition of amides will not generate hydrogen
but rather, NH3. On the contrary, the H in hydrides,
especially ionic hydrides, is negatively charged (Hd)
[11]. The abnormal high potential of the union of Hd+
and Hd to H2 may induce amides and hydrides to react with each other. The significant implication of this
interaction lies in that similar reactions between a variety of amides and hydrides combination may occur
which will lead to the development of new chemical
processes/materials for hydrogen storage. In this paper,
investigations on hydrogen storage in metal–N–H systems were reviewed based on the works published since
2002. The chemical processes involved are summarized
in Table 1.
1359-6462/$ - see front matter Ó 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.scriptamat.2007.01.001
818
P. Chen et al. / Scripta Materialia 56 (2007) 817–822
Table 1. Summary of the chemical reactions between various amides and hydrides
Systems
Reactions
Refs.
Binary
B-1
B-2
B-3
B-4
B-5
B-6
B-7
LiNH2 + 2LiH M Li3N + 2H2
LiNH2 + LiH M Li2NH + H2
Mg(NH2)2 + MgH2 ! 2MgNH + H2
Mg(NH2)2 + 2MgH2 ! Mg3N2 + 4H2
2CaNH + CaH2 M Ca3N2 + 2H2
CaNH + CaH2 M Ca2NH + H2
Ca(NH2)2 + CaH2 ! 2CaNH + 2H2
[8,12,13]
[8,12,13]
[17]
[18]
[21]
[8,21]
[22]
Ternary
T-1
T-2
T-3
T-4
T-5
T-6
T-7
T-8
2LiNH2 + CaH2 M Li2 Ca(NH)2 + 2H2
Mg(NH2)2 + 2LiH M Li2Mg(NH)2 + 2H2
3Mg(NH2)2 + 8LiH M 4Li2NH + Mg3N2 + 8H2
Mg(NH2)2 + CaH2 ! MgCa(NH)2 + 2H2
2LiNH2 + LiBH4 ! Li3BN2 + 4H2
2LiNH2 + LiAlH4 ! Li3AlN2 + 4H2
LiNH2 + 2LiH + AlN M Li3AlN2 + 2H2
4LiNH2 + 2Li3AlH6 ! Li3AlN2 + Al + 2Li2NH + 3LiH + 15/2H2
[23]
[23,24]
[26]
[31]
[32,33]
[36]
[36]
[37]
Multinary
M-1
M-2
NaNH2 + LiAlH4 ! NaH + LiAlNH + 2H2
3Mg(NH2)2 + 3LiAlH4 ! Mg3N2 + Li3AlN2 + 2AlN + 12H2
[39]
[40]
2. Binary metal–N–H systems
10
Abs.
Wt % H2
8
6
4
Des.
2
0
50
100 150 200 250 300 350 400 450
Temperature (ºC)
Figure 1. Weight variations during the hydrogen absorption and
desorption processes of an Li3N sample.
+
Intensity (a.u.)
A few binary systems, such as Li–N–H [8,12–16],
Mg–N–H [17–20] and Ca–N–H [21,22], are listed in
Table 1. Hydrogen desorption from a LiNH2 and LiH
mixture (molar ratio = 1:2) is a two-step process [8,12].
As shown in Figure 1, the first desorption step occurs
at temperatures above 150 °C. Approximately 6 wt.%
of hydrogen is desorbed at 200 °C when a high vacuum
was applied. The second desorption step occurs at temperatures above 320 °C. It takes a long time for complete desorption at temperatures around 420 °C. The
subsequent hydrogenation takes place at temperatures
above 150 °C. X-ray diffraction measurements have revealed that the half-dehydrogenated sample contains
Li2NH and LiH, whereas the fully dehydrogenated sample is red–brown Li3N (see Fig. 2) [12]. Thermodynamic
analyses have shown that hydrogen desorption from
LiNH2–2LiH and LiNH2–LiH is a highly endothermic
process, with the heat of desorption of 80 and
66 kJ mol1 H2 [8], respectively. Therefore, the operation temperature at 1.0 bar equilibrium desorption pressure is above 250 °C, which is too high for its practical
application. It has also been reported that additives such
+
#
+
c
+
+
#
b
#
a
10
*
*
20
30
2θ
40
50
Figure 2. Structural changes during the dehydrogenation of an
LiNH2–LiH (1/2) sample: (a) pristine sample; (b) half-dehydrogenated
sample; (c) fully dehydrogenated sample. (s) LiNH2; (*) LiH; (#)
Li2NH and (+) Li3N.
as TiCl3 and Li2O can improve the kinetics of hydrogenation and dehydrogenation [13,14]. A few simulation
studies have been performed on LiNH2 and Li2NH
[15,16].
The Mg–N–H system, on the other hand, has relatively mild thermodynamics [17,18]. Hydrogen desorption from Mg(NH2)2 and MgH2 (at either a 1:1 or 1:2
molar ratio, Reactions B-3 and B-4) can occur under a
mechanochemical reaction condition, i.e. energetic ball
milling. As shown in Figure 3, hydrogen evolved from
the Mg(NH2)2–MgH2 (1:2) mixture shortly after the ball
milling [18]. The hydrogen desorption was accelerated
after the mixture was ball milled for 5 h. The hydrogen
release rate slowed down after about 20 h of milling.
Mg(NH2)2 and MgH2 are stable in the milling jar if ball
milled alone, indicating that the hydrogen generation is
due to the chemical reaction between these two chemicals. In total, 4 equiv. H2 per [Mg(NH2)2 + 2MgH2]
were released at the end of ball milling [18]. As expected,
the solid residue is Mg3N2. Thermodynamic analysis
revealed that the average heat of desorption is
3.5 kJ mol1 H2 [18] for Reaction B-4, indicating that
P. Chen et al. / Scripta Materialia 56 (2007) 817–822
200
0
180
Release
160
1
140
2
120
Wt % H2
Pressure increase (psi)
819
100
80
3
4
60
soak
40
5
20
0
6
10
20
30
40
50
Milling time (hour)
60
70
0
80
Figure 3. Time dependence of the hydrogen pressure increase in the
milling jar.
Mg3N2 can hardly be converted to Mg(NH2)2 and
MgH2 under normal hydrogenation conditions. It took
72 h to complete the hydrogen desorption even if the
dehydrogenation was allowed thermodynamically.
Therefore, the overall kinetics for hydrogen desorption
should be rather slow, which might be the reason why
investigations by other researchers have failed to detect
the reaction between Mg(NH2)2 and MgH2 [19].
3. Ternary systems
300
1000
Abs
100
Des
10
1
0.1
0
Even though they have a high H content, the binary
systems have thermodynamic problems for practical
applications. Compositional alteration is a viable way
to change the thermodynamic parameters of a chemical
process. As discussed in Section 1, the strong potential
for the union of Hd+ in amides and Hd in hydrides will
drive varieties of amides and hydrides to react with each
other and liberate hydrogen molecules. A series of
ternary systems have been developed since 2004. Li–
Mg–N–H, Li–B–N–H and Li–Al–N–H are among the
most investigated systems [23–37].
Two processes can lead to the formation of
Li2Mg(NH)2. One is via the chemical reaction of
Mg(NH2)2 and 2LiH [23], the other is via 2LiNH2 and
MgH2 [24]. The hydrogenation of Li2Mg(NH)2 yields
Mg(NH2)2 + 2LiH (Reaction T-2) [23]. This is because
Mg(NH2)2 + 2LiH is thermodynamically more favored
than 2LiNH2 + MgH2 [25]. Volumetric release and soak
measurements reveal that more than 5 wt.% of hydrogen
can be reversibly stored in the Li2Mg(NH)2 sample
(Fig. 4) in the temperature range of 100–300 °C [28].
The thermodynamic properties are improved considerably compared with those of the binary Li–N–H and
Mg–N–H systems. Pressure–composition–temperature
(PCT) measurements show that the dehydrogenation
of Mg(NH2)2 + 2LiH exhibits a pressure plateau region
and a slope region (Fig. 5) [23,24,28]. The heat of
desorption hydrogen within the pressure plateau is
38.9 kJ mol1 H2 [28]. The overall reaction heat is
44 kJ mol1 H2 [28]. The results of the van’t Hoff plot
indicate that the temperature to desorb hydrogen at
1.0 bar equilibrium pressure is 90 °C [28], which is
100
200
Temperature (˚C)
Figure 4. Volumetric soak and release measurements of a Li2Mg(NH)2
sample.
Pressure (psi)
0
1
2
3
H / Li2MgN2H2
4
Figure 5. PCT curve of Li2Mg(NH)2 at 180 °C.
close to the operation temperature of proton exchange
membrane fuel cell (PEM FC). However, the high kinetic barrier associated with the hydrogen desorption
puts a restriction onto the low-temperature applications.
The detailed investigations show that the activation energy for hydrogen desorption is higher than 88 kJ mol1
H2 [29]. Isotopic exchange and other designed experiments indicate that the kinetic barrier may come from
the interface reactions and mass transport through the
product layer, depending on the progression of the reaction [29]. Catalytic modification is important to bring
the system a step forward towards practical usage.
Altering the molar ratio of Mg(NH2)2 and LiH will lead
to different dehydrogenation products, such as nitride
(Reaction T-3) [26,27,30]. The H content can be higher,
but the temperature for the hydrogen desorption moves
to the higher end [30]. Continuous compositional optimizations are being conducted in this promising system
to obtain better thermodynamic and kinetic properties.
Four H atoms can be detached from the mixture of
Mg(NH2)2 and CaH2 (Reaction T-4) [31]. The average
heat of desorption was measured to be 28 kJ mol1 H2
[31], which enables it to be considered as a low-temperature system. However, the activation energies for both
dehydrogenation and hydrogenation are very high.
LiNH2–LiBH4 and LiNH2–LiAlH4 have recently
become the subjects of increasing research activities
P. Chen et al. / Scripta Materialia 56 (2007) 817–822
0
Release
1
2
Wt %
[32–38]. It was reported that more than 11 wt.% of
hydrogen can be desorbed from the mixture of
2LiNH2 + LiBH4 in the temperature range of 100–
350 °C [32,33]. Li3BN2 is the final product (Reaction
T-5). The overall reaction is an exothermic process, indicating that direct hydrogenation may not be feasible under normal conditions. Interestingly, the partial
hydrogenation of an Li–B–N–H sample was noticed,
which was probably due to an alternative reaction route
[34]. Investigations on the LiNH2–LiAlH4 system (with
molar ratios of 1:1 and 2:1) revealed that the transition
of [AlH4] to [AlH6]3 is fairly easy in the presence of
LiNH2 [35,36]. A large amount of hydrogen was desorbed from the LiNH2–LiAlH4 system under a ball milling condition. Nuclear magnetic resonance (NMR)
measurements of the samples collected at different intervals of ball milling showed that an Al–N bond was
established immediately after mixing these two chemicals (Fig. 6) [36], which implied the direct interaction between LiNH2 and [AlH4]. Such an interaction may
induce the dissociation of [AlH4]. It was noticed that
most of the hydrogen desorbed during the mechanical
ball milling was hard to be recharged back to the material, probably due to thermodynamic reasons. Interestingly, hydrogen can be reversibly stored by Li3AlN2,
which is the fully dehydrogenated state of the postmilled 2LiNH2–LiAlH4 sample (Reaction T-6). As
shown in Figure 7, 5.1 wt.% of hydrogen can be
absorbed and desorbed in a Li3AlN2 sample in the temperature range of 100–500 °C. The fully hydrogenated
sample contains AlN, LiH and LiNH2 (Reaction T-7)
[36].
The LiNH2–Li3AlH6 combination was also investigated [37,38]. Different results were observed by different
groups. Kojima et al. detected the formation of
Li3AlN2, Li2NH, Al and LiH after the dehydrogenation
of Li3AlH6–2LiNH2 (Reaction T-8). Rehydrogenation
of the sample cannot reproduce Li3AlH6 [37]. The Al–
N bonding was not observed by Lu et al. A fully reversible reaction between Li3AlH6 and LiNH2 (molar ration
1:3) was reported [38].
3
4
Soak
5
6
0
Intensity (a.u.)
200
300
400
In addition to the emerging ternary systems, a slight
compositional change within a given ternary system
may induce considerable variations in the reaction path.
Those interesting features enable a relatively broader
scope for material optimization and engineering.
4. Multinary systems
Materials containing more than two metal or metalloid elements are categorized as multinary systems.
Two systems, NaNH2–LiAlH4 [39] and Mg(NH2)2–
LiAlH4 [40], have been investigated so far. The chemical
process of a multi-component system is more complicated than those of binary and ternary systems. The
5
wt%
7x10
5
6x10
5
5x10
5
4x10
5
3x10
5
2x10
5
1x10
0
0
-1
-2
-3
-4
-5
signal of hydrogen in TPD
volumetric release
a
150
100
50
0
-50
-100
Chemical shift (ppm)
Figure 6. 27Al magic angle spinning NMR spectra of a LiNH2–LiAlH4
(2:1) sample collected at different stages of ball milling: (a) ball milled
for 40 min with one H atom detached from the starting chemicals and
(b) ball milled for 12 h with four H atoms detached.
NaNH2-LiAlH4 1-1
LiAlH4
NaNH2
endo exo
Heat Flow (a.u.)
b
Li3AlH6
500
Figure 7. Volumetric soak and release of a Li3AlN2 sample.
Al
LiAlH4
100
Temperature (˚C)
TPD Intensity
820
50
100
150 200 250 300
Temperature (°C)
350
400
Figure 8. Temperature programmed desorption volumetric release and
DSC measurements on the mixed NaNH2–LiAlH4 (1:1) sample. For
comparison, the decomposition of LiAlH4 and NaNH2 are illustrated
in DSC.
P. Chen et al. / Scripta Materialia 56 (2007) 817–822
reaction between NaNH2 and LiAlH4, for example, is a
multi-step reaction involving continuous phase changes
(Reaction M-1). The overall reaction is an exothermic
process, which quickly liberates 5 wt.% of hydrogen
at temperature around 110 °C (see Fig. 8). Ball milling
the two chemicals at ambient temperature also resulted
in desorption of the same amount of hydrogen within
a short period of time [39].
5. Pending issues
H content in metal–N–H systems is comparatively
high. Systems with storage capacity of 10 wt.% or above
are available. More than 10 attractive systems have been
developed just within 5 years of research. More promising systems are emerging. In the meanwhile, considerable knowledge of the chemistry of N-containing
compounds has been accumulated, which will greatly
facilitate the subsequent research and development.
The broad range of materials and large number of possibility and varieties of approaches in the material optimization and engineering enable metal–N–H systems
among the most promising candidates for hydrogen
storage.
There are a few challenges that metal–N–H systems
have to face. One of them is material stability. Most
metal–N–H systems are unstable when heated to high
temperatures or if they encounter moisture and oxygen.
NH3 is the major gaseous by-product of dehydrogenation, which may deactivate the PEM FC if the concentration is high [41]. Systems of stronger chemical
bonding between metal (or metalloid) and N that ensure
the retention of N should have a greater chance of solving this problem.
The other challenge that metal–N–H systems have to
overcome is the slow kinetics of the hydrogenation and
dehydrogenation, which is an intrinsic property of solidstate reactions. The kinetics problem could be more
prominent for a system of multiple phases and/or undergoing a stepwise reaction. Catalytic modification is
important. The effective catalytic modification depends
strongly on the identification of the origin of the kinetic
barrier(s); therefore, mechanistic understanding of the
corresponding reaction is essential. Numerous investigations into the reaction mechanism have been performed
by different research groups [13,29,42,43]. Although
gained from different viewpoints, the information collected considerably enhances the understanding of different aspects of a reacting system. In general, the
mass transport and interface reaction should be the
main sources of kinetic barriers. Catalytic modification
should focus on the improvement of the rate of mass
transport and the facilitation of the chemical bond
reconstruction along the phase boundaries [29].
References
[1] L. Schlapbach, A. Zuttel, Nature 414 (2001) 353.
[2] B. Bogdanovic, M. Schwickardi, J. Alloy Compd. 253
(1997) 1.
[3] C.M. Jensen, K.J. Gross, Appl. Phys. A 72 (2001) 213.
821
[4] A. Gutowska, L.Y. Li, Y.S. Shin, et al., Angew. Chem.
Int. Ed. 44 (2005) 3578.
[5] S.C. Amendola, S.L. Sharp-Goldman, M.S. Janjua, et al.,
Int. J. Hydrogen Energy 25 (2000) 969.
[6] M. Hirscher, M. Becher, M. Haluska, et al., J. Alloy
Compd. 330 (2002) 654.
[7] A.M. Seayad, D.M. Antonelli, Adv. Mater. 16 (2004) 765.
[8] P. Chen, Z.T. Xiong, J.Z. Luo, et al., Nature 420 (2002)
302;
P. Chen, et al., Oral & poster presentations at the MRS
Spring Meeting 2002, San Francisco, CA.
[9] A. Hinchliffe, J.C. Dobson, Theor. Chim. Acta 38 (1975)
17.
[10] E.U. Würthwein, K.D. Sen, J.A. Pople, R. Schleyer P
von, Inorg. Chem. 22 (1983) 496.
[11] S. Yamamura, S. Kasahara, M. Takata, Y. Sugawara,
M. Sakata, J. Phys. Chem. Solids 60 (1999) 1721.
[12] P. Chen, Z.T. Xiong, J.Z. Luo, et al., J. Phys. Chem. B
107 (2003) 10967.
[13] T. Ichikawa, S. Isobe, N. Hanada, et al., J. Alloy Compd.
365 (2004) 271.
[14] Y.H. Hu, E. Ruckenstein, Ind. Eng. Chem. Res. 43 (2004)
2464.
[15] K. Miwa, N. Ohba, S. Towata, et al., Phys. Rev. B 71
(2005) 195109.
[16] B. Magyari-Kope, V. Ozolins, C. Wolverton, Phys. Rev.
B 73 (2006) 220101.
[17] J.J. Hu, Z.T. Xiong, G.T. Wu, P. Chen, J. Power Sources
159 (2006) 120.
[18] J.J. Hu, G.T. Wu, Y.F. Liu, Z.T. Xiong, P. Chen, J.
Phys. Chem. B 110 (2006) 14688.
[19] Y. Nakamori, G. Kitahara, S. Orimo, J. Power Sources
138 (2004) 309.
[20] H.Y. Leng, T. Ichikawa, S. Isobe, S. Hino, N. Hanada,
H. Fujii, J. Alloy Compd. 404–406 (2005) 443.
[21] Z.T. Xiong, P. Chen, G.T. Wu, et al., J. Mater. Chem. 13
(2003) 1676.
[22] S. Hino, T. Ichikawa, H.Y. Leng, et al., J. Alloy Compd.
398 (2005) 62.
[23] Z.T. Xiong, G.T. Wu, J.J. Hu, et al., Adv. Mater. 16
(2004) 1522;
P. Chen, et al., Oral presentation at the MRS Fall Meeting
2003, Boston, MA.
[24] W.F. Luo, J. Alloy Compd. 381 (2004) 284.
[25] W. Luo, E. Ronnebro, J. Alloy Compd. 404 (2005) 392.
[26] H.Y. Leng, T. Ichikawa, S. Hino, et al., J. Phys. Chem. B
108 (2004) 8763.
[27] Y. Nakamori, S. Orimo, J. Alloy Compd. 370 (2004) 271.
[28] Z.T. Xiong, J.J. Hu, G.T. Wu, et al., J. Alloy Compd.
398 (2005) 235.
[29] P. Chen, Z.T. Xiong, G.T. Wu, et al., J. Phys. Chem. B
110 (2006) 14221.
[30] Z.T. Xiong, G.T. Wu, J.J. Hu, et al., J. Alloy Compd.
417 (2006) 190.
[31] Y.F. Liu, J.J. Hu, Z.T. Xiong, G.T. Wu, et al., J. Alloy
Compd. (2007) in press, doi:10.1016/j.jallcom.2006.05.124.
[32] F.E. Pinkerton, G.P. Meisner, M.S. Meyer, et al., J.
Phys. Chem. B 109 (2005) 6.
[33] M. Aoki, K. Miwa, T. Noritake, et al., J. Alloy Compd.
404 (2005) 427.
[34] F.E. Pinkerton, M.S. Meyer, G.P. Meisner, et al., J.
Phys. Chem. B 110 (2006) 7967.
[35] Z.T. Xiong, G.T. Wu, J.J. Hu, P. Chen, J. Power Sources
159 (2006) 167;
P. Chen, Oral presentation at APS March Meeting 2005,
Los Angeles, CA.
[36] Z.T. Xiong, G.T. Wu, J.J. Hu, Chen P, submitted.
[37] Y. Kojima, M. Matsumoto, Y. Kawai, et al., J. Phys.
Chem. B 110 (2006) 9632.
822
P. Chen et al. / Scripta Materialia 56 (2007) 817–822
[38] J. Lu, Z.Z. Fang, H.Y. Sohn, J. Phys. Chem. B 110 (2006)
14236.
[39] Z.T. Xiong, J.J. Hu, G.T. Wu, Y.F. Liu, P. Chen, Catal
Today, (2007) in press, doi:10.1016/j.cattod.2006.09.006.
[40] Y.F. Liu, Z.T. Xiong, G.T. Wu, J.J. Hu, P. Chen et al,
submitted.
[41] F.E. Pinkerton, J. Alloy Compd. 400 (2005) 76.
[42] H.Y. Leng, T. Ichikawa, S. Hino, et al., J. Phys. Chem. B
109 (2005) 10744.
[43] W.F. Luo, S. Sickafoose, J. Alloy Compd. 407 (2006)
274.