Well-defined molecular magnesium hydride clusters: Relationship between size and hydrogen-elimination temperature

Article abstract of DOI:10.1002/chem.201300684

A new tetranuclear magnesium hydride cluster, [{NN-(MgH)₂} ₂], which was based on a N-N-coupled bis-β-diketiminate ligand (NN^2-), was obtained from the reaction of [{NN-(MgnBu) ₂}₂] with PhSiH₃. Its crystal structure reveals an almost-tetrahedral arrangement of Mg atoms and two different sets of hydride ions, which give rise to a coupling in the NMR spectrum (J=8.5 Hz). To shed light on the relationship between the cluster size and H₂ release, the thermal decomposition of [{NN-(MgH)₂}₂] and two closely related systems that were based on similar ligands, that is, an octanuclear magnesium hydride cluster and a dimeric magnesium hydride species, have been investigated in detail. A lowering of the H₂-desorption temperature with decreasing cluster size is observed, in line with previously reported theoretical predictions on (MgH₂)_n model systems. Deuterium-labeling studies further demonstrate that the released H₂ solely originates from the oxidative coupling of two hydride ligands and not from other hydrogen sources, such as the β-diketiminate ligands. Analysis of the DFT-computed electron density in [{NN-(MgH)₂}₂] reveals a counterintuitive interaction between two formally closed-shell H ^- ligands that are separated by 3.106 A. This weak interaction could play an important role in H₂ desorption. Although the molecular product after H₂ release could not be characterized experimentally, DFT calculations on the proposed decomposition product, that is, the low-valence tetranuclear Mg(I) cluster [(NN-Mg₂)₂], predict a structure with two almost-parallel, localized Mg-Mg bonds. As in a previously reported β-diketiminate Mg^I dimer, the Mg-Mg bond is not characterized by a bond critical point, but instead displays a local maximum of electron density midway between the atoms, that is, a non-nuclear attractor (NNA). Interestingly, both of the NNAs in [(NN-Mg₂)₂] are connected through a bond path that suggests that there is bonding between all four Mg^I atoms. Size matters: The temperature that is required for the elimination of H₂ in magnesium hydride clusters is dependent on the cluster size (see scheme). Detailed experimental and/or theoretical data for magnesium hydride and magnesium(I) clusters are reported. Copyright

Full text of DOI:10.1002/chem.201300684

DOI: 10.1002/chem.201300684
Well-Defined Molecular Magnesium Hydride Clusters: Relationship between
Size and Hydrogen-Elimination Temperature
Julia Intemann,[a] Jan Spielmann,[b] Peter Sirsch,*[c] and Sjoerd Harder*[d]
Abstract: A new tetranuclear magne-
ACHTUNGTRENNUNGsium hydride cluster, [{NN-(MgH)2}2],
decreasing cluster size is observed, in
line with previously reported theoreti-
cal predictions on (MgH2)n model sys-
tems. Deuterium-labeling studies fur-
ther demonstrate that the released H2
solely originates from the oxidative
coupling of two hydride ligands and
not from other hydrogen sources, such
as the b-diketiminate ligands. Analysis
of the DFT-computed electron density
in [{NN-(MgH)2}2] reveals a counterin-
tuitive interaction between two formal-
ly closed-shell Hꢀ ligands that are sepa-
rated by 3.106 ꢀ. This weak interaction
could play an important role in H2 de-
sorption. Although the molecular prod-
uct after H2 release could not be char-
acterized experimentally, DFT calcula-
tions on the proposed decomposition
product, that is, the low-valence tetra-
nuclear Mg(I) cluster [(NN-Mg2)2],
predict a structure with two almost-par-
ꢀ
which was based on a N N-coupled
bis-b-diketiminate ligand (NN2ꢀ), was
obtained from the reaction of [{NN-
(MgnBu)2}2] with PhSiH3. Its crystal
structure reveals an almost-tetrahedral
arrangement of Mg atoms and two dif-
ferent sets of hydride ions, which give
rise to a coupling in the NMR spec-
trum (J=8.5 Hz). To shed light on the
relationship between the cluster size
and H2 release, the thermal decomposi-
tion of [{NN-(MgH)2}2] and two closely
related systems that were based on
similar ligands, that is, an octanuclear
magnesium hydride cluster and a di-
meric magnesium hydride species, have
been investigated in detail. A lowering
of the H2-desorption temperature with
ꢀ
allel, localized Mg Mg bonds. As in a
previously reported b-diketiminate MgI
ꢀ
dimer, the Mg Mg bond is not charac-
terized by a bond critical point, but in-
stead displays a local maximum of elec-
tron density midway between the
atoms, that is, a non-nuclear attractor
(NNA). Interestingly, both of the
NNAs in [(NN-Mg2)2] are connected
through a bond path that suggests that
there is bonding between all four MgI
atoms.
Keywords: cluster compounds
·
elimination · hydrides · hydrogen
storage · magnesium
Introduction
[a] J. Intemann
Stratingh Institute
University of Groningen
Nijenborgh 4, 9747AG Groningen (Netherlands)
Magnesium hydride is receiving widespread attention as a
potential hydrogen-storage material.[1] This compact, light-
weight salt-like material is rich in hydrogen (7.7 wt%) and
advantageously displays reversible hydrogen-release/uptake:
MgH2QMg+H2. However, its drawback is the relatively
high temperatures (about 3008C) that are required for hy-
drogen release. This requirement is directly related to unfav-
orable thermodynamics for this formal reductive elimination
[b] Dr. J. Spielmann
Anorganische Chemie
Universitꢁt Duisburg-Essen
Universitꢁtsstrasse 5, 45117 Essen (Germany)
[c] Dr. P. Sirsch
Institut fꢂr Anorganische Chemie
Universitꢁt Tꢂbingen
Auf der Morgenstelle 18, 72076 Tꢂbingen (Germany)
E-mail: Peter.Sirsch@uni-tuebingen.de
step (DH=74.4(3) kJmolꢀ1
;
DS=135.1(2) Jmolꢀ1 Kꢀ1).[2]
The high enthalpic contribution to the stability of bulk
MgH2 is primarily due to the enormous lattice energy for
[d] Prof. Dr. S. Harder
the salt-like rutile structure of [Mg2+Hꢀ ]1 (DH=
2
Anorganische Chemie
Universitꢁt Erlangen-Nꢂrnberg
Egerlandstrasse 1, 91058 Erlangen (Germany)
E-mail: Sjoerd.Harder@chemie.uni-erlangen.de
2718 kJmolꢀ1).[3]
Recent calculations have shown that thermodynamics
and, thus, the desorption temperature can be influenced to
some extent by a variation in particle size.[4,5] Downsizing
(MgH2)n particles results in a rapid increase in the surface/
volume ratio and in the number of less strongly bound sur-
face hydride ions. This effect starts to become noticeable for
(MgH2)n clusters with n<19 (i.e., particles smaller than
1.3 nm) and is dramatically amplified for clusters in the sub-
nanometer range. At the extreme, monomeric linear MgH2
is calculated to have a negative decomposition enthalpy
Supporting information for this article, including details on the crystal
structures of [{NN-(MgnBu)2}2], [{NN-(MgnBu)ACTHUNTRGNEU(GN MgH)}2], and [{NN-
(MgH)2}2], NMR spectra of [{NN-(MgH)2}2], PGSE-NMR spectra of
[(DIPP-nacnacMgH·THF)2] and [(DIPP-nacnacCaH·THF)2], cartesi-
an coordinates and geometrical parameters for the calculated struc-
tures of [{NN’-(MgH)2}2] and [(NN’-Mg2)2], topological parameters
for the H···H interaction in [{NN’-(MgH)2}2] at different levels of
theory, and plots of the interatomic surfaces for the H1 and H2 atoms
in [{NN’-(MgH)2}2], is available on the WWW under http://dx.doi.org/
10.1002/chem.201300684.
8478
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 8478 – 8489

FULL PAPER
(DH=ꢀ5.5 kJmolꢀ1).[4] Although such size effects have
been experimentally verified to some extent, (MgH2)n clus-
ters in the sub-nanometer range are not easily obtained by
using the typical solid-state approaches.[5] However, a molec-
ular “bottom-up” approach allows for the synthesis of small,
well-defined magnesium hydride clusters that could be seen
as the smallest possible magnesium hydride particles that
are encapsulated by a ligand framework. Owing to their low
hydride content, these molecular clusters are impracticable
for efficient hydrogen-storage purposes; however, they
could serve as well-defined model systems. Their molecular
nature allows for their in-depth analysis and precise investi-
gation down to the atomic level.
Recently, dimeric b-diketiminate magnesium hydride
complex 1, [(DIPP-nacnacMgH)2], has been reported[6]
(DIPP-nacnacꢀ =[(2,6-iPr2C6H3)NC(Me)CHC(Me)N(2,6-
iPr2C6H3)]ꢀ). As in the previously reported heavier calcium
congener,[7] this complex is stable towards ligand exchange
and the formation of the insoluble salt (MgH2)n, on account
of the sterically demanding 2,6-di-isopropylphenyl substitu-
ents (DIPP). The thermal release of H2 could possibly result
in formation of the structurally known MgI complex 2,
[(DIPP-nacnacMg)2].[8] However, complex 1 is described as
extraordinarily stable (the reported decomposition tempera-
ture is 3008C) and can only be converted into MgI complex
2 by reduction with K.[9] Conversely, pressurizing complex 2
with H2 does not give hydride complex 1; however, the oxi-
dation of MgI with AlH3 resulted in the formation of com-
plex 1.[9] It seems that the direct reductive elimination of H2
requires the combined action of more than two metal atoms.
Therefore, we systematically investigated synthetic methods
towards multi-metallic magnesium hydride complexes.
lar magnesium hydride clusters and on the proposed decom-
position product, a low-valence [(MgI)4] cluster.
Results and Discussion
Synthesis and structures: The magnesium hydride complex
[NN-(MgH)2] was obtained from the reaction of the ligand
NN-H2 with nBu2Mg and the stepwise conversion of the two
nBu groups into hydrides by reaction with PhSiH3
(Scheme 1). It was found that the first MgnBu functionality
Scheme 1. Synthesis of [NN-(MgH)2].
is considerably more reactive than the second. The first nBu
anion could be smoothly transformed into a hydride by reac-
tion with PhSiH3 in toluene at 608C (1 h), however, conver-
sion of the second nBu anion required higher temperatures
and prolonged reaction times (808C, 48 h). This difference
in reactivity allowed for the isolation and characterization of
the intermediate [NN-(MgnBu)ACTHNUGTRNEUNG(MgH)] species, which was
obtained as luminescent yellow crystalline blocks in 41%
yield.
We recently introduced a set of bridged bis-b-diketiminate
ligands with large DIPP substituents.[10] A dianionic ligand
with a para-phenylene bridge (PARA2ꢀ) was previously
used in the isolation of a large molecular magnesium hy-
dride cluster, [PARA3Mg8H10].[11] This cluster contains a
6+
Mg8H10 core that is encapsulated by three PARA2ꢀ ligands
and releases five equivalents of H2 at a temperature of only
2008C. Herein, we report the synthesis, structure, and ther-
mal decomposition of a tetranuclear magnesium hydride
The starting complex [NN-(MgnBu)2], the intermediate
complex [NN-(MgnBu)ACTHUNGTRENNG(U MgH)], and the final magnesium
hydride product [NN-(MgH)2], which was obtained as
yellow pyramid-shaped crystals in 48% yield (the complex
was also highly luminescent in the yellow-wavelength
range), were all characterized by single-crystal X-ray diffrac-
tion. Their structures are shown in Figure 1 and selected
bond lengths are summarized in Table 1. All of the com-
ꢀ
cluster that is based on the N N-coupled bis-b-diketiminate
ligand NN2ꢀ. With three clusters of various sizes in hand, we
evaluated the thermally initiated elimination of H2 as a
function of cluster size. In addition, we report a first theoret-
ical investigation of possible Hꢀ···Hꢀ interactions in molecu-
Chem. Eur. J. 2013, 19, 8478 – 8489
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8479

P. Sirsch, S. Harder et al.
Table 1. Selected distances [ꢀ] for the crystal structures shown in
Figure 1.[a]
ACHUNTRGEN[NUG {NN-AHCTUNGTRNE(NUGN MgnBu)2}2]
ꢀ
ꢀ
Mg1 N1
2.068(2)
2.046(2)
2.225(3)
2.211(3)
2.751(1)
Mg2 N3
2.049(2)
2.084(2)
2.241(3)
2.242(3)
2.755(1)
ꢀ
ꢀ
Mg1 N2
Mg2 N4
ꢀ
ꢀ
Mg1 C35
Mg2 C39
ꢀ
ꢀ
Mg C35’
Mg2 C39’
Mg1···Mg1’
Mg2···Mg2’
AHCTUNGTREGNU[NN {NN-ACHTUNREGTG(NNUN MgnBu)ACUTHNGTRENNUNG
ꢀ
ꢀ
Mg1 N1
Mg2 N3
2.056(1)
2.049(1)
2.247(2)
1.83(2)
3.605(1)
3.517(1)
ꢀ
ꢀ
Mg1 N2
Mg2 N4
ꢀ
ꢀ
Mg1 C35
Mg2 C35
ꢀ
ꢀ
Mg1 H1’
Mg2 H1
Mg1···Mg1’
3.767(1)
3.165(1)
Mg2···Mg2’
Mg2···Mg1’
Mg1···Mg2
ACHUTNRGEN[NUG {NN-CAHTUNGTREN(NUNG MgH)2}2]
ꢀ
Mg N1
2.025(1) [2.055]
2.010(1) [2.034]
1.771(12) [1.849]
1.845(6) [1.828]
4.65(2) [4.713]
2.93(4) [3.106]
Mg···Mg’
3.586(1) [3.555]
3.030(1) [3.154]
3.492(1) [3.580]
3.492(1) [3.580]
3.030(1) [3.154]
3.586(1) [3.555]
ꢀ
Mg N2
Mg···Mg’’
Mg···Mg’’’
Mg’···Mg’’
Mg’···Mg’’’
Mg’’···Mg’’’
ꢀ
Mg H1
ꢀ
Mg H2
H1···H1’
H2···H2’
[a] Theoretical DFT values for [{NN’-(MgH)2}2] are given in square
brackets.
hedron. Bridging nBuꢀ and Hꢀ ions result in the formation
of an eight-membered ring instead of two separated four-
membered rings. Further conversion into [NN-(MgH)2] af-
fords a dimeric aggregate with a similar tetrahedral assem-
bly of Mg2+ nuclei and an eight-membered Mg4H4 ring. The
transition from a “double dimer” structure into a cyclic
structure upon the replacement of nBuꢀ groups with Hꢀ
groups is due to the fact that the spherically symmetrical hy-
dride ligand favors a more-linear bridging mode. For exam-
ple, in [{NN-(MgBu)ACHTUNTRGNEUNG(MgH)}2], the Mg-H-Mg and Mg-nBu-
Mg angles are 157.5(1)8 and 89.4(1)8, respectively.
The structure of [{NN-(MgH)2}2] contains a tetrahedron
of Mg2+ nuclei, in which the Mg···Mg’ distances range from
3.030(1) to 3.586(1) ꢀ. The vertices of this tetrahedron are
either bridged by the NN2ꢀ ligands or by Hꢀ groups. This
pattern is comparable to a recently reported tetranuclear hy-
dride-rich magnesium cluster,[12] which crystallized with a
similar “adamantane-like” structure; however, the [{NN-
(MgH)2}2] cluster is highly symmetrical (D2). The asymmet-
ric unit, which represents one-quarter of the tetranuclear ag-
gregate, is repeated by three perpendicular two-fold rotation
Figure 1. Crystal structures of: a) [{NN-(MgnBu)2}2], a crystallographic
two-fold axis is positioned horizontally in the plane of projection;
b) [{NN-(MgnBu)ACHTUNGTRENNUNG(MgH)}2], a crystallographic two-fold axis is positioned
vertically in the plane of projection; and c) [{NN-(MgH)2}2], three per-
pendicular crystallographic two-fold axes run through the H1 and H1’
ꢀ
axes (one through the middle of the N N bonds, one
ꢀ
atoms, through the H2 and H2’ atoms, and through the middle of the N
N bond (D2 symmetry). In all of these structures, the iPr substituents and
through the H1 and H1’ atoms, and one through the H2 and
most H atoms are omitted for clarity.
H2’ atoms). The b-diketiminate units in the NN ligand are
ꢀ
twisted and rotation around the N N bond is somewhat hin-
dered. Therefore, the NN ligand shows axial chirality that is
similar to that in BINOL ligands. The cluster contains two
NN ligands of equal axial chirality and, therefore, is homo-
chiral (D2 symmetry, Scheme 2).
plexes form dimeric aggregates that contain four Mg2+
nuclei, albeit in different arrangements and with differences
in the bridging of nBuꢀ and Hꢀ ions.
A
E
The structure of the tetranuclear cluster [{NN-(MgH)2}2]
is maintained in solution. This property is nicely demonstrat-
ed by low-temperature 1H NMR spectroscopy: Two different
four Mg2+ nuclei are more or less situated in one plane. The
nBuꢀ ions each bridge two Mg2+ centers, thereby resulting
in two separated four-membered Mg2nBu2 rings. The partial
conversion of the nBuꢀ ions into Hꢀ ions gives a structure in
which the Mg2+ nuclei are situated at the corners of a tetra-
hydride ligands, H1 and H2, give two triplets at d=2.85 and
2
3.64 ppm at ꢀ508C, with a coupling of J
ACHTUNGTREN(NGNU H,H)=8.5 Hz (see
the Supporting Information). The relatively large difference
8480
www.chemeurj.org
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 8478 – 8489

Well-Defined Molecular Magnesium Hydride Clusters
FULL PAPER
study of their thermal decomposition as a function of cluster
size. These investigations allow for verification of the calcu-
lational result that smaller magnesium hydride clusters show
lower enthalpies for the elimination of hydrogen.[4]
In a previous report, we described the thermal decompo-
sition of the octanuclear magnesium hydride cluster [PAR-
A3Mg8H10].[11] It was found that a solution of [PAR-
A3Mg8H10] in toluene was surprisingly stable, even when
heated at 1808C for 2 days, that is, well above its boiling
point, in a J-Young NMR tube. However, heating this solid
material on a high-vacuum line with an attached Tçpler
pump afforded decomposition of the complex at 2008C,
during which process the yellow microcrystalline material
became reddish brown. Gas quantification by pumping the
gas quantitatively into a calibrated burette of a Tçpler
pump system indicated the release of the expected
4.9
G
gate. Moreover, the gas does not condense in liquid N2, but
is fully converted into condensable water after leading it
over CuO at 3008C. After leading the gas into deuterated
THF, a clear H NMR resonance could be observed at d=
Scheme 2. Schematic representation and chirality interconversion for
[{NN-(MgH)2}2].
4.55 ppm, which is the chemical shift for H2 in this sol-
vent.[13]
Complex [{NN-(MgH)2}2] showed similar behavior: Heat-
ing a solution of the complex in toluene in a J-Young tube
at 1808C for 2 days did not show any noticeable decomposi-
tion of the complex. However, the stepwise heating of the
solid material on a vacuum line afforded a color change
from yellow to red at 1758C and showed signs of gas release.
This was confirmed by gas measurements using the Tçpler
pump. Quantitative gas release was slow and required
between their chemical shifts (Dd=0.79 ppm) is presumably
due to the ASIS effect of the aryl substituents. The high-
field signal is assigned to the H1 atom, which is situated be-
tween two aryl rings. The chemical shift of the low-field
signal (d=3.64 ppm) also compares well to the signal for
the hydride ligands in [NN-(MgnBu)ACTHNUTRGNEUNG(MgH)], which appears
as a singlet at d=3.59 ppm. The herein observed magnetic
coupling between two hydrides in a magnesium complex is
ACHTUNGTRENNUNG
only the second reported example of such an interaction.
N
2
The coupling constant, J
G
than those observed in [PARA3Mg8H10] (4.5 and 5.2 Hz).[11]
The magnesium hydride complex [{NN-(MgH)2}2] is more
dynamic than [PARA3Mg8H10] in toluene. Whereas the
ꢀ
latter complex shows hydride hydride coupling at 1008C,
This observation represents new evidence for the fact that
smaller molecular magnesium hydride clusters can release
H2 at substantially lower temperatures than bulk MgH2
itself (decomposition temperature>3008C). In agreement
with earlier calculations by de Jong and co-workers,[4] the
smaller magnesium hydride cluster [{NN-(MgH)2}2] already
released H2 at 1758C, that is, significantly lower than the
2008C that was required for [PARA3Mg8H10]. This result
led us to reinvestigate the thermal stability of the small di-
meric magnesium hydride cluster 1 under similar conditions.
Solid complex 1 was stepwise heated on a high-vacuum
line with a Tçpler pump. At 1258C, the light-yellow powder
changed into an orange-red melt that released gas by bub-
bling. Similar to the results described above, gas quantifica-
signals for the H1 and H2 atoms in [{NN-(MgH)2}2] already
coalesce at 258C (DG° =56(1) kJmolꢀ1, see the Supporting
Information). This result is likely due to conformational
ꢀ
changes, rather than Mg H bond-breaking/making process-
es. A change in the axial chirality of one of the NN ligands
in the D2-symmetric cluster would result in the achiral (het-
erochiral) S4-symmetric structure (Scheme 2), in which all of
the hydride ligands are equal. A further change in the chir-
ality of the second NN ligand would afford a homochiral
D2-symmetric cluster that is a mirror image of the original
D2-symmetric cluster. It should be noted that the positions
of the H1 and H2 atoms are exchanged during this process.
1
Therefore, the coalescence of the H1/H2 H NMR resonan-
ces can simply be explained as a change in the axial chirality
tion indicated that 1.0
(ꢁ0.05) mol equivalents of hydrogen
ꢀ
of both ligands through 908 rotations about their N N axes.
gas were released (the nature of the released gas was con-
firmed by oxidation over hot CuO and independently by
1H NMR spectroscopy). We noted that compound 1 also
started to decompose at 1258C in a capillary under a N2 at-
mosphere. The mismatch with the reported decomposition
Thermal-decomposition studies: The availability of a set of
magnesium hydride clusters with similar b-diketiminate-
based ligand systems but of different nuclearities enables a
Chem. Eur. J. 2013, 19, 8478 – 8489
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8481

P. Sirsch, S. Harder et al.
temperature (>3008C) is likely due to the fact that the de-
composition requires a short induction period; thus, fast
heating rates during melting-point determination may ex-
plain this discrepancy. Although we only have a limited
amount of data points, the H2-elimination temperature
seems to decrease with cluster size.
hydride coupling constants in [{NN-(MgH)2}2] (8.5 Hz) and
[PARA3Mg8H10] (4.5 and 5.2 Hz), but still remote from the-
ꢀ
oretical predictions for the value of geminal H H coupling
in linear MgH2 (55–80 Hz).[14] At present, it is unclear
ꢀ
whether the hydride hydride NMR coupling is due to
through-bond or through-space mechanisms (or a combina-
tion). Although the exact positions of the hydride ligands
from X-ray diffraction data should be treated with care, it
seems likely that there is no correlation between the cou-
The nature of the product(s) after the thermal decomposi-
tion of [{NN-(MgH)2}2] is hitherto unknown. Most of the
reddish solid residue is soluble in aromatic solvents.
1H NMR analysis shows the presence of several ligand sig-
nals and there is no indication of paramagnetic behavior.
We could not characterize any products by crystallization
and subsequent structure determination. In addition, the
thermal decomposition of complex 1 gave a reddish powder,
which, when dissolved in [D6]benzene, showed several sets
ꢀ
pling constant and the Mg H bond length. The average
ꢀ
Mg H bond lengths in [PARA3Mg8H10] and [{NN-
(MgH)2}2] are both 1.81(2) ꢀ. The various polymorphs of
ACHUTNGRENdNUG imeric complex 1 or its dimeric THF adduct, 1·ACHTUNGTRENNUNG(THF)2,
ꢀ
show somewhat (partially significantly) larger Mg H bond
lengths of 1.91(5) ꢀ and 1.93(2) ꢀ, respectively.[6,15] The
larger coupling constant in complex 1 might be explained by
the fact that the hydride ligands in the dimer are part of a
four-membered ring system, whereas, in the larger aggre-
gates, they are integrated in more-extended ring systems.
The smaller ring causes a significantly shorter hydride···hy-
dride distance. We determined a reasonably accurate value
of 2.32(2) ꢀ for the H···H distance in the THF adduct
1·(THF)2,[15] whereas a distance of 2.44(4) ꢀ was found in
another polymorph.[6a] The H···H distances in two poly-
morphs of THF-free complex 1 are 2.35(8) and 2.65(4).[6]
Therefore, the average H···H distance of 2.44(4) ꢀ is signifi-
cantly shorter than the H···H distances in [{NN-(MgH)2}2]
(2.76(4) ꢀ) and [PARA3Mg8H10] (average: 2.86(4) ꢀ) and
could explain the larger observed magnetic coupling; this
correlation between the hydride···hydride distance and its
magnetic coupling constant hints that a significant part of
the coupling proceeds through space.
1
of H NMR signals that may include those of complex 2. Al-
though this result could not be unambiguously confirmed, it
is tempting to believe that low-valence MgI clusters are
formed. First of all, the color change to red may indicate
such a formation (the THF adduct of low-valence compound
2 is described as orange-red).[6a] Secondly, the Mg···Mg dis-
tances in complexes 1 and 2 are quite similar. Therefore, a
simple elimination of H2 might generate a tetranuclear MgI
cluster [(NN-Mg2)2] with a comparable tetrahedral arrange-
ment of the metal atoms and, therefore, of similar global
shape. However, larger MgI clusters are hitherto unknown
and it is questionable whether four valence electrons would
be sufficient to stabilize the bonding in a Mg4 tetrahedron
(see the DFT calculations below). We could not convert the
red residues back into magnesium hydride complexes by
pressurizing with H2 (<100 bar).
Deuterium-labeling studies: Deuterium-labeled species can
provide important information regarding the source of hy-
drogen upon thermal decomposition. In addition, informa-
tion regarding the dynamics of hydride exchange can be ob-
tained. For this purpose, we synthesized a series of magnesi-
um deuteride clusters by using the original synthetic proce-
dures but substituting PhSiH3 for PhSiD3.
This above-described H/D exchange process is highly
AHCTUNGTREGUNsNN olvent dependent: No exchange is observed in pure
[D6]benzene, but the addition of [D8]THF results in com-
plete exchange into a statistical mixture of complexes 1-H2
(25%), 1-H,D (50%), and 1-D2 (25%) within 1 h. This nota-
ble influence of THF indicates that H/D exchange likely
proceeds through a smaller aggregate. Concerning the ag-
gregation state in [D6]benzene, it should be noted that
PGSE NMR measurements show that not only the magnesi-
um hydride [(DIPP-nacnacMgH·THF)2] but also its calcium
analogue, [(DIPP-nacnacCaH·THF)2], are dimeric in aro-
matic solvents (see the Supporting Information). The hydro-
dynamic radii for [(DIPP-nacnacMgH·THF)2] and [(DIPP-
nacnacCaH·THF)2] are both 6.8(1) ꢀ. This result is signifi-
cantly larger than the hydrodynamic radii for [DIPP-nac-
nacZnH] (5.2(1) ꢀ)[16] and [Mes-nacnacZnH] (Mes=mesi-
tyl, 5.0(2) ꢀ),[17] both of which are monomeric in solution.
The ratio of the hydrodynamic radii between the magnesium
and zinc hydride complexes or the calcium and zinc hydride
complexes, 1.3:1, corresponds to a volume ratio of about
2.2:1, thus indicating that the THF adducts of the magnesi-
um and calcium hydride complexes are dimeric in aromatic
solvents (at least at room temperature). The observation of
Hꢀ exchange between magnesium hydride clusters: The di-
meric hydride complex 1-H2 and its deuteride dimer (1-D2)
were dissolved in C6D6 and possible H/D exchange between
the complexes was monitored by 1H NMR spectroscopy.
After several hours at room temperature, no changes were
observed. Whilst performing the same experiments in a mix-
ture of [D6]benzene/[D8]THF (4:1), an exchange of Hꢀ and
Dꢀ ions was observed: A new signal for the Hꢀ ligand ap-
peared at slightly higher field. This 1H NMR signal is as-
signed to the mixed dimer with bridging Hꢀ and Dꢀ ligands
ꢀ
and showed a characteristic 1:1:1 triplet, owing to H D cou-
pling (2J
(H,D)=4.4 Hz). Although we could not determine
the coupling constant between the two identical Hꢀ ligands
ꢀ
in complex 1-H2, the observed H D coupling corresponds
ꢀ
to an estimated equivalent H H coupling constant of
ꢀ
28.6 Hz (based on the gyromagnetic ratios for H and D).
H D magnetic coupling in complex 1-H,D is additional evi-
ꢀ
This value is substantially larger than the observed hydride
dence for the dimeric state of the magnesium hydride com-
8482
www.chemeurj.org
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 8478 – 8489

Well-Defined Molecular Magnesium Hydride Clusters
FULL PAPER
plex under exchange conditions. Therefore, THF-stimulated
H/D exchange likely proceeds through a short-lived, THF-
stabilized, monomeric state.
Interestingly, both larger clusters, that is, [{NN-(MgH)2}2]
and [PARA3Mg8H10], do not show H/D exchange with deu-
terated clusters (also not after prolonged heating at 608C or
after THF addition). This result is likely due to their inher-
ent higher stability towards dissociation and underscores the
proposed dissociative mechanism for hydride exchange be-
tween aggregates.
Hꢀ/H2 exchange: The s-bond metathesis reaction, which is
well-established in organolanthanide chemistry,[18] has re-
cently also been shown to be operative in Group 2 chemis-
try.[19,20] A solution of the dimeric calcium deuteride com-
plex [(DIPP-nacnacCaD·THF)2] in benzene already showed
D/H exchange with H2 (1 bar) at room temperature (90%
exchange was observed within 20 min).
Under similar conditions, a solution of magnesium deuter-
ide complex 1-D2 in [D6]benzene did not show any reaction
with H2 (1 bar). However, heating this solution to 608C led
ꢀ
to partial exchange and the formation of H D and complex
1-H,D in the reaction mixture, as detected by 1H NMR spec-
troscopy. Keeping the sample at this temperature overnight
led to 85% exchange. Therefore, magnesium hydride com-
plexes are much less reactive than their corresponding calci-
um hydrides. The exchange process is not only accelerated
by heat but also by the addition of THF. A solution of 1-D2
in [D6]benzene/[D8]THF (4:1) already gave slow exchange
at room temperature (10% after 1 h, 50% after 5 days).
Scheme 3. Possible routes for D/H exchange.
This result was clearly visible by the detection of dissolved
1H NMR spectrum showed the characteristic triplets for
1
ꢀ
ꢀ
H D and complex 1-H2 by H NMR spectroscopy (see the
magnetic H H coupling. The inability of these larger clus-
Supporting Information). In contrast to the experiment in
pure C6D6, we could not detect the intermediate mixed com-
plex 1-H,D in the presence of THF. We tentatively explain
this result with the mechanism shown in Scheme 3. In THF,
dimer 1-D2 is in equilibrium with small amounts of mono-
meric magnesium deuteride species, on which any empty co-
ordination sites are likely filled by THF ligands. These
highly reactive monomers exchange Dꢀ for Hꢀ by s-bond
metathesis with H2. A subsequent recombination of mono-
mers gives dimeric complex 1-H2. The absence of detectable
amounts of complex 1-H,D can be explained by assuming a
slow, rate-determining deaggregation step and a considera-
bly faster s-bond metathesis. In pure C6D6, either the s-
bond metathesis is much slower or D/H exchange takes
place along a route in which the dimer only breaks open
(Scheme 3).
ters to exchange their Hꢀ ligand with H2 might be an indica-
tion that smaller monomeric metal hydride complexes—or
at least opened clusters—are required for this process.
Whereas dimers like complex 1 could be in a dimer/mono-
mer equilibrium, the larger multimetallic/multihydride clus-
ters clearly are not: A solution of [PARA3Mg8H10] in tol-
uene even shows clear Hꢀ···Hꢀ NMR coupling at 1008C.[11]
The importance of a monomeric state for hydride/H2 ex-
change has also been suggested in [(Cp2*LnH)2] complexes
(Ln=lanthanide, Cp=cyclopentadienyl).[21]
Thermal-decomposition studies with deuterium labeling:
Within the closed system of a J-Young NMR tube, various
magnesium deuterides were thermally decomposed as solids
and the gasses were analyzed in THF by H NMR spec
1
G
AHCTUNGTREGcNNUN opy (for technical details, see the Supporting Information).
Interestingly, neither of the larger magnesium deuteride
clusters, [{NN-(MgD)2}2] and [PARA3Mg8D10], exchange
their deuteride ligands with H2, even at higher temperatures
(608C). Partially deuterated clusters might give broad
1H NMR signals for the remaining Hꢀ ligand (due to H/D
magnetic coupling) that might not be clearly visible at the
start. Therefore, we also investigated the H/D exchange by
subjecting [{NN-(MgH)2}2] to D2. No exchange was observed
(even after heating at 608C for several days) and the
The decomposition of [{NN-(MgD)2}2] showed no significant
ꢀ
quantities of H D or H2 (small quantities can be explained
by the incomplete labeling of the magnesium compounds:
PhSiD3 with 95% isotopic purity was used). Also, the ther-
mal decomposition of the larger magnesium deuteride clus-
ꢀ
ter [PARA3Mg8D10] showed no significant formation of H
D. In contrast, clusters that contained 50% hydride and
50% deuteride (which were prepared by using a 1:1 mixture
ꢀ
of PhSiD3 and PhSiH3) eliminated large quantities of H D.
Chem. Eur. J. 2013, 19, 8478 – 8489
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8483

P. Sirsch, S. Harder et al.
Table 3. Selected atomic properties that were derived from the theoreti-
cal electron densities of [{NN’-(MgH)2}2] and [(NN’-Mg2)2].
This result means that hydrogen elimination from [{NN-
(MgH)2}2] and [PARA3Mg8H10] is solely based on the cou-
pling of hydride ligands. The involvement of ligand protons
can largely be excluded.
q(W) [e]
E(W) [hartree]
Vol(W) [ꢀ3][a]
Loc(W)[b]
ACHUNTRGEN[NUG {NN’-CAHTUNGTREN(NUNG MgH)2}2]
Mg
N1
N2
H1
H2
AHCTUNGTERNN[GUN (NN’-Mg2)2]
Mg
N1
N2
+1.67
ꢀ0.93
ꢀ1.23
ꢀ0.81
ꢀ0.81
ꢀ200.103
ꢀ55.155
ꢀ55.282
ꢀ0.719
6.7
14.4
16.6
18.9
17.0
96.2
78.5
80.2
81.3
79.2
We also investigated the thermal decomposition of solid
[{NN-(MgH)2}2] or [PARA3Mg8H10] under a D2 atmosphere
(1 bar). In both cases, not only H2 but also significant
ꢀ
amounts of H D could be detected. This result could either
ꢀ0.734
be due to H/D exchange prior to the decomposition or to a
reversible H2-elimination process. Because previous studies
have shown (see above) that the larger magnesium deuter-
ide clusters are inert towards D/H exchange with H2, we
+1.25
ꢀ0.95
ꢀ1.24
ꢀ0.70
ꢀ200.181
ꢀ55.153
ꢀ55.272
ꢀ0.084
16.7
14.9
16.9
14.6
94.1
78.3
80.1
30.6
NNA
ꢀ
think that the observed H D originates from a reversible
[a] At 1(r)=0.001; [b] Loc(W) is dimensionless.
process. However, considering the harsher conditions that
are needed for H2 elimination (175–2008C), we do not fully
exclude partial H/D exchange of hydride-only clusters with
D2 prior to the elimination step.
and a high ratio of G/1b is observed, which represents the ki-
netic energy per electron. The topological parameters for
ꢀ
DFT calculations and charge-density analysis: To explore
the bonding situation in [{NN-(MgH)2}2] in more detail, we
carried out a topological analysis of its electron density,
1(r), as derived from DFT calculations, by using the “atoms
in molecules” approach.[22] For our theoretical model
system, we replaced the DIPP substituents on each ligand
by methyl groups; this model system will be denoted as
[{NN’-(MgH)2}2]. The experimental twist angle between b-
diketiminate units, 83.4(2)8, is retained in the optimized ge-
ometry (84.98) and all other parameters are in good agree-
ment with their experimental counterparts, bearing in mind
that positions of hydrogen atoms are always approximate in
X-ray diffraction studies (for a detailed geometry, see the
Supporting Information).
the Mg H moieties are comparable to those of their coun-
terparts in solid b-MgH2, which has recently been investigat-
ed in a theoretical study.[25] However, in this bulk material,
the hydride anions are linked to three different Mg atoms,
ꢀ
which renders the Mg H bonds slightly longer and weaker
than in the molecular species reported herein.
The atomic charges, q(W), as obtained by integrating the
electron density over the respective atomic basins, reveal
that 0.81 electrons are transferred from the Mg atoms onto
each hydride, thus leaving the Mg atom with a charge of
+1.67, that is, close to its oxidation state (Table 3). The addi-
tional charge on the hydrides significantly increases their
atomic volumes, which are now similar to those of N atoms.
For comparison, H atoms that are attached to ligand carbon
atoms in [{NN’-(MgH)2}2] carry charges of between ꢀ0.02
and +0.02 and display atomic volumes of Vol(W)=7.0–
7.6 ꢀ3, that is, more than two-times smaller.
The topological parameters of 1(r) for selected bonds in
[{NN’-(MgH)2}2] are listed in Table 2 and selected atomic
ꢀ
properties are shown in Table 3. The values for the Mg H
bonds confirm their strongly ionic character:[23] At the bond
The large size and polarizability of the hydridic H atoms
in [{NN’-(MgH)2}2] allow them to engage in mutual interac-
tions over a relatively long distance. As shown in the con-
tour plot of 1(r) in Figure 2, there is an accumulation of
electron density between two of the hydrides (H2 and H2’,
3.106 ꢀ apart) and a corresponding bond path between
these atoms can be identified. As expected for such a long-
range contact, the topological parameters at the BCP, as
shown in Table 2, point to a weak, closed-shell interaction:
The value for 1(r) at the BCP (1b) is low, albeit somewhat
2
critical point (BCP),[24] 1(r) is relatively small and 5 1(r),
the Laplacian of the electron density, is clearly positive. In
addition, the total energy (Hb) at the BCP is close to zero,
Table 2. Selected topological parameters at the BCPs for the theoretical
electron densities in [{NN’-(MgH)2}2] and [(NN’-Mg2)2].[a]
2
1b
[eꢀꢀ3
5 1b
e
Hb
G/1b
d
]
[eꢀꢀ5
]
[hartreeꢀꢀ3
]
[hartreeeꢀ1
]
higher than for typical van der Waals interactions,[26] 5 1b is
2
ACHTUNGTRENNUNG[{NN’-ACHTUNGTRENNUNG(MgH)2}2]
ꢀ
Mg N1
0.327 +7.171 0.10 +0.04
0.348 +7.662 0.10 +0.03
2.014 ꢀ7.673 0.02 ꢀ1.46
0.249 +3.366 0.03 ꢀ0.01
0.247 +3.473 0.01 ꢀ0.01
1.43
1.45
0.46
0.98
1.01
0.41
0.17
0.19
1.20
0.18
0.17
0.05
relatively small and positive, and the energy density (Hb) is
positive and close to zero. Furthermore, two ring critical
points (RCPs)[27] are close (0.36 ꢀ each) to the H2···H2’
ꢀ
Mg N2
ꢀ
N1 N1’
ꢀ
Mg H1
BCP, with values of 1ACHTNUGRTENUNG(RCP) that are only slightly smaller
ꢀ
Mg H2
(0.046 eꢀꢀ3), thus showing that the electron density is rela-
tively flat in this region and that the H···H bond path is po-
tentially unstable.[28] The topological parameters for this in-
ꢀ
H2 H2’
0.047 +0.327 3.43
0.00
[(NN’-Mg2)2]
ꢀ
Mg N1
0.320 +6.999 0.11 +0.04
0.329 +7.133 0.10 +0.03
1.997 ꢀ7.425 0.00 ꢀ1.43
0.195 ꢀ0.418 0.07 ꢀ0.06
1.42
1.42
0.46
0.15
0.15
0.20
0.22
1.20
0.35
0.03
ꢀ
Mg N2
ꢀ
ꢀ
teraction resemble those of homopolar C H···H C interac-
tions, for which the term hydrogen hydrogen bonding has
recently been introduced. Although, in this case, the hydro-
gen atoms only carry negligible or very small (mostly posi-
ꢀ
N1 N1’
ꢀ
ꢀ
Mg NNA
NNA1···NNA2 0.046 +0.074 0.13
0.00
[a] Values e and d are dimensionless.
8484
www.chemeurj.org
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 8478 – 8489

Well-Defined Molecular Magnesium Hydride Clusters
FULL PAPER
on the thermal release of H2 in ammonia borane showed
that similar BH···HB interactions are highly relevant.[33]
The first identification of intramolecular hydride···hydride
interactions in the molecular species reported here points to
the possibility of a more-ubiquitous bonding feature in com-
pounds that contain hydridic H atoms and also emphasizes
their potential role on the reaction coordinate of the dehy-
drogenation reaction. Moreover, it offers a unique opportu-
nity to directly compare two pairs of hydridic hydrogen
atoms in an identical chemical environment, that is, one that
engages in H···H interactions (H2, H2’) and one that does
not (H1, H1’).
As mentioned before, the [{NN’-(MgH)2}2] cluster displays
D2 symmetry and it is interesting to note that the Mg-H-Mg
angles are significantly different for the H1 and H2 atoms:
The Mg-H2-Mg angle (153.08) is clearly larger than the Mg-
H1-Mg angle (117.18); this more-linear arrangement helps
to bring the H2 and H2’ atoms into closer proximity to one
another, in line with their mutual interaction. Whereas the
atomic charges on both the H1 and H2 atoms are the same,
the values of their atomic volume, Vol(W), differ to some
extent (Table 3): H2 is about 10% smaller than H1, which is
comparable to the volume changes that are associated with
Figure 2. Contour plot of 1(r) in the Mg’’-H2’-Mg’’’-H2 plane of [{NN’-
(MgH)2}2]; selected atoms and bond paths are superimposed. The BCPs
and RCPs are represented by green and red circles, respectively. Lines
are drawn at 2ꢄ10ꢀn, 4ꢄ10ꢀn, 8ꢄ10ꢀn (n=ꢀ3, ꢀ2, ꢀ1, 0, +1, +2), and
0.0065 au.
[29]
ꢀ
hydrogen hydrogen bonding.
The value of the electron
localization, Loc(W), is also lower for the H2 atom, thus in-
dicating that its electrons exchange slightly more than those
of the H1 atom. Finally, the H2 and H2’ atoms share a sig-
nificant interatomic surface (for details, see the Supporting
Information) and the weak interaction between these atoms
is further reflected in the value of the delocalization index,
d, which typically measures the number of electron pairs
that are shared between two atoms, irrespective of existing
tive) charges,[29] this result might indicate that these interac-
tions represent two sides of the same coin.
Whilst a bonded radius of 1.553 ꢀ is observed for the H2
and H2’ atoms (i.e., the distance between the nuclei and the
H···H BCP), the nonbonded radius can also be estimated by
tracing the extension of the electron density in the opposite
direction, towards the outside of the complex. By taking the
0.001 envelope for 1(r), nonbonded radii of about 1.7 ꢀ are
obtained for the H2 and H2’ atoms, which is clearly greater
than the accepted van der Waals radius of 1.1 ꢀ for hydro-
gen.[30]
These radii suggest that there is significant interference
between hydridic hydrogen atoms H2 and H2’ and a con-
comitant redistribution of the electron density in the intera-
tomic region, which results in a stabilizing interaction, as in-
dicated by the H···H bond path. The atomic energies, E(W),
confirm that both the H2 and H2’ atoms are more stable by
39 kJmolꢀ1 compared to the non-interacting H1 and H1’
atoms. Stabilizing interactions between significantly hydridic
H atoms have recently been observed in a DFT study of a
range of binary and complex hydrides of light main group
metals, in which it was shown that these interactions could
occur at a wide range of H···H separations (2.3–3.1 ꢀ).[25]
Although anion···anion bonding interactions might seem
contradictory at first, they have been known for some time
and are thought to play an important role in the stabiliza-
tion of crystals.[31] Hydride···hydride interactions have also
been observed in the extended structures of alkali-metal
amidoboranes.[32] Very recently, NMR deuteration studies
bond paths.[34] Table 2 shows that the value for d
small but non-zero, whereas, for atoms H1 and H1’, which
are separated by 4.713 ꢀ, d(H1,H1’) equals zero.
ACHTUNGTRENUN(NG H2,H2’) is
AHCTUNGTRENNUNG
The differences between the atomic energies of the H1
and H2 atoms are significant (see above); however, there is
ꢀ
no direct method to obtain a precise value for the H H in-
teraction energy from these values, because the formation of
a H···H interaction can also raise or lower the atomic ener-
gies of other atoms in the compound.[25a] By using the Espi-
nosa correlation,[35] which has previously been employed for
hydrogen bonds, as well as for other weak interactions,[35,36]
the H···H interaction energy in [{NN’-(MgH)2}2] would be
about 3.0 kJmolꢀ1, given that this equation also holds true
ꢀ
for hydride hydride interactions. This value is comparable
in magnitude to those of very weak hydrogen bonds[37] and
lies within the range of other previously characterized weak
ꢀ
interactions, such as intermolecular O···O, N···O, C H···O,
[36]
ꢀ
ꢀ
and C H···H C interactions.
DFT calculations are also a perfect tool for studying the
possible low-valence residue that remains after H2 elimina-
tion. To evaluate the possible existence of a tetrahedral
[(MgI)4] cluster, we simply optimized the DFT geometry of
the hydride-free complex [(NN’-Mg2)2], by taking the coor-
dinates of [{NN’-(MgH)2}2] (without the four hydride moiet-
ies) as a starting point. The obtained DFT structure is
Chem. Eur. J. 2013, 19, 8478 – 8489
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8485

P. Sirsch, S. Harder et al.
Approximately 0.7 electrons can be allocated to each of
the two identical pseudo-atoms in [(NN’-Mg2)2], which com-
prise an NNA along with its corresponding basin (Table 3).
The electron density at the position of these NNAs
(0.197 eꢀꢀ3) is only marginally higher than at the corre-
sponding Mg-NNA BCPs (0.28 ꢀ away, Table 3), which
points to a relatively flat profile of 1(r) along this axis. The
extremely small value for the kinetic energy per electron at
the NNAs (G/1ACTHNUTRGNEUNG
(NNA)=0.03 hartreeeꢀ1) is typical for
pseudo-atoms and has been interpreted as evidence for the
loosely bound and potentially mobile nature of their elec-
tron density.[39b] This is also reflected in the decreased locali-
zation of electrons within the boundaries of the pseudo-
atoms, as expressed by the parameter Loc(W). As shown in
Table 3, the majority of electrons on the Mg and N atoms
are localized and only exchange to a limited extent, whereas
the electrons of the pseudo-atoms are clearly more delocal-
ized.
The topology of 1(r) in the region between the Mg atoms
and the NNAs displays the typical features of metall–metall
bonding (Table 2): The Laplacian and the total energy, H(r),
at the Mg-NNA BCPs are small and negative, which is indi-
cative of weak covalent bonding, whereas the G/1(r) ratio is
clearly smaller than 1, thus again confirming the shared
character of the interaction. As expected for the weak
Figure 3. DFT-optimized geometry of [(NN’-Mg2)2]; all hydrogen atoms
ꢀ
are omitted for clarity. Selected distances [ꢀ] and angles [8]: Mg N1
ꢀ
ꢀ
2.063, Mg N2 2.057, N1 N2 1.430, Mg···Mg’ 4.242, Mg···Mg’’ 2.854,
Mg···Mg’’’ 3.776, Mg’···Mg’’ 3.776, Mg’···Mg’’’ 2.854, Mg’’···Mg’’’ 4.242;
C1-N1-N1’’’-C1’’’ 91.05.
shown in Figure 3, along with selected structural parameters.
The geometry of [(NN’-Mg2)2] is similar to that of its parent
compound, albeit with significant changes in the Mg4 core.
Starting from an almost-tetrahedral Mg4 core, with Mg···Mg
distances that ranged from 3.154 to 3.580 ꢀ, the structure
optimized towards a minimum in which two of the four
Mg···Mg distances were now clearly shorter (2.854 ꢀ),
whereas the other two distances were noticeably elongated
(4.242 ꢀ). Given a Mg···Mg distance of 3.20 ꢀ in Mg
metal,[38] this suggests the presence of a dimer with two lo-
ꢀ
calized Mg Mg bonds (as in complex 2), rather than a Mg4
tetrahedral cluster with delocalized bonding.
This result was confirmed by a topological analysis of the
electron density in [(NN’-Mg2)2]. Figure 4 shows plots of
2
1(r) and 5 1(r) in a contour map that contains two of the
Mg atoms that are in closer contact; selected bond paths
and critical points are also shown. Clearly, 1(r) accumulates
ꢀ
along the Mg Mg axis and a bond path can be identified be-
tween the atoms. Another striking feature is that the Mg
atoms are not directly linked to each other through a bond
critical point, but instead are linked through a local maxi-
mum of 1(r) midway between the atoms. Such non-nuclear
attractors (NNAs) have been observed before on rare occa-
sions; however, most of these reports were for bulk metals
or model metal clusters.[39] NNAs have also been linked to
defects and color F centers in crystals, as well as to solvated
electrons.[40] Very recently, an experimental NNA was re-
ported for the first time in a stable molecular system that
was similar to the one reported herein, that is, the afore-
mentioned compound 2, in which two MgI atoms were
joined together in a dimeric MgI species.[41]
2
Figure 4. Contour plots of 1(r) (top) and the negative Laplacian, ꢀ5 1(r)
(bottom), in the Mg’-Mg’’’-NNA plane of [(NN’-Mg2)2]; selected atoms
and bond paths are superimposed. The BCPs and RCPs are represented
by green and red circles, respectively. Lines are drawn at 2ꢄ10ꢀn, 4ꢄ
10ꢀn, 8ꢄ10ꢀn au (n=ꢀ3, ꢀ2, ꢀ1, 0, +1, +2), and 0.028 au (1(r) only),
for the Laplacian with both positive (dashed) and negative values (solid).
Solid and dashed lines represent charge concentration and depletion,
AHCTUNGTREGUNrNN espectively.
8486
www.chemeurj.org
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 8478 – 8489

Well-Defined Molecular Magnesium Hydride Clusters
FULL PAPER
nature of the bonding, the value of 1
small. However, the observed value is more than twice as
large as those previously reported for metals (e.g., Be) or
(BCP) is relatively
within the same range as those of NNA···NNA interactions
that have been reported for some metals and larger metal
clusters in theoretical studies.[39b] It has been proposed that
it is this network of linked pseudo-atoms that might be re-
sponsible for the delocalized binding and conducting proper-
ties in some of these systems.[39b] In turn, this proposal
would suggest that, for [(NN’-Mg2)2], a description of the
bonding solely in terms of a dimer that is linked by two lo-
small metal clusters, such as Li4 and Na4, for which 1ACHTUNTRGNEUNG(BCP)
typically lies within the range 0.04–0.09 eꢀꢀ3 for a metal···
NNA interaction.[39a,b,e] A similar value has only been report-
ed very recently in the above-mentioned experimental study
on a dimeric MgI molecular species.[41]
ꢀ
As in this aforementioned study, the pseudo-atoms in
[(NN’-Mg2)2] are fairly large and comparable in size to the
other atoms in this compound (see Vol(W), Table 3). The
large region that is claimed by the concentration of charge
around the NNA is also noticeable; in Figure 4, this region
calized Mg Mg bonds might not be fully adequate; the
bonding between the four Mg atoms is also of a delocalized
nature, at least to a small extent.
ꢀ
extends by more than 2.2 ꢀ perpendicular to the Mg Mg
Conclusion
axis. As expected for a MgI species, the atomic volume for
the Mg atoms is significantly increased compared to that of
the MgII compound [{NN’-(MgH)2}2] and is now more than
double the size. The lower oxidation state of Mg in [(NN’-
Mg2)2] is not reflected to the same extent in its atomic
charge, which is only decreased by 0.42 electrons. However,
it should be taken into account that about 0.35 electrons per
Mg atom have been transferred onto an NNA, which gives
an effective Mg charge of +0.95 if added to the value given
above.
The use of directly coupled b-diketiminate ligands gave rise
to the formation of a small tetranuclear magnesium hydride
cluster, [{NN-(MgH)2}2], in which the Mg atoms are situated
at the corners of a tetrahedron. In the cluster, there are two
1
sets of different hydride positions and, in its H NMR spec-
trum, a coupling constant of J=8.5 Hz is observed, which
confirms that the solid-state structure is maintained in solu-
tion. The hydride positions exchange through a configura-
ꢀ
tional change in the ligand, rather than through Mg H bond
Finally, Figure 5 shows a 3D representation of the nega-
tive Laplacian in [(NN’-Mg2)2], which also reveals a note-
worthy feature: To the best of our knowledge, this is the
breaking: Enantiomers of the chiral (D2-symmetric) cluster
exchange through a non-chiral (S4-symmetric) intermediate.
Also, the NMR coupling constant between the two hydride
ligands in [(DIPP-nacnacMgH)2] (1, J=28.6 Hz) could be
determined. Whilst the coupling constant in various molecu-
lar magnesium hydride clusters does not seem to correlate
ꢀ
to the Mg H bond lengths, a comparison of the hydride···hy-
dride distances and their NMR coupling constants does
show a correlation: Shorter distances relate to larger cou-
pling constants. This correlation could be an argument in
favor of a through-space coupling mechanism.
The smaller tetranuclear magnesium hydride cluster
[{NN-(MgH)2}2] eliminates H2 at
a lower temperature
(1758C) than the larger cluster [PARA3Mg8H10] (2008C). A
reinvestigation of hydrogen release in dimeric [(DIPP-nac-
nacMgH)2] already shows the loss of one equivalent of H2 at
1208C. As predicted by theory, the H2-elimination tempera-
tures seem to decrease with cluster size. Deuterium-labeling
studies show that H2 elimination is based on the coupling of
hydride ligands and is not due to deprotonation of the
ligand by the hydride ligands. These studies also show that,
as a solution in toluene, hydride exchange between the clus-
ters is either not observed, such as in [PARA3Mg8H10] and
[{NN-(MgH)2}2] (also not at higher temperatures), or is slow,
as in [(DIPP-nacnacMgH)2]. For the latter compound, this
process can be accelerated by the addition of the polar sol-
vent THF. Also, Hꢀ/H2 exchange is not observed for larger
clusters and is slow for [(DIPP-nacnacMgH)2]; again, the
process is accelerated by the addition of THF. The thermal
decomposition of [{NN-(MgH)2}2] under a D2 atmosphere
Figure 5. 3D-envelope map of the negative Laplacian of [(NN’-Mg2)2] for
2
5 1(r)=ꢀ0.1 eꢀꢀ5, superimposed on bond paths and selected gradient
paths. The BCPs and RCPs are represented by green and red circles,
ACHTUNGTRENNUNGrespectively.
first example for a molecular species in which two NNAs
are linked by a bond path. The topological parameters, as
listed in Table 2, show that this interaction is clearly weaker
than the Mg-NNA interactions; however, these values are
ꢀ
gave significant amounts of H D. This result could be due
to the microscopic reversibility of the hydrogen-elimination
Chem. Eur. J. 2013, 19, 8478 – 8489
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8487

P. Sirsch, S. Harder et al.
3
3
12H; iPr), 1.31 (d, J
G
E
process, although H/D exchange through another route
cannot be excluded.
12H; iPr), 1.61 (s, 12H; Me backbone), 1.78 (s, 12H; Me backbone), 2.85
(br s, 2H; MgH), 3.06 (sept, 3J(H,H)=7.0 Hz, 4H; iPr), 3.31 (sept, 3J-
(H,H)=6.8 Hz, 4H; iPr), 3.50 (br s, 2H; MgH), 4.53 (s, 4H; H back-
From an analysis of the electron density of a DFT-opti-
mized model system of [{NN-(MgH)2}2], we identified a hy-
dride···hydride interaction between two of the four hydride
ions at the cluster core. This interaction can be characterized
as a weak closed-shell interaction that is similar in magni-
tude to other weak interactions that have been reported to
date. This first identification of such an interaction within a
molecular system highlights a potential role on the reaction
coordinate that leads to the release of H2 and also points to
a wider significance of this type of interaction in molecular
compounds, outside the solid state.
DFT calculations also demonstrate that the possible de-
composition product following H2 desorption, a low-valence
[(NN’-Mg)4] cluster, does not have a structure in which MgI
atoms are situated at the corners of a tetrahedron. This tet-
ranuclear MgI compound is best described as a dimer that is
AHCTUNGTRENNUNG
bone), 7.02–7.14 ppm (m, 12H; aryl H); 13C NMR (75 MHz, [D6]benzene,
208C): 20.8 (Me backbone), 23.7 (iPr), 23.8 (iPr), 24.5 (iPr), 24.6 (iPr),
26.3 (Me backbone), 28.6 (iPr), 28.8 (iPr), 91.7 (backbone), 124.0 (aryl
C), 124.1 (aryl C), 125.7 (aryl C), 143.0 (aryl C), 143.0 (aryl C), 144.7
(aryl C), 167.2 (backbone), 168.1 ppm (backbone); elemental analysis
(%) calcd for C68H100Mg4N8 (Mw =1126.83): C 72.48, H 8.95; found:
C 73.36, H 8.13; deviations are likely owing to the partial incorporation
of n-hexane (which appears in the crystal-structure analysis as
severely disordered molecule): elemental analysis (%) calcd for
C68H100Mg4N8·C6H14 (Mw =1213.01): C 73.27, H 9.47.
Crystal-structure determination: For details on the crystal structures of
[{NN-(MgnBu)2}2], [{NN-(MgnBu)
the Supporting Information. CCDC-893202 [{NN-(MgH)2}2], CCDC-
893203 [{NN-(MgnBu)(MgH)]2, and CCDC-893204 [{NN-(MgnBu)2}2]
contain the supplementary crystallographic data for this paper. These
data can be obtained free of charge from The Cambridge Crystallograph-
ic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
ACHTNUGRTEN(NNUG MgH)}2], and [{NN-(MgH)2}2], see
Computational details: DFT calculations were performed with the Gaus-
ꢀ
linked through two localized Mg Mg bonds. However, a
sian 09 program suite[42] by using the B3LYP density functional[43] and the
6–311GACHTUNGTRNEUNG
(d,p) basis set.[44] An additional polarization and a diffuse func-
closer analysis of the bonding situation revealed a highly in-
teresting topology of the electron density, with two intercon-
nected non-nuclear attractors. This topology has previously
been observed in metals and metal clusters and it suggests a
small degree of electron delocalization between the two
tion were employed for all magnesium and nitrogen atoms, as well as for
the hydridic hydrogen atoms at the cluster core. All of the geometry opti-
mizations were performed without imposing any symmetry constraints
and the structures were confirmed as true minima by calculating the ana-
lytical frequencies. Structure optimizations were also performed by using
other functionals and basis sets of different size, which gave very similar
geometries and electron densities (for further details, see the Supporting
Information). In particular, the H···H interaction was invariant to the
choice of basis set and level of theory and it was also confirmed by an
MP2 calculation at the previously determined B3LYP geometry. The top-
ology of the electron density was analyzed by using the software package
AIMALL.[45] The reported atomic energies for the hydridic hydrogen
atoms in [{NN’-(MgH)2}2] were estimated by multiplying their atomic ki-
ꢀ
Mg Mg bonds. We are currently engaged in the isolation
and crystallization of multinuclear MgI species for experi-
mental studies on such unusual bonding interactions.
Experimental Section
netic energies by the factor ꢀ(gꢀ1), with g=ꢀV/T, as described, for ex-
N
All experiments were carried out under an argon atmosphere by using
standard Schlenk techniques and freshly dried degassed solvents. The fol-
lowing starting materials were prepared according to literature proce-
dures: NN-H2 and [NN-(MgnBu)2].[10b]
Synthesis of [NN-ACHTUNGTRENNUNG(MgnBu)ACHUTNGTRENG(NUN MgH)]: A solution of [NN-(MgnBu)2]
(240 mg, 0.355 mmol) and PhSiH3 (38 mg, 0.351 mmol) in toluene (8 mL)
was heated at 608C for 1 h. The resulting solution was concentrated to a
volume of 1 mL and slowly cooled to ꢀ278C, which led to the crystalliza-
Acknowledgements
tion of [{NN-(MgBu)ACHTUNGTRENNUNG(MgH)}2] as large yellow blocks. Yield: 90 mg
We thank Prof. Dr. R. Boese and D. Blꢁser for performing the X-ray
data collection, M. Zꢁhres for performing the PGSE measurements, and
we acknowledge the DFG for financial support. We also thank the
bwGRiD project, as part of the German D-Grid initiative, for providing
computing facilities.
(0.068 mmol, 41%); 1H NMR (300 MHz, [D6]benzene, 208C): d=ꢀ0.44
ACHTUNGTRENNUNG
(t, 3J(H,H)=9.4 Hz, 4H; nBu), 0.14–0.30 (m, 4H; nBu), 0.48 (t, 3J-
ACHTUNGTRENNUNG(H,H)=7.1 Hz, 6H; nBu), 0.71 (m, 2H; nBu), 1.05 (m, 2H; nBu), 1.18
3
3
(d, J
G
G
3
3
(d, J
(H,H)=6.7 Hz, 12H; iPr), 1.49 (d, J
(s, 12H; Me backbone), 1.91 (s, 12H; Me backbone), 3.21–3.31 (m, 8H;
iPr), 3.59 (s, 2H; MgH), 4.66 (s, 4H; H backbone), 7.05–7.16 ppm (m,
12H; aryl H); 13C NMR (75 MHz, [D8]THF, 208C): 6.3 (nBu), 14.0
(nBu), 21.0 (Me backbone), 24.7 (Me backbone), 24.7 (iPr), 24.8 (iPr),
25.3 (iPr), 25.6 (iPr), 28.4 (iPr), 29.5 (nBu), 29.8 (iPr), 30.3 (nBu), 92.5
(backbone), 124.5 (aryl C), 124.9 (aryl C), 125.6 (aryl C), 142,9 (aryl C),
143.2 (aryl C), 146.9 (aryl C), 166.3 (backbone), 167.5 ppm (backbone);
elemental analysis (%) calcd for C76H116Mg4N8·C6H6 (Mw =1239.05):
C 73.67, H 9.44; found: C 73.13, H 9.18.
[1] a) K.-F. Aguey-Zinsou, J.-R. Ares-Fernꢅndez, Energy Environ. Sci.
2010, 3, 526–543; b) I. P. Jain, C. Lal, A. Jain, Int. J. Hydrogen
Energy 2010, 35, 5133–5144.
[2] J. F. Stampfer Jr., C. E. Holley Jr., J. F. Suttle, J. Am. Chem. Soc.
1960, 82, 3504–3508.
[3] C. M. Stander, R. A. Pacey, J. Phys. Chem. Solids 1978, 39, 829–832.
[4] R. W. P. Wagemans, J. H. van Lenthe, P. E. de Jongh, A. J. van Dill-
en, K. P. de Jong, J. Am. Chem. Soc. 2005, 127, 16675–16680.
[5] P. E. de Jongh, P. Adelhelm, ChemSusChem 2010, 3, 1332–1348.
[6] a) S. P. Green, C. Jones, A. Stasch, Angew. Chem. 2008, 120, 9219–
9223; Angew. Chem. Int. Ed. 2008, 47, 9079–9083; b) S. J. Bonyhady,
C. Jones, S. Nembenna, A. Stasch, A. J. Edwards, G. J. McIntyre,
Chem. Eur. J. 2010, 16, 938–955.
Synthesis of [NN-(MgH)2]:
A solution of [NN-(MgnBu)2] (400 mg,
0.592 mmol) and PhSiH3 (170 mg, 1.571 mmol) in toluene (10 mL) was
heated at 808C for 48 h. After the removal of all of the volatile com-
pounds under vacuum, the residue was dissolved in n-hexane (5 mL).
Slowly cooling this solution to ꢀ278C gave yellow crystals in the form of
tetrahedrons that were strongly yellow luminescent under UV light.
Yield: 150 mg (0.133 mmol, 48%); 1H NMR (300 MHz, [D6]benzene,
[7] S. Harder, J. Brettar, Angew. Chem. 2006, 118, 3554–3558; Angew.
Chem. Int. Ed. 2006, 45, 3474–3478.
[8] S. P. Green, C. Jones, A. Stasch, Science 2007, 318, 1754–1757.
3
3
208C): d=0.80 (d, J
(H,H)=6.8 Hz, 12H; iPr), 1.16 (d, J
N
8488
www.chemeurj.org
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 8478 – 8489

Well-Defined Molecular Magnesium Hydride Clusters
FULL PAPER
[9] S. J. Bonyhady, D. Collis, G. Frenking, N. Holzmann, C. Jones, A.
Stasch, Nat. Chem. 2010, 2, 865–869.
[33] D. J. Wolstenholme, K. T. Traboulsee, Y. Hua, L. A. Calhoun, G. S.
McGrady, Chem. Commun. 2012, 48, 2597–2599.
[10] a) D. F.-J. Piesik, S. Range, S. Harder, Organometallics 2008, 27,
6178–6187; b) D. F.-J. Piesik, R. Stadler, S. Range, S. Harder, Eur. J.
Inorg. Chem. 2009, 3569–3576.
[11] S. Harder, J. Spielmann, J. Intemann, H. Bandmann, Angew. Chem.
2011, 123, 4242–4246; Angew. Chem. Int. Ed. 2011, 50, 4156–4160.
[12] M. Arrowsmith, M. S. Hill, D. J. MacDougall, M. F. Mahon, Angew.
Chem. 2009, 121, 4073–4076; Angew. Chem. Int. Ed. 2009, 48, 4013–
4016.
[34] X. Fradera, M. A. Austen, R. F. W. Bader, J. Phys. Chem. A 1999,
103, 304–314.
[35] E. Espinosa, E. Molins, C. Lecomte, Chem. Phys. Lett. 1998, 285,
170–173.
[36] a) Y. V. Nelyubina, I. L. Dalinger, K. A. Lyssenko, Angew. Chem.
2011, 123, 2944–2946; Angew. Chem. Int. Ed. 2011, 50, 2892–2894;
b) C. F. Matta, N. Castillo, R. J. Boyd, J. Phys. Chem. B 2006, 110,
563–578.
[13] G. R. Fulmer, A. J. M. Miller, N. H. Sherden, H. E. Gottlieb, A. Nu-
delman, B. M. Stoltz, J. E. Bercaw, K. I. Goldberg, Organometallics
2010, 29, 2176–2179.
[14] I. Alkorta, P. F. Provasi, G. A. Aucar, J. Elguero, Magn. Reson.
Chem. 2008, 46, 356–361.
[15] J. Spielmann, D. F.-J. Piesik, S. Harder, Chem. Eur. J. 2010, 16,
8307–8318.
[16] J. Spielmann, D. F.-J. Piesik, B. Wittkamp, G. Jansen, S. Harder,
Chem. Commun. 2009, 3455–3456.
[17] S. Schulz, T. Eisenmann, D. Schuchmann, M. Bolte, M. Kirchner, R.
Boese, J. Spielmann, S. Harder, Z. Naturforsch. 2009, 64b, 1397–
1400.
[18] a) P. L. Watson, J. Am. Chem. Soc. 1983, 105, 6491–6493; b) G.
Jeske, H. Lauke, H. Mauermann, H. Schumann, T. J. Marks, J. Am.
Chem. Soc. 1985, 107, 8111–8118.
[37] U. Koch, P. L. A. Popelier, J. Phys. Chem. 1995, 99, 9747–9754.
[38] J. Emsley, The Elements, 2nd ed., Clarendon Press, Oxford, 1995.
[39] a) C. Gatti, P. Fantucci, G. Pacchioni, Theor. Chim. Acta 1987, 72,
433–458; b) W. L. Cao, C. Gatti, P. J. MacDougall, R. F. W. Bader,
Chem. Phys. Lett. 1987, 141, 380–385; c) A. M. Pendꢅs, M. A.
Blanco, A. Costales, P. Mori-Sꢅnchez, V. LuaÇa, Phys. Rev. Lett.
1999, 83, 1930–1933; d) V. LuaÇa, P. Mori-Sꢅnchez, A. Costales,
M. A. Blanco, A. M. Pendꢅs, J. Chem. Phys. 2003, 119, 6341–6350;
e) J. Friis, G. K. H. Madsen, F. K. Larsen, B. Jiang, K. Marthinsen,
R. Holmestad, J. Chem. Phys. 2003, 119, 11359–11366.
[40] a) R. F. W. Bader, J. A. Platts, J. Chem. Phys. 1997, 107, 8545–8553;
b) G. K. H. Madsen, C. Gatti, B. B. Iversen, L. Damjanovic, G. D.
Stucky, V. I. Srdanov, Phys. Rev. B 1999, 59, 12359–12369; c) A.
Taylor, C. F. Matta, R. J. Boyd, J. Chem. Theory Comput. 2007, 3,
1054–1063.
[19] J. Spielmann, F. Buch, S. Harder, Angew. Chem. 2008, 120, 9718–
9718; Angew. Chem. Int. Ed. 2008, 47, 9576–9580.
[41] J. A. Platts, J. Overgaard, C. Jones, B. B. Iversen, A. Stasch, J. Phys.
Chem. A 2011, 115, 194–200.
[20] G. Zeng, S. Li, Inorg. Chem. 2010, 49, 3361–3369.
[21] L. Maron, O. Eisenstein, J. Am. Chem. Soc. 2001, 123, 1036–1039.
[22] R. F. W. Bader, Atoms in Molecules: A Quantum Theory, Oxford
University Press, Oxford, 1990.
[23] For a classification of chemical bonding, based on the AIM ap-
proach, see: P. Macchi, A. Sironi, Coord. Chem. Rev. 2003, 238–239,
383–412.
[24] The BCP represents the point of minimum electron density, 1(r),
along the bond path between two atoms, in which 1(r) is locally
greater than in any direction away from the path.
[25] P. Sirsch, F. C. Che, J. T. Titah, G. S. McGrady, Chem. Eur. J. 2012,
18, 9476–9480.
[42] Gaussian 09, Revision C.01, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V.
Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X.
Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Son-
nenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa,
M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven,
J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J.
Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J.
Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar,
J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W.
Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A.
Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, ꢇ.
Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian,
Inc., Wallingford CT, 2009.
[26] R. G. A. Bone, R. F. W. Bader, J. Phys. Chem. 1996, 100, 10892–
10911.
[27] A ring critical point is a minimum of 1(r) in two dimensions.
[28] a) D. Cremer, E. Kraka, T. S. Slee, R. F. W. Bader, C. D. H. Lau,
T. T. Nguyen-Dang, P. J. MacDougall, J. Am. Chem. Soc. 1983, 105,
5069–5075; b) R. F. W. Bader, C. F. Matta, F. Cortꢆs-Guzmꢅn, Orga-
nometallics 2004, 23, 6253–6263.
[29] a) C. F. Matta, J. Hernꢅndez-Trujillo, T.-H. Tang, R. F. W. Bader,
Chem. Eur. J. 2003, 9, 1940–1951; b) C. F. Matta in Hydrogen Bond-
ing: New Insights, Chapter 9: Challenges and Advances in Computa-
tional Chemistry and Physics Series (Ed.: S. J. Grabowski), Springer,
Dordrecht, 2006, pp. 337–375.
[43] a) A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652; b) C. Lee, W.
Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785–789; c) B. Miehlich, A.
Savin, H. Stoll, H. Preuss, Chem. Phys. Lett. 1989, 157, 200–206.
[44] a) A. D. McLean, G. S. Chandler, J. Chem. Phys. 1980, 72, 5639–
5648; b) R. Krishnan, J. S. Binkley, R. Seeger, J. A. Pople, J. Chem.
Phys. 1980, 72, 650–654.
[45] T. A. Keith, AIMALL (Version 12.11.09), TK Gristmill Software,
Overland Park KS, USA, 2012 (aim.tkgristmill.com).
[30] R. S. Rowland, R. Taylor, J. Phys. Chem. 1996, 100, 7384–7391.
[31] a) E. A. Zhurova, V. G. Tsirelson, A. I. Stash, A. A. Pinkerton, J.
Am. Chem. Soc. 2002, 124, 4574–4575; b) Y. V. Nelyubina, M. Y.
Antipin, K. A. Lyssenko, Russ. Chem. Rev. 2010, 79, 167–187.
[32] D. J. Wolstenholme, J. T. Titah, F. N. Che, K. T. Traboulsee, J. Floge-
ras, G. S. McGrady, J. Am. Chem. Soc. 2011, 133, 16598–16604.
[46] G. A. Zhurko, CHEMCRAFT (http://www.chemcraftprog.com).
Received: February 21, 2013
Published online: May 9, 2013
Chem. Eur. J. 2013, 19, 8478 – 8489
ꢃ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8489