Communications
Next, the macrocyclic trialdehydes 2a and 2b were
allowed to react with the C3-symmetrical triamine 1,1,1-
tris(4-aminophenyl)pentane (3a; Scheme 1). The reactions
were performed at room temperature without a template or
catalyst in a mixture of CH2Cl2 and MeOH (1.75:1 v/v) as the
solvent. After several days, we observed the formation of one
major compound (4a and 4b, respectively), which precipi-
tated in the form of an orange powder upon adiabatic removal
of CH2Cl2 in vacuum. The products showed remarkably
simple 1H and 13C NMR spectra (Supporting Information,
Figure S3,S4), with only one set of signals for the protons of
arene p ligands and one signal for the imine protons and
carbon atoms. The completeness of condensation was indi-
cated by the lack of aldehyde signals in the d = 10.0–10.5 ppm
region. Although NMR spectroscopy did not allow unambig-
uous determination of the stoichiometry and topology of the
products, formation of the dodecanuclear complexes resulting
from a [4+4] condensation was confirmed by low- and high-
resolution ESI mass spectrometry (Supporting Information,
Figure S7–S9). The complexes 4a and 4b were isolated with
yields of 55 and 45%, respectively (on a 50 mmol scale), which
corresponds to a 93–95% efficiency for each of the 12 imine
condensations.[15]
Trimers based on {(arene)Ru} complexes and 3-hydroxy-
2-pyridone ligands are known to undergo ligand exchange
reactions in polar solvents, such as methanol or water.[16]
However, attempted ligand exchange between 2a and 2b
under the conditions used for the synthesis of 4a and 4b
(room temperature, CD2Cl2/CD3OH 1.75:1 v/v) did not result
in the formation of mixed complexes after two days, as
determined by 1H NMR spectroscopy. Therefore, the trialde-
hydes 2a and 2b should be regarded as inert building blocks
on the timescale of the imine condensation process.
Diffusion of Et2O vapors into solutions of 4a and 4b in
CHCl3/MeOH (95:5) resulted in the growth of large, well-
shaped single crystals, which despite appearances diffracted
X-rays rather poorly. A satisfactory data set was nonetheless
obtained for a crystal of 4a[17] whose the structure was solved
in I2/a, an alternative setting of the monoclinic space group
C2/c. The solid-state structure of 4a is shown in Figure 1. The
complex displays approximate tetrahedral symmetry (T),
which is in line with the simple NMR spectra observed in
solution. The trinuclear metallamacrocycles occupy the four
vertices, whilst the triphenylmethane units span each of the
four faces. Notably, within a given molecule, all four metal-
lamacrocycles have identical relative configuration with
respect to rotation about the pseudo threefold axes that
connects each vertex with its opposing face (Figure 1a). The
crystal as a whole is, however, racemic and thus contains an
equal number of opposite stereoisomers.
The X-ray diffraction analysis also revealed a remarkable
feature: crystals of 4a are perforated by large solvent-
accessible voids; less that 30% of the total unit cell volume
is accounted for by the complex alone. This is partly due to the
relatively large internal volume of the cage, which we
estimate as being approximately 500 ꢀ3 (the volume of a
polyhedron which fits inside the cage; see Supporting
Information, Figure S14). The main contributing factor is,
however, the highly inefficient manner in which the cages
Figure 1. Molecular structure of complex 4a in the solid state. a) View
along the pseudo C3 axis, b) view along the crystallographic C2 axis
(Ru: spheres), and c) packing diagram as viewed down the crystallo-
graphic a axis (opposite enantiomers are shown in different shades of
gray). All hydrogen atoms have been omitted for clarity.
pack. The shortest points of contact between any two nearest
neighbors are localized on the metallamacrocycles and this
gives rise to large pores of up to 45 ꢀ in the cross-section,
which propagate along the crystallographic a axis (Figure 1c).
The presence of highly disordered solvent molecules in these
regions of the structure is most likely the reason why the
crystal does not diffract much beyond 1.2 ꢀ in resolution.
The high percentage of solvent-accessible volume in
crystalline 4a prompted us to investigate whether a material
with permanent porosity could be generated.[3a] N2 sorption
measurements were performed at 77 K with a sample of crude
amorphous 4a, and also with a sample of an X-ray-quality
crystalline product, after prolonged drying in vacuum. The
calculated Brunauer–Emmett–Teller surface areas were 30
and 15 m2 gÀ1 for the amorphous and dried crystalline
products, respectively, which indicates that larger voids
between the cages are no longer present, presumably because
of a structural collapse during the drying process. This
conclusion is further supported by the poor match between
the powder X-ray diffraction pattern of dried crystalline 4a
with that calculated from the single-crystal diffraction data
(Supporting Information, Figure S2).
The synthetic concept described above can be used to
create even larger cages. This was demonstrated by the
synthesis of complexes 5a and 5b from metallamacrocycles
2a and 2b and an expanded triamine, the trixenylmethane
derivative 3b (Scheme 1). Indeed, products of the [4+4]
condensation were obtained, albeit in lower yields than 4a
and 4b (20% and 18% for 5a and 5b, respectively). The
lower yields of the isolated products may indicate that the
geometry of the expanded triamine 3b is less favorable for
reaction with metallamacrocyclic trialdehydes. At the same
5516
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 5515 –5518