Synthesis and conformation studies of a dodecaazanonacyclotetratetracontane

Article abstract of DOI:10.1002/(SICI)1521-3765(19991001)5:10<3055::AID-CHEM3055>3.0.CO;2-N

Compound 3 was prepared by self-assembly of 1,3,5-pentanetriamine and aqueous formaldehyde in quantitative yield (Figure 1). This molecule can exist in four well-defined diamond-lattice conformations of symmetries D(2d), S₄, C(2v) and D(2d). Low-temperature ¹³C NMR spectroscopy indicates the existence of three main conformers; their relative populations depend on the solvent used. An extra set of low- intensity lines is also observed. A conformation interconversion scheme is proposed; it involves two additional less populated quasi-diamond-lattice intermediates derived from a helix compressed along its axis. One of these is trapped as a 1:2 clathrate with 1,4-dioxane; its crystal structure is reported.

Full text of DOI:10.1002/(SICI)1521-3765(19991001)5:10<3055::AID-CHEM3055>3.0.CO;2-N

FULL PAPER
Synthesis and Conformation Studies of a Dodecaazanona-
cyclotetratetracontane
M. Rachel Suissa,*[a,b] Christian Rùmming,[b] and Johannes Dale[b]
Abstract: Compound 3 was prepared by
self-assembly of 1,3,5-pentanetriamine
and aqueous formaldehyde in quantita-
tive yield (Figure 1). This molecule can
exist in four well-defined diamond-lat-
tice conformations of symmetries D2d,
S4, C2v and D2d. Low-temperature
13C NMR spectroscopy indicates the
existence of three main conformers;
their relative populations depend on
the solvent used. An extra set of low-
intensity lines is also observed. A con-
formation interconversion scheme is
proposed; it involves two additional less
populated quasi-diamond-lattice inter-
mediates derived from a helix com-
pressed along its axis. One of these is
trapped as a 1:2 clathrate with 1,4-
dioxane; its crystal structure is reported.
Keywords: azaalkanes ´ conforma-
tion analysis ´ NMR spectroscopy ´
polycycles ´ self-assembly
Introduction
Hexamethylenetetramine
1
(HMTA, Figure 1), first pre-
pared in 1860 by Butlerow[1] in
a reaction of gaseous ammonia
with paraformaldehyde, is
probably the earliest known
example of the formation of
multicyclic compounds by self-
assembly.[2] This symmetric ada-
mantane-like structure is read-
ily made in quantitative yield
because of the unique ability of
saturated six-membered rings
to adopt a perfect diamond-
lattice conformation. In 1955,
Krässig observed that 1,3-pro-
panediamine and formaldehyde
Figure 1. A homologous series of 8-, 16-, and 24-membered rings, prepared by condensation of ammonia or
amines with formaldehyde.
form a pentacyclooctaaza compound (2; Figure 1) with
remarkable ease.[3] Its conformation in a 1:1 clathrate[‡] with
benzene was determined in 1974 by Murray-Rust, who
pointed out that the observed conformation was the less
symmetric (S4) of the two possible diamond-lattice confor-
mations (D2d and S4).[4] Low-temperature NMR spectroscopy
of a toluene solution of compound 2, done in our laboratory,
revealed a mixture of the two conformers, while a CD2Cl2
solution contained practically only the D2d form.[5] Crystals
grown from CH2Cl2 were then shown by Rùmming to be a
stable complex of the D2d conformer with two solvent
molecules.[5]
From these structural data, it was predicted that the product
obtained from 2,2-dimethyl-1,3-propanediamine (Table 1, 4)
should be stable only in the D2d conformation, as was indeed
observed.[6] This in turn led us to the conclusion that a
bridging pentamethylene chain would be long enough to span
[a] Dr. M. R. Suissa
HF Oslo College, 0167 Oslo (Norway)
[b] Dr. M. R. Suissa, Prof. C. Rùmming, Prof. J. Dale
Department of Chemistry, University of Oslo
P.O. Box 1033, 0315, Oslo (Norway)
Fax : (‡ 47)22855441
E-mail: michals@ki-mail.uio.no
‡
[ ] By clathrates we mean enclosure compounds in which guest molecules
interact by weak van der Waals forces with their host molecules, while
in complexes there is a defined coordination between the macrocycle
and its guest.
Chem. Eur. J. 1999, 5, No. 10
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3055 $ 17.50+.50/0
3055

M. R. Suissa et al.
FULL PAPER
Table 1. The mass spectra (EI) of the multicyclic amines correspond mainly to those of di- or oligo-imines.
Macrocycle
Imine
Mass Spectrum [%][a]
Ref.
[1]
1
8
2
140 (70), 112 (11), 85 (9), 71 (5), 42 (100)
168 (80), 139 (17), 126 (21), 111 (12), 99 (31), 85 (74), 56 (25), 49 (16),
42 (100), 18 (39)
[3]
259 (28), 197 (34), 99 (82), 70 (53), 56 (59), 42 (100)
379 (14), 253 (31), 127 (100), 84 (51), 42 (52)
[3, 5]
4
5
[6]
[7]
584 (80), 583 (100), 293 (22)
3
6
152 (100), 125 (7), 98 (35), 82 (16), 70 (22), 56 (32).
[8]
[9]
154 (57), 112 (55), 98 (23), 85 (36), 70 (62), 56 (100), 42 (98), 28 (80)
664 (0.5), 499 (3), 333 (11), 167 (33), 137 (45), 124 (76), 109 (17), 95
(44), 81 (26), 67 (44), 42 (100)
7
[21]
[a] The mass spectra have been done by us.
across the 16-membered ring, one above and one below, to
produce a cagelike septicyclic molecule with a compact
ªtetrahedralº van der Waals surface of D2d symmetry (Ta-
ble 1, 5). The required 2,2,8,8-tetrakis(aminomethyl)nonane
was synthesized and condensed with formaldehyde to give
compound 5 in 81% yield.[7] Recently, we reported the
synthesis of a 24-membered ring 3 (1,3,7,9,11,15,17,19,23,25,27,
34-dodecaazanonacyclo[25.5.3.26,9.214,17.222,25.13,7.111,15.119,23.030,34]-
tetratetracontane) from 1,3,5-pentanetriamine and formalde-
hyde.[8] A full discussion of its properties is the subject of this
paper.
The three compounds mentioned above (1 ± 3) have cyclic
structures formed by a chain of alternating carbon and
nitrogen atoms, on to which four ªcarbon atom handlesº are
fused and which have an overall diamond-lattice structure.
Figure 1 shows an oblique view (a) and a vertical projection
Abstract in Norwegian: Forbindelse 3 ble syntetisert i kvanti-
tativt utbytte ved selvkondensasjon av 1,3,5-pentantriamin og
vandig formaldehyd. Molekylet kan opptre i tre veldefinerte
diamantgitter-konformasjoner med henholdsvis D2d, S4 og C2v
symmetri, samt i en mindre stabil D2d-form. Lavtemperatur
13C NMR-spektroskopi bekrefter at de tre hovedkonformerene
er tilstede i mengdeforhold som avhenger av det anvendte
lùsningsmiddel, og at den fjerde konformer gir svakere NMR
signaler. Mekanismen for utveksling av konformasjoner er
Ê
Ê
illustrert i et skjema som ogsa ma innbefatte to mellom-
produkter av ªkvasi-karakterº, dannet ved at en diamantgit-
terhelix komprimeres langs helixaksen. En av disse danner et
1:2 klatrat med 1,4-dioksan, og krystall-strukturen av dette
rapporteres.
3056
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3056 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 10

Polycycles
3055±3065
(b) of the D2d conformation of the eight-membered ring 1, the
16-membered ring 2, and of the 24-membered-ring 3. One
may consider HMTA as a lower homologue of this series,
since each of the four 8-membered rings that can be defined
follow the diamond-lattice structure (Figure 1, 1b). However,
the pairs of nitrogen atoms are too close, and two transannular
CH2 bridges are needed to fill the interior space. Higher
homologues should have fewer restrictions, and hence several
conformations become possible, depending on the size of the
molecule. As mentioned, the 16-membered ring system can
exist in two conformations, the most symmetrical of which is
shown in Figure 1 (2a, 2b). The next member of this series has
a 24-membered ring (Figure 1, 3a, 3b), which consists of four
rigid trans-decalin-like units connected by four CH2 groups. It
can exist in four well-defined conformations of symmetries
D2d, S4, C2v and D2d.
13. Curtius rearrangement, followed by hydrolysis, gave the
corresponding tris-ammonium-trichloride. The free amine
had to be distilled carefully. Condensation of the triamine
with formaldehyde at 108C (molar ratio 1:3) gave, after
approximately 30 minutes, compound 3 in quantitative yield.
The mechanism of the condensation reaction: Condensation
of amines with formaldehyde gives in most cases a polymer,
which can be isomerized, usually in an organic solvent like 1,4-
dioxane, to the corresponding multicyclic compound in high
yields. This condensation reaction meets the criteria of strict
covalent self-assembly.[2] A fundamental feature of this
process is its reversibility. Dynamic disassembly and re-
assembly enable recovery from initial mismatching and lead
ultimately to stable structures at thermodynamic equilibrium.
Clearly, cyclic products formed from polymers are expected to
be thermodynamically favoured both by enthalpy (less
conformational strain) and by entropy (larger number of
molecules). Normally the monomer is strained, whereas the
chosen oligomer would have a strain-free conformation.
The reaction between ammonia and formaldehyde to
give HMTA is perhaps the most extreme case: a perfect
diamond lattice of the highest possible symmetry [Eq. (1)].
The ªsquareº conformation of these molecules is made of
four identical parts joined together by four CH2 ªcornersº that
have adjacent gauche bonds of identical sign. These four CH2
ªcornersº are located in the same plane and carry geminally
identical hydrogen atoms.
We have already reported briefly on the synthesis and
crystal structure of a tetrahydrate of compound 3.[8] In the
present paper, we describe in detail the synthesis of the
starting material 1,3,5-pentanetriamine and we include the
spectroscopic data. A conformation analysis is given to help
the assignment of the NMR signals. Detailed conformation
interconversion paths are proposed, supported by the crystal
structure of one of the less populated conformers trapped as a
2:1 clathrate with 1,4-dioxane. A general mechanism for the
self-assembly reaction of amines with formaldehyde is also
proposed. A second full paper will follow that will also deal
with a 24-membered ring (Table 1, 6) prepared by the
condensation of 1,1-bis(2-aminoethyl)hydrazine with form-
aldehyde.[9, 10]
ˆ
ˆ
6CH2O ‡ 4NH3 > H2C N CH2 N CH2 ‡ 6H2O > C6H12N4 ‡ 6H2O (1)
The literature about this condensation is quite confusing, and
a careful review is needed, but is beyond the scope of this
paper. However, a few examples may illustrate the confusion:
Evans proposed 1,3,5-triazacyclohexane as the intermediate
in the condensation of ammonia with formaldehyde.[12]
Nielsen et al. claim to have proved the formation of this
precursor and of 1,3,5,7-tetraazabicyclo[3.3.1]nonane by
NMR spectroscopy,[13] but we have not been able to reproduce
their NMR results. The mechanism they propose for this
Results and Discussion
Synthesis:
1,3,5-Pentanetri-
amine (Scheme 1, 15) was pre-
pared by the method of Rüs-
sel.[11] Since Rüssel gave no
spectroscopic data, and a very
high purity of the triamine is
essential, we describe the syn-
thesis in detail. The compound
3,3-di(carboethoxy)-1,7-hep-
tanedinitrile 9 was prepared
from diethyl malonate and
acrylonitrile in 91% yield.
Long reaction times were need-
ed to hydrolyse the dinitrile 9 to
the corresponding tricarboxylic
acid 10. Esterification of the
tricarboxylic acid gave the best
yield when catalysed by sulfuric
acid. Triamine 15 was obtained
after conversion of trihydrazine
12 to the corresponding triazide
Scheme 1. The synthesis of 1,3,5-pentanetriamine.[11]
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999 0947-6539/99/0510-3057 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 10
3057

M. R. Suissa et al.
FULL PAPER
condensation seems too complicated for such a simple, fast
and very clean reaction.
Methanimine has been obtained by gas-phase elimination
of HCl from N-chloromethanamine or by flash vacuum
thermolysis of 2-azabicyclo[2.2.n]alkenes.[14] The NMR spec-
trum of methanimine was recorded at
958C, as the
compound decomposes to HMTA and to polymeric materials
above 808C.[14] The authors proposed that this decomposi-
tion occurs via 1,3,5-triazacyclohexane as reported by Nielsen
et al.[13] We have no reason to believe that methanimine
decomposes to HMTA via 1,3,5-triazacyclohexane and we
Scheme 3. Formation of four bonds gives the corresponding tetracyclic
compound directly. The two dimeric units have opposite configurations in
both cases.
ˆ
ˆ
think instead that the di-imine CH2 N CH2 N CH2 is
formed and that it dimerizes rapidly to HMTA.
As shown in Scheme 2, if HMTA is to be formed via 1,3,5-
triazacyclohexane, three formaldehyde molecules have to be
oriented in axial positions; this will cause unfavourable 1,3-
diaxial interactions. The product cannot then equilibrate
further with its corresponding imine (by loosing water) and
then react with the last ammonia molecule to finish the
building of the HMTA structure. Hydrated hemi-aminal
adducts can hardly be involved as important intermediates
in this reaction, since aldehyde-ammonia adducts are not
stable (depending on pH) and readily undergo dehydration
and polymerization. The simplest imaginable way to form
HMTA is to let two di-imine units of opposite conformations
mass of the corresponding di-imine with its fragments
(Table 1, 8), but there is no peak corresponding to tetrahy-
droimidazole. Repeating this reaction, we isolated a jelly-like
‡
‡
(g g and g g ) come together like in the folding of hands
(Scheme 3). The fit is perfect for the formation the four bonds
needed.
Scheme 4. Does condensation of aromatic aldehydes with ammonia form
HMTA derivatives?
Condensation of 1,2-ethanediamine with formaldehyde
gives
1,3,6,8-tetraazatricyclo[4.4.1.13,8]dodecane,[3, 4, 15, 16]
which is an analogue of HMTA (Table 1, 8). It can be
described as a tetraazacyclooctane with two transannular
ethylene bridges, while HMTA is a tetraazacyclohexane with
two transannular methylene bridges. Krässig proposed a
wrong structure for this molecule, assuming that tetrahydroi-
midazole was the direct precursor in this reaction.[3] Tetrahy-
droimidazole can be excluded, since its five-ring structure is
not retained in the tricyclo-structure of 8. Furthermore, the
mass spectrum of 8 contains mainly the molecular ion and the
polymer by distilling water from the reaction mixture. This
polymer isomerized quickly in boiling 1,4-dioxane to 8. Again,
the simplest way to form this multicyclic product is to assume
that two di-imine units react together as shown in Scheme 3.
The fact that condensation of aromatic aldehydes with
ammonia gives stable di-imine products supports the general
mechanism we propose.[17±19] Ahmad et al. claim to have
synthesized various HMTA derivatives (and their correspond-
ing di-imines) by replacing formaldehyde with aromatic
aldehydes, but they have not
proven their structures.[17] Be-
cause of severe steric problems,
as shown in Scheme 4, six mol-
ecules of the aldehydic substitu-
ent R can not be accommodat-
ed on the surface of HMTA,
which can be considered equiv-
alent to four cyclohexane sur-
faces. In each ring, one can
define three axially related
1,3,5-positions, of which only
one can accommodate an aro-
matic R group to avoid the
forbidden 1,3-interaction. The
other two R groups must be in
equatorial positions if this is to
be sterically tolerated (equato-
rial R group of one ring becom-
ing an axial R group in the
Scheme 2. The simplest way to form HMTA is from two di-imine units (b) and not from
1,3,5-triazacyclohexane (a).[12]
3058
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3058 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 10

Polycycles
3055±3065
other ring). This is possible only for two of the four cyclo-
hexane rings. The other two rings must then carry two axial R
groups in 1,3-positions and a third in an equatorial position,
which creates two forbidden 1,3-diaxial interactions. Thus,
only four of these six aromatic substituents can be accom-
modated on the surface of HMTA.
The reaction between 1,3-propanediamine and formalde-
hyde was first believed to lead directly to the favoured 16-
membered ring 2.[16] It is now clear that rapid condensation to
a polymer occurs. The polymer then dissolves in refluxing 1,4-
dioxane to give the multicyclic product in high yield, with
small amounts of 1-formyl-3-methyl-1,3-diazacyclohexane
and 3-oxa-1,5-diazabicyclo[3.3.1]nonane, and trace amounts
of 1,3-diazacyclohexane. Evans proposed a mechanism for this
reaction involving 1,3-diazacy-
of the oligo-imine units in the mass spectra of these molecules
supports our conclusion that the imine structure is retained in
the macrocyclic system and, hence, is the final precursor for its
formation.
Summing up, all of these compounds are formed in high
yields by condensation of amines with formaldehyde. As the
geometry of the oligo-imines is preserved in the multicyclic
structures, we propose them as the ultimate precursors in this
condensation. All the stages in the process are reversible; a
polymer and its parent oligo-imine are in equilibrium with the
corresponding multicycle. Reversible and irreversible by-
products are possible when the geometry of the intermediates
allows it, provided that the polymer formed is soluble enough
to participate in equilibrium processes. Scheme 5 shows the
clohexane as the intermedi-
ate.[12] This molecule cannot be
the final precursor in this reac-
tion, but rather an intermediate
to the corresponding di-imine.
The stable conformation of 1,3-
diazacyclohexane is the one
with diaxial N H, which has
repulsive interactions between
the two lone-pair orbitals. After
this has reacted with one form-
aldehyde molecule, it must re-
act further to the di-imine,
which can in turn either
polymerize or dimerize to
molecule 2.
In the case of compound 3,
we did not observe any polymer
formation, and the macrocycle
was formed in water after a
very short reaction time. When
we condensed 1,2-bis(2-amino-
ethyl)hydrazine with formalde-
Scheme 5. A proposed mechanism for the self-assembly of 1,3,5-pentanetriamine with formaldehyde.
hyde,[9] we obtained a soluble polymer that isomerized in
boiling 1,4-dioxane to give the corresponding tetramer
(Table 1, 6; see also Ref. [10]). A highly insoluble polymer
was formed in the case of the cage-like septicyclic molecule
5[7] (Table 1), and it isomerized to give the corresponding
macrocycle within a few hours in boiling 1,4-dioxane. Con-
densation of 1,1-bis(aminomethyl)cyclohexane with formal-
dehyde resulted in soluble polymers that isomerized in a few
minutes to give macrocycle 7 in boiling dichloromethane
(Table 1).[20]
The mass spectra of these molecules are summarized in
Table 1. Mass spectra of compounds 1, 5 and 8 show mainly
the molecular ions, while the mass spectra of the other
molecules display mainly the masses of the corresponding
imine and its fragments. The mass spectra of these three
compounds can be explained by the fact that the fragmenta-
tion of these noncage molecules into two halves requires four
C N bonds to be simultaneously broken in the case of
molecules 1 and 8, resulting in two bis-imine units. In the case
of molecule 5, eight C N bonds need to be broken. The other
five molecules produce four imines each. The predominance
self-assembly mechanism of 1,3,5-pentanetriamine with form-
aldehyde as an example. The intermediates in this case
exclude the possibility of the formation of irreversible and
reversible by-products that are favoured by entropy over time,
as was the case in the condensation of 1,3-propanediamine
with formaldehyde.[16]
The best solvent we found for the isomerization is 1,4-
dioxane. The less soluble the polymer, the higher the temper-
ature and the longer the reaction time needed to complete the
isomerization reaction. When no polymer formation is
observed, we believe that the reason for this may be high
solubility of the polymer, relatively high stability of the oligo-
imine, high thermodynamic stability of the macrocycle or a
combination of these factors.
Conformational analysis: Adopting the standard approach to
the conformation analysis of macrocyclic alkanes,[21] we
considered the possible diamond-lattice conformations for
the central 24-membered ring, assuming that each of the four
laterally fused trans-decalin-like moieties is already rigidly
fixed in a diamond-lattice pattern. We further assume that
Chem. Eur. J. 1999, 5, No. 10
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3059 $ 17.50+.50/0
3059

M. R. Suissa et al.
FULL PAPER
each of the four methylene
groups linking the trans-decalin
moieties can be attached either
equatorially or axially to the
nitrogen atoms. This results in
four possible conformations
[Figure 2 and Figure 5 (see lat-
er)]: conformer A with all eight
bonds of the four CH2 groups
attached equatorially (e) to the
nitrogens, B in which each CH2
group is attached equatorially
to one nitrogen and axially (a)
to the other in the sequence
e,a,e,a,e,a,e,a, C with one CH2
group attached equatorially to
one nitrogen and axially to the
other, while the second CH2
group is attached axially to
one nitrogen and equatorially
to the other, and so on alter-
Figure 2. The four conformers and two intermediates derived from the diamond lattice: the helical symmetry of
the intermediates K and L is seen clearly here.
nately
in
the
sequence
a,e,e,a,a,e,e,a, and D with eight
axial attachments. Any other combination will be incompat-
ible with the diamond lattice and will therefore introduce
strain. The relative stabilities of these conformers on an
enthalpy basis must be directly dependent on the relative
stabilities of equatorial and axial N-alkyl substituents in
azacyclohexanes. Assuming equatorial preference, we can
conclude that the stability order would be: A > B ˆ C > D.
The stability order on an entropy basis can roughly be
estimated from the molecular rotational symmetry, all four
conformers being achiral. Conformers A and D have D2d
symmetry (s ˆ 4), B has S4 symmetry (s ˆ 2) and C has C2v
symmetry (s ˆ 2). By increasing the temperature, the entropy
will favour the four conformers in the order: B ˆ C > A ˆ D.
If the enthalpy differences are neglected, a statistical mixture
of 17% A, 33% B, 33% C and 17% D is expected. In practice,
the choice of conformation is dominated by interaction with
the solvent, either by external solvation or by internal
clathrate or complex formation.
between the conformers. The ªverticalº projections of K
and L shown in Scheme 6 might suggest that these are also
strain-free and of the diamond-lattice types. However, they
have been derived from one turn of a strain-free helix by
vertical compression along the helix axis. Hence, they are
chiral like the parent helix (C2 symmetry) and somewhat
strained (Figure 2). The term ªquasi-diamond-lattice typeº is
proposed for them. Finally, six even more highly strained
intermediates lacking symmetry are needed to complete the
scheme shown in Scheme 6, with one intermediate (not
drawn) between each pair of ªarrowsº. Each of the six pairs
of ªarrowsº implies that changes have been made successively
at two diametrically[=] opposed ªcornersº (Scheme 7). The
population of these intermediates is assumed to be negligible
at any instant.
The precise mechanism of each interconversion step is
thought to involve a slow local nitrogen inversion followed by
a fast adjustment of the torsion angles, primarily in the
adjustment of macrocyclic C N bonds. This unit process is
repeated in successive steps so that all values for the energy
barriers are expected to have much the same height.
Conformational interconversion paths: Interconversion be-
tween B and A, between C and A or between C and B
requires four nitrogen inversion steps of relatively high
energy, while interconversion between A and D requires
eight nitrogen inversion steps. Each nitrogen inversion is
accompanied by appropriate adjustments of the torsion angles
in the eight C N bonds that connect the four unchanged trans-
decalin-like units. These adjustments occur over lower barriers.
The simplest and most direct interconversion paths con-
necting any pair of these conformers is through three
intermediate non-diamond-lattice conformers, as shown in
Scheme 6 and Scheme 7. The four conformers A, B, C and D
are of perfect diamond-lattice type and are expected to be the
most stable in solution. Their interconversion requires other
less stable conformers as intermediates. Two of these, K and
L, are of special interest, since both of them are common
intermediates of the three direct interconversion paths
NMR spectroscopy: The high-temperature 1H NMR spectrum
in CD2Cl2 of compound 3 is given in Figure 3 and consists of
nine resonances corresponding to the constitutional symme-
try. The absorptions are generally broad as this is an averaged
spectrum of all the conformers present. The molecule contains
a pair of protons b (A2). The equality of the chemical shifts of
these two protons is a result of the existence of a C2 symmetry
operator interconverting the two nuclei. Two absorptions
(AX) belong to protons a with 2Ja1,a2 ˆj 9 j . This value is due to
the neighboring nitrogens and reflects the conformational
rigidity of the trans-decalin units of the molecule. These two
[=] Choosing two adjacent ªcornersº results in intermediates lacking any
degree of symmetry.
3060
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3060 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 10

Polycycles
3055±3065
Scheme 6. Interconversion among the four low energy conformers.
Scheme 7. Detailed interconversion among the four possible conformers. A, B, C and D are strain-free diamond-lattice rings. K and L are chiral diamond-
lattice helices compressed along the helical axis. The others are unpopulated chiral intermediates. Because this 24-membered macrocycle consists of four
conformationally rigid bicyclic units, the total conformation may be described by a sequence of letters a and e, giving the mode of attachment of the corner
atom (equatorial or axial attachment) to the rigid bicyclic units.
Chem. Eur. J. 1999, 5, No. 10
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3061 $ 17.50+.50/0
3061

M. R. Suissa et al.
FULL PAPER
ly). On heating they should
merge to give five lines that
correspond to the chemical
symmetry. If only conformer A
is present, its 13C NMR spec-
trum should consist of five lines
and it should remain unaffected
at variable temperatures, as A
already has constitutional sym-
metry. No low-energy processes
should be observed in this case.
If A and D are present, the
high-temperature 13C NMR
spectrum should consist of five
lines (averaged spectrum) that
on cooling should split into 10
lines. The same is true for the
other possible combinations.
Figure 3. 1H NMR spectrum of compound 3 (500 MHz) in CD2Cl2 at 298 K (s ˆ solvent).
[D2]Dichloromethane or
a
mixture of freons and [D2]di-
chloromethane was used for
these low-temperature experi-
ments. The higher-temperature
13C NMR spectrum in CD2Cl2
consists of five lines as shown in
Figure 4. On cooling, the lines
broaden, followed by splitting
and then sharpening of the new
lines, to give the final slow-
exchange spectrum. In the sol-
vent mixture CD2Cl2:CCl3F
(1:1) at 160 K, the spectrum
consists of 22 lines: five lines
for A, eight for B and nine for C
in a ratio of 3:2:2 respectively.
In addition to these 22 lines, the
spectrum includes an extra set
of low intensity lines; some of
them are marked with an aster-
isk. This set of lines could be
ascribed to the fourth conform-
er D, but the species K and L of
low population are also possible
candidates, even though we do
Figure 4. 13C NMR spectrum of compound 3 (600 MHz) in CD2Cl2 at 250 K and in CD2Cl2:CCl3F (1:1) at 160 K
(s ˆ solvent).
signals at d ˆ 2.8 and 3.8 represent the inner and outer a
protons, or equatorial and axial protons, respectively. The
value of 2J for the c protons is lower than that for the d protons
not see as many lines as expected. The possibility of complex
formation between the different conformers and solvent
molecules seems to be excluded, since the complexes should
cause large shifts, and we did not observe such shifts. We also
expect that some of these complexes would lose their
symmetry and give rise to additional resonances.
Geminal site exchange of CH2 protons in compound 3 is
rendered impossible by the presence of one CH group in each
ªdecalinº unit. If these were all replaced by N, thus allowing
inversion, geminal exchange would become possible.[10]
2
2
as expected; the values are Jc1,c2 ˆj 11.9 j and Jd1,d2 ˆj 12.7j,
respectively.
The interconversion of the four possible conformers is more
easily studied by 13C NMR spectroscopy, since the spectra are
simple because of the high symmetries of these conformers.
The low-temperature 13C NMR spectrum for a mixture of the
four conformers A, B, C and D should consist of 27 lines: five
lines for each of the conformers A and D, Cb(1) Ca(1) Ce(1)
Cc(1) Cd(1) in the ratio of 1:2:1:2:2 respectively; eight lines for
conformer B, Cb(1) Ca(2) Ce(1) Cc(2) Cd(2), (ratio of 1:1:1:1:1:
1:1:1, respectively); and nine lines for conformer C, Cb(1)
Ca(2) Ce(2) Cc(2) Cd(2), (ratio of 2:2:2:1:1:2:2:2:2, respective-
Complexation: Molecule 3 is built of four trans-decalin-like
units, each containing three pairs of nonbonding electrons
from the nitrogens. Because all of its stable conformers can be
derived from the diamond lattice, the four decalin units in
3062
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3062 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 10

Polycycles
3055±3065
each of them are divided into two pairs, one pointing above
and one pointing below the central 24-membered ring, to
produce a ªdouble sandwichº (Figure 2).
one above the other. The central ring is much smaller in this
conformer than in A and B.
Although we have tried to crystallize all these four con-
formers by forming complexes or clathrates with other
molecules, nearly all of the crystals we have obtained
contained only the A conformer. Crystallization of molecule
3 from ethyl acetate gave in one case the A conformer as
tetrahydrate.[8] In another case, the A conformer crystallized
with a disordered ethyl acetate molecule in the cleft of
compound 3. Surprisingly, a crystalline 1,4-dioxane clathrate
of the K conformer (C2 intermediate) was formed (Figure 6).
The orientation of the electron pairs is dependent on the
symmetry of the different conformers, as shown in Figure 5.
Conformer A has two parallel rows of electron pairs (3 and 3)
above, and two below the central 24-membered ring. These
electrons lie on the edge of the cleft on either side, and each of
these two rows can form complexes with molecules that fit
into the cleft. In conformer B, the rows of electrons consist of
only two pairs each. Four pairs of electrons are oriented out
from the central ring and are available for interaction with
solvent molecules. The size of the central ring is, however,
identical to the one in A conformer. The central ring is
particularly long and narrow in conformer C, with six electron
pairs on the edges of one cleft (3 and 3) and two pairs on the
edges of the other cleft (1 and 1). As with conformer B, four
electron pairs are oriented away from the central ring.
Conformers B and C were always present in a ratio of 1:1 in
all the solvent mixtures that we used for the DNMR experi-
ments, while the relative population of A to (B ‡ C) was
solvent dependent. The fourth conformer, D, has only four
electron pairs on the edges of the clefts, two above (1 and 1)
and two below (1 and 1) the central ring. The remaining eight
electron pairs are oriented in the corners out from the central
ring and are divided into four groups. Each group, together
with the NCH2N corner unit to which it is connected, creates a
small helix of C2 symmetry, and the electron pairs are located
Figure 6. ORTEP plot of the C2 intermediate.
This intermediate is stabilized in the crystal structure by
van der Waals interactions between dioxane molecules locat-
ed in the center of the 24-membered ring and dioxane
molecules located between the macrocycles. This crystal
structure supports the interconversion pathway we propose
(Scheme 5 and 6) for this molecule.
Experimental Section
The 1H and 13C NMR spectra were recorded with Bruker DPX300, Bruker
DRX500, and DRX600 instruments. The temperatures were calibrated
with CD3OD. Temperatures below the freezing point of methanol were
determined by extrapolating with
a calibration graph. Standard 2D
experiments were used to confirm the assignments of H and C resonances.
The mass spectra under impact conditions (EI) were usually recorded at
70 eV ionization potential, and methane or ammonia was used for chemical
ionization (CI). Melting points are uncorrected.
Preparation of 3,3-di(carbethoxy)-1,7-heptanedinitrile (9): A stirred so-
lution of diethyl malonate (160 g, 1.0 mol) and tetraethylammonium
hydroxide pentahydrate (10 g, 0.06 mol) in 1,4-dioxane (400 mL) was
cooled to 108C. A solution of acrylonitrile (199 mL, 3.0 mol) in 1,4-dioxane
(50 mL) was added dropwise over 2 h, and the temperature was kept under
208C. The reaction was left overnight at room temperature. The mixture
was acidified with HCl (3m), was poured into cold water (1 L), and was
stirred well. After a short time, the compound crystallized and was pure
enough to continue to the next stage (243 g, 91% yield). It could be further
recrystallized from benzene. M.p. 638C; MS (CI, NH3): m/z (%): 284
‡
(100%) [M‡NH4] ; MS (CI): m/z (%): 266 (1), 221 (14), 154 (34), 108
1
(100), 69 (41), 53 (19), 41 (20), 29 (50); H NMR (500 MHz, CDCl3): d ˆ
Figure 5. The four stable conformers with their nonbonding electrons.
4.22 (q, J ˆ 7.1 Hz, 4H; CH2CH3), 2.42 (dt, J ˆ 7.7 Hz, 4H; H3,5), 2.21 (dt,
Chem. Eur. J. 1999, 5, No. 10
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3063 $ 17.50+.50/0
3063

M. R. Suissa et al.
FULL PAPER
J ˆ 7.7 Hz, 4H, H2,6), 1.25 (t, J ˆ 7.1 Hz, 4H; CH2CH3); 13C NMR
while warm. Compound 3 was obtained as a white solid on cooling (5.2 g,
100%). M.p. 2218C; MS (EI): m/z (%): 153 (31), 152 (100), 125 (7), 98 (35),
97 (20), 83 (16), 82 (16), 70 (22), 57 (27), 56 (32), 55 (27); 1H NMR
(500 MHz, CD2Cl2): d ˆ 3.91 (d, J ˆ 9.2 Hz, 1H; Ha1), 3.09 (s, 2H; Hb),
2.29 (ddd, J ˆ 11.8, 4.3, 2.9 Hz, 1H; Hc1), 2.41 (d, J ˆ 9.2 Hz, 1H; Ha1),
2.16 (dt, J ˆ 12.1, 12.1, 2.9, 1H; Hc2), 1.80 (ddt, J ˆ 11.0, 2.7 Hz, 1H; He),
1.61 (dddd, J ˆ 12.6, 12.6, 11.0, 4.4 Hz, 1H; Hd1), 1.36 (dddd, J ˆ 12.8,
2.6 Hz, 1H; Hd2); 13C NMR (500 MHz, CD2Cl2, 298 K): d ˆ 74.9 (Cb), 73.1
(Ca), 62.3 (Ce), 50.0 (Cc), 30.1 (Cd); 13C NMR (600 MHz, CCl3F:CD2Cl2
(ꢀ1:1), 160 K): d ˆ 77.3, 76.8 (D2d), 73.9, 73.7, 73.5, 72.9 (D2d), 72.1, 70.0 (Cb
and Ca 8 lines); 62.6, 62.0, 61.3 (D2d), 61.2 (Ce 8 lines); 51.4, 51.3 (D2d), 45.4,
44.4 (Cc 5 lines); 31.8, 31.3, 30.7 (D2d), 25.0, 24.7 (Cd 5 lines).
ˆ
(500 MHz, CDCl3): d ˆ 169.0 (C O), 118.5(CN), 62.4 (CH2CH3), 55.5
(C4), 29.5 (C2,6), 13.8 (CH2CH3), 13.0 (C3,5).
1,3,5-Pentanetricarboxylic acid (10): 3,3-Di(carbethoxy)-1,7-heptanedini-
trile (210 g, 0.80 mol) dissolved in HCl (6m, 500 mL) was refluxed and
stirred continuously for 3 days. The acid crystallized at low temperature
(ꢀ58C) as the monohydrate. Concentration of the mother liquor and
cooling resulted in a second crop of acid (140 g, 87% yield). M.p. 2088C
(dehydrates at 1148C); MS (CI): m/z (%): 186 (2), 140 (13), 114 (95), 86
1
(71), 55 (100), 42 (59); H NMR (500 MHz, D2O): d ˆ 2.31 (m, 1H; H3),
2.24 (m, 2H; H1,5), 1.70 (m, 2H; H2,4); 13C NMR (500 MHz, D2O ‡
ˆ
ˆ
CD3OD): d ˆ 180.4 (C3 C O), 178.5 (C1,5 C O), 44.7 (C3), 32.3 (C2,4),
27.4 (C1,5).
X-ray crystallography: X-ray data were collected on a Siemens SMART
CCD diffractometer,[22] with graphite monochromated MoKa radiation.
Data collection method: w-scan, range 0.68, crystal to detector distance
5 cm; further information is given in Table 2. Data reduction and cell
Triethyl-1,3,5-pentanetricarboxylate (11): 1,3,5-Pentanetricarboxylic acid
(70 g, 0.34 mol) in absolute ethanol (700 mL) and concentrated sulfuric
acid (50 mL) was refluxed and stirred continuously for 3 ± 4 days (the acid
dissolved gradually). The solvent was carefully distilled off, and cold water
was added to the residue. The ester was extracted with diethyl ether several
times. The ether phase was washed with sodium carbonate solution and
water, and then dried. Purification by vacuum distillation (b.p. 182 ± 1868C/
12 mmHg) gave 65 g (83% yield) of the ester. MS (CI, CH4): m/z (%): 289
(7), 243 (100), 214 (23), 185 (8), 169 (20), 155 (8), 141 (14), 123 (1), 114 (15),
99 (6), 71 (6), 55 (8); MS (EI): m/z (%): 288 (0.3), 243 (7), 214 (33), 185 (15),
169 (28), 155 (14), 141 (24), 123 (20), 114 (30), 99 (12), 71 (13), 55 (21), 41
(11), 29 (28), 28 (41), 18 (100); 1H NMR (500 MHz, CDCl3): d ˆ 4.04 (q, J ˆ
7.1 Hz, 4H; CH2CH3), 4.02 (q, J ˆ 7.1 Hz, 2H; CH2CH3), 2.33 (m, 1H; H3),
2.22 (m, 4H; H1,5) 1.82 (m, 2H; H2,4), 1.73 (m, 2H; H2,4), 1.14 (q, J ˆ
Table 2. Crystal data for compound 3.
empirical formula
Mw
C40H76N12O4
394.56
T [K]
150 (2)
l [Š]
0.71073
crystal system
space group
a [Š]
orthorhombic
C2/c
24.415(1)
b [Š]
6.461(1)
c [Š]
27.360(2)
4301.2(2)
4, 1.219
0.081
7.2 Hz, 9H, CH2CH3); 13C NMR (500 MHz, CD2Cl2): d ˆ 174.6 (C3 C O),
ˆ
V [Š3]
ˆ
172.6 (C5 C O), 60.3 (CH2CH3), 60.2 (CH2CH3), 43.7 (C3), 31.6 (C1,5),
3
Z, 1calcd [Mgm
m(MoKa) [mm
F(000)
]
26.9 (C2,4), 14.1 (CH2CH3), 14.0 (CH2CH3).
1
]
1,3,5-Triaminopentane (15): Triethyl-1,3,5-pentanetricarboxylate (115.2 g,
0.40 mol) dissolved in absolute ethanol (400 mL) was stirred under N2
atmosphere at room temperature. Anhydrous hydrazine (51.6 g, 1.6 mol),
carefully dissolved in absolute ethanol (100 mL), was added dropwise to
the triester solution. The mixture was refluxed for 24 h and then cooled.
The product was filtered off, dissolved in HCl solution (6m, 250 mL), and
cooled to 108C, before diethyl ether (250 mL) was added. The temper-
ature of the mixture was stabilized at 08C, and sodium nitrite (81.8 g,
1.2 mol) dissolved in water (150 mL) was added dropwise. After stirring for
3 h, the mixture was separated, and the water phase was extracted with
diethyl ether (3 Â 100 mL). CaCl2 was added to the combined ether phases,
and the mixture was left in the refrigerator (ꢀ58C) overnight. The ether
phase was decanted, and the same volume of absolute ethanol was added to
it. This mixture was refluxed and stirred continuously for 48 h until no more
nitrogen evolved. The solvents were carefully evaporated, and the residue
(ꢀ76 g) was dissolved in HCl solution (3m, 500 mL) and refluxed and
stirred continuously for 12 hours. Water was carefully evaporated (in small
portions in a one-liter flask), and the crude triammonium chloride (62 g)
was dissolved in absolute ethanol (200 mL) with an equivalent amount of
sodium ethoxide [Na (19.3 g, 0.84 mol) in ethanol] and was refluxed for
2 hours. NaCl was filtered off, and the ethanol was evaporated carefully in
small portions. The crude triamine (30 g, 64% yield) was further purified by
vacuum distillation (0.01 mmHg) at 84 ± 1008C (bath temperature was
2058C).
1728
crystal size [mm]
q range for data collection [8]
limiting indices
reflections collected
refinement method
data / restraints / parameters
goodness-of-fit on F 2
final R indices [I > 2s(I)]
0.45 Â 0.2 Â 0.12
1.49 to 30.51
34 ꢁ h ꢁ 34, 8 ꢁ k ꢁ 9, 39 ꢁ l ꢁ 37
unique 22640 / 6505 [R(int) ˆ 0.045]
full-matrix least-squares on F 2
5330 / 0 / 405
1.204
R1 ˆ 0.063, wR2 ˆ 0.128
R1 ˆ 0.095, wR2 ˆ 0.173
0.367 / 0.190
R indices (all data)
3
largest D1 [eŠ
]
determination were carried out with the SAINT and XPREP programs.[22]
Absorption corrections were applied by the use of the SADABS
program.[23] The structure was determined and refined using the SHELXTL
program package.[24] The non-hydrogen atoms were refined with aniso-
tropic thermal parameters; hydrogen positions were calculated from
geometrical criteria and refined with isotropic thermal parameters.
Crystallographic data (excluding structure factors) for the structure
reported in this paper have been deposited with the Cambridge Crystallo-
graphic Data Centre as supplementary publication no. CCDC-102850.
Copies of the data can be obtained free of charge on application to CCDC,
12, Union Road, Cambridge CB2 1EZ, UK (fax: (‡44)1223-336-033;
e-mail: deposit@ccdc.cam.ac.uk)
Trihydrazide (12): MS (CI, CH4): m/z (%): 246 (0.5), 215 (9), 198 (6), 183
(100), 166 (6), 155 (20), 141 (6), 55 (11); MS (EI): m/z (%): 215 (12), 183
(100), 155 (47), 141 (10), 113 (7), 69 (8), 55 (30), 41 (12), 32 (24), 31 (14);
1H NMR (500 MHz, CF3CO2D, d ˆ 11.50): d ˆ 4.59 (m, J ˆ 4.6 Hz, 1H;
H3), 4.42 (t, J ˆ 7.4 Hz, 4H), 3.97 (m, J ˆ 7.4 Hz, 2H; H2,4), 3.87 (m, J ˆ
7.42 Hz; H2,4); 13C NMR (500 MHz, CF3CO2D, d ˆ 164.2): d ˆ 178.2 (C3
Acknowledgments
We thank MR-Senter (Norway) and Dr. Tore Skjetne, in Trondheim, for
allowing us to use the 600 MHz NMR instrument. Special thanks are given
to Odd Inge Optun, NTNU (Norway) for assistance in using this instru-
ment.
ˆ
ˆ
C O), 176.6 (C1,5 C O), 45.5 (C3), 32.9, (C1,5), 29.3 (C2,4).
Triamine 15: MS (CI, CH4): m/z (%): 118 (100), 83 (6), 71 (11); MS (EI):
m/z (%): 83 (22), 71 (40), 70 (13), 56 (14), 44 (100), 42 (12), 30 (97);
1H NMR (500 MHz, CDCl3): d ˆ 2.64 (h, J ˆ 8.3, 4.6 Hz, 1H, H3), 2.51 (m,
4H; H1,5), 1.28 (m, 2H, H2,4), 1.12 (m, 2H; H2,4), 0.89 (brs, 6NH2);
13C NMR (500 MHz, CDCl3): d ˆ 47.1 (C3), 41.6 (C1,5), 38.9 (C2,4).
[1] A. Butlerow, Ann. 1860, 115, 322.
Molecule 3: Aqueous formalin (8.80 mL, 0.102 mol, 37%) diluted with
water (10 mL) was added dropwise to a stirred ice-cold solution of 1,3,5-
triaminopentane (4.00 g, 0.034 mol) in water (25 mL). After 30 min, a white
precipitate formed. The mixture was heated slightly (ꢀ708C) and filtered
[2] D. Philp, J. F. Stoddart, Angew. Chem. 1996, 108, 1242; Angew. Chem.
Int. Ed. Engl. 1996, 35, 1154.
[3] H. Krässig, Makromol. Chem. 1955, 17, 77.
[4] P. Murray-Rust, Acta Crystallogr. Sect. B 1974, 31, 583.
3064
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3064 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 10

Polycycles
3055±3065
[5] J. Dale, C. Rùmming, T. Sigvartsen, Acta Chem. Scand. 1991, 45, 1071.
[6] J. Dale, F. Rise, C. Rùmming, T. Sigvartsen, Acta Chem. Scand. 1992,
46, 1200.
[7] J. Dale, C. Rùmming, M. R. Suissa, J. Chem. Soc. Chem. Commun.
1993, 1283.
[15] J. Hocker, D. J. Wendish, J. Chem. Res. 1977, 236.
[16] J. Dale, C. Rùmming, T. Sigvartsen, Acta Chem. Scand. 1991, 45, 1064.
[17] A. Kamal, A. Ahmad, A. Ali Qureshi, Tetrahedron 1963, 869.
[18] Y. Ogata, A. Kawasaki, N. Okumura, J. Org. Chem. 1964, 29, 1985.
[19] T. L. Crowell, R. K. McLeod, J. Org. Chem. 1967, 32, 4030.
[20] M. R. Suissa, C. Rùmming, J. Dale, Chem. Commun. 1997, 113.
[21] J. Dale, Top. Stereochem. 1976, 9, 199.
[8] J. Dale, C. Rùmming, M. R. Suissa, J. Chem. Soc. Chem. Commun.
1995, 1631.
[9] J. Dale, C. Rùmming, M. R. Suissa, J. Chem. Soc. Chem. Commun.
1995, 1633.
[10] M. R. Suissa, C. Rùmming, J. Dale, unpublished results.
[11] H. Rüssel, Chem. Ber. 1956, 89, 679.
[12] R. F. Evans, Aust. J. Chem. 1967, 20, 1643.
[13] A. T. Nielsen, D. W. Moore, M. D. Ogan, R. L. Atkins, J. Org. Chem.
1979, 44, 1678.
[22] SMARTand SAINTArea-detector Control and Integration Software,
Siemens Analytical X-ray Instruments Madison, Wisconsin (USA).
[23] G. M. Sheldrick. private communication, 1996.
[24] G. M. Sheldrick, SHELXTL, version 5. Siemens Analytical X-ray
Madison, Wisconsin (USA).
[14] B. Brailion, M. C. Lasne, J. L. Ripoll, Nouv. J. Chim. 1982, 6, 121.
Received: January 5, 1999 [F1527]
Chem. Eur. J. 1999, 5, No. 10
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0510-3065 $ 17.50+.50/0
3065