Bisamides derived from azulene-1,3- and -5,7-dicarboxylic acids as new building blocks for anion receptors

Article abstract of DOI:10.1002/chem.200700592

Bisamides based on the azulene moiety were investigated as building blocks for anion receptors. In the course of these studies, derivatives of azulene-1,3- and -5,7-dicarboxylic acid were synthesized and thoroughly characterized. The anion affinities of t

Full text of DOI:10.1002/chem.200700592

DOI: 10.1002/chem.200700592
Bisamides Derived from Azulene-1,3- and -5,7-dicarboxylic Acids as New
Building Blocks for Anion Receptors
Tomasz Zielin´ski,[a] Mariusz Ke¸dziorek,[a] and Janusz Jurczak*[a,b]
Dedicated to Professor Marek Chmielewski on the occasion of his 65th birthday
Abstract: Bisamides based on the azu-
lene moiety were investigated as build-
ing blocks for anion receptors. In the
course of these studies, derivatives of
azulene-1,3- and -5,7-dicarboxylic acid
were synthesized and thoroughly char-
acterized. The anion affinities of the
derivatives based on functionalization
in the five-membered ring and in the
seven-membered ring were determined
by 1H NMR titration. The structural
analysis of these building blocks was
performed by X-ray diffractometry,
molecular modelling and 2D NMR
spectroscopy. The five-membered ring
derivatives are easy to obtain, offer a
binding site preorganized in the syn–
syn conformation and bind anions with
a strength similar to those of pyrrole-
based analogues. There is also strong
evidence for aromatic CH···anion inter-
actions. The ligands substituted at the
5- and 7-positions offer a binding cleft
with an uncommon geometry that orig-
inates from the seven-membered ring
and seems to be complementary to the
chloride anion.
Keywords: amides · anion recogni-
tion · azulene · ligand design ·
thioamides
Introduction
in a very precise manner within the host structure.[6] Howev-
er, our knowledge of structure/affinity relationships is still
limited[7] and there are many examples of how small changes
in ligand constitution can drastically alter the binding prop-
erties.[8,9] For this reason, it is important to have a large pool
of structural subunits—building blocks—that can help in
fine-tuning the host–guest interactions.
Immense progress in anion coordination chemistry has been
achieved during the last twenty years.[1] The importance of
anions in biological systems, the environment and medicine
is the main driving force for research in this area of supra-
molecular chemistry.[2] Artificial neutral receptors became
attractive targets of studies because of their analogy to natu-
ral systems[3] and because their selectivities were higher than
in the case of charged ligands. In neutral hosts, anion bind-
ing is usually accomplished through hydrogen bonds.[4,5] As
a single hydrogen bond is typically weak, strong binding can
only be achieved by multiple interactions of this type. In
order to achieve perfect complementarity with a target
guest, it is necessary to arrange the hydrogen bond donors
Amide groups are one of the most popular hydrogen
bond donors used in anion recognition[5] and are often incor-
porated into ligand structures in the form of aromatic bisa-
mides: namely isophthalamides (type 1),[10] dipicolinic bisa-
mides (type 2)[11] and pyrrole derivatives (type 3)[12]
(Scheme 1). Not only do these building blocks introduce the
binding sites into a ligand, but they can also be used as or-
ganizing elements that determine the shape of a host.[8,13]
We decided to extend the family of known aromatic
building blocks to derivatives of azulene, with its seven-
membered ring geometry. We thus prepared and studied the
bisamide and bisthioamide of azulene-5,7-dicarboxylic acid
(5, 6) as reported in our previous paper.[14] Azulene is an in-
teresting subunit for the construction of anion receptors,[15]
not only because of the uncommon geometry of its seven-
membered ring. Azulene is a strong chromophore and can
be regarded as a combination of a cyclopentadienyl anion
and a tropylium cation, which results in its large dipole
moment of 0.8 D.[16] Azulene-based bisamides therefore rep-
´
[a] Dr. T. Zielinski, Dr. M. Ke¸dziorek, Prof. J. Jurczak
Institute of Organic Chemistry
Polish Academy of Sciences
Kasprzaka 44/52, 01-224 Warsaw (Poland)
Fax : (+48)2263-26681
E-mail: jurczak@icho.edu.pl
[b] Prof. J. Jurczak
Department of Chemistry, Warsaw University, Pasteura 1
02-093Warsaw (Poland)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
838
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 838 – 846

FULL PAPER
ceeded in preparation of the desired amide 5a only by per-
forming the reaction in neat butylamine. Because of the
lower nucleophilicity of aniline, we were unable to prepare
the aromatic amide (5b, R=Ph) in the same manner. To
solve this problem, we hydrolysed the ester 11 into the
diacid and tried methods known from peptide bond chemis-
try. Neither the mixed-anhydride method nor the carbodii-
mide method led to the phenyl amide 5b, however (in both
cases we observed only formation of the activated inter-
mediate).
It seems that the chemistry of 5,7-dicarboxylic acid deriva-
tives of azulene might hamper the introduction of this build-
ing block into more sophisticated systems. Only aliphatic
amides can be easily prepared, which constrains the poten-
tial ligand structures; moreover, the popular method of mac-
rocyclization—aminolysis of esters—cannot be used for the
preparation of seven-membered ring derivatives.
Scheme 1. Aromatic building blocks for anion receptors.
The five-membered ring dicarboxylic acid derivatives
were prepared by direct acylation of azulene (12) with tri-
fluoroacetic acid anhydride, followed by hydrolysis to afford
the acid 14 (Scheme 2).[18] The acid 14 was converted into
the acid dichloride 15 and subjected to treatment with
amines, yielding the amides 7. The availability of the acid di-
chloride 15 is a great asset for the five-membered deriva-
tives, since it offers a convenient synthetic route to more
elaborate ligands.
From our previous experience with thioamide-based li-
gands,[19] we knew that thioamide groups “sensitize” recep-
tors to the presence of anions. In order to obtain optical sen-
sors, we decided to convert the butylamides 5a and 7a into
their thioamide analogues 6a and 8a, respectively, using
Lawessonꢁs reagent.[20] This transformation was performed
in boiling THF and gave the thioamides in good yields.
Having prepared the model ligands, we set out to deter-
mine their anion-binding properties. All azulene derivatives
are colourful, so it was interesting to see whether the azu-
lene-based bisamides would change their optical properties
upon anion complexation. Unfortunately though, interaction
of receptors 5a and 7a with anions does not lead to signifi-
resent the attractive fusion of binding site with signalling
subunit. Moreover, we expected an enhancement of anion
binding, owing to interaction of the anionic guest with the
azulene dipole or through the formation of an additional hy-
drogen bond with 6-CH. However, our preliminary results
showed limitations of these advantageous interactions be-
tween anions and azulene skeleton.[14] To check how the dif-
ferent geometries and electronic structures of the five- and
seven-membered rings would influence the anion binding,
we performed a comparison of azulene derivatives based on
functionalization in the seven-membered and in the five-
membered rings. Here we present our studies on azulene-
1,3- and -5,7-dicarboxylic acid bisamides (types 5 and 7) as
building blocks for anion receptors. In order to help readers
to distinguish between the two isomers, we will use the
terms “five-membered” and “seven-membered” derivative/
compound etc. as abbreviations for statements such as “de-
rivative of azulene-1,3-dicarboxylic acid”, as we do not
report any macrocyclic structures in this work; we hope that
such descriptions will not be misleading.
Results and Discussion
The azulene substituted with
carboxyl groups in the seven-
membered ring was obtained by
skeleton synthesis according to
literature
procedures
(Scheme 2).[17] The availability
of the diester 11 was very
promising, as one of the most
convenient methods for macro-
cyclization involved aminolysis
of diesters with a,w-diamines.
Unfortunately though, the ester
11 was unreactive under stan-
Scheme 2. Conditions: a) quinoline, 1008C; b) BuNH2, neat, 608C; c) Lawessonꢁs reagent, THF, reflux;
d) (CF3CO)2O, neat, 08C!RT; e) 1) KOH, EtOH, H2O, 60 8C; 2) HCl; f) SOCl2, CH2Cl2, reflux; g) RNH2,
dard conditions, and we suc- CH2Cl2, RT; h) Lawessonꢁs reagent, THF, reflux.
Chem. Eur. J. 2008, 14, 838 – 846
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
839

J. Jurczak et al.
cant colour changes, even in noncompetitive solvent such as
CH2Cl2. Introduction of an additional chromophore—the
nitro group in 7d—also failed to trigger spectacular colour
changes. However, the hues of CH2Cl2 solutions of the thio-
amide-based ligands 6a and 8a are slightly modified by the
presence of anions (Figure 1).
colour changes. Nevertheless, these results show that simple
acyclic bisamides based on azulene cannot act as optical sen-
sors for anions.
We then embarked on quantitative determination of the
receptorsꢁ affinities towards anions. As our studies of bind-
ing properties were intended as a reference for further gen-
erations of azulene-based receptors, we decided to deter-
mine the binding constants by 1H NMR titration in
DMSO+0.5% H2O. We were aware that our simple ligands
would interact weakly with anions in this very competitive
solvent. However, the recently published anion receptors
bind anionic guests efficiently in such demanding media as
DMSO, methanol or even water,[22] so we chose DMSO to
allow further comparison with more elaborate systems.
Upon titration with anions, the signals assigned to amide
or thioamide NH protons underwent downfield shifts indi-
cating complex formation that was reversible on the NMR
timescale. The lack of disappearance or broadening of NH
signals excluded deprotonation, so the main process that
takes place in solution is complexation. The changes in
chemical shift allowed us to calculate the values of the bind-
ing constants (Table 1), and the data gave consistent fit with
a 1:1 model, as was confirmed by Job plots (see Figure S2,
Supporting Information).
Figure 1. Colour changes of ligands in CH2Cl2 in the presence of anions
(3equiv of TBA salts).
Table 1. Binding constants [mꢀ1] for the formation of 1:1 complexes of model ligands with various anions in
[D6]DMSO+0.5% H2O or acetone at 298 K.[a]
In view of the facts that no
such colour changes were ob-
servable in more demanding
media such as DMSO and that
the optical response was the
most pronounced for the highly
basic fluoride anion, it was
likely that we were observing
an effect of deprotonation of
thioamide groups rather than
complexation. To check this as-
sumption we performed experi-
Anion
Solv.
4a[b]
4b[b]
5a
6a
7a
7b
7d
8a
Clꢀ
D
D
D
D
A
A
1.7
–
49
150
3.8
–
80
20327
6.1
–
1346
104
500
230
13
–
1.2
–
105
9.3
–
17
–
2.0
–
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
Brꢀ
ꢀ
PhCO2ꢀ
H2PO4
Clꢀ
27
73496
550
17
1400
82
[d]
46500
1480
240
185
1080
900
–
–
1620
520
Brꢀ
[d]
[a] Determined by 1H NMR titration. Errors estimated to be <10%. TBA salts were used as the source of
anions. Solvent: D=[D6]DMSO+0.5% H2O, A =acetone. [b] Ref. [19] [c] Interaction too weak to be mea-
sured. [d] Insoluble in acetone.
ments similar to those reported by Fabbrizzi[21] for urea-/thi-
ourea-based systems, comparing UV/Vis spectra of ligands
upon addition of TBA salts and TBA hydroxide (see Fig-
ure S1, Supporting Information). However, these UV/Vis
measurements did not give the final answer for our ligands.
Unlike in the case of urea-based receptors,[21] the presence
of the TBA hydroxide did not drastically change the ligandsꢁ
spectra, and no new band was developed. Since the spectra
of the free ligands and those after addition of TBA salts and
TBA hydroxide are quite similar, it is impossible to identify
the deprotonated forms of ligands and to distinguish un-
equivocally between complexation and deprotonation. How-
ever, keeping in mind the facts that i) more acidic thioa-
mides respond better than amides, ii) the effect takes place
for basic anions in noncompetitive, nonpolar CH2Cl2, and
iii) the spectral changes are moderate, it may be assumed
that anion complexation is accompanied by a small degree
of deprotonation, which is responsible for the observed
During titration, we also observed the downfield shift of
the signal of the “middle” CH proton (6-CH for the seven-
membered derivatives 5 and 6 or 2-CH for the five-mem-
bered derivatives 7 and 8). This might in some cases indicate
the presence of an advantageous CH···anion interaction
helping to bind the anionic guest,[23,24] although in the case
of the seven-membered derivatives (5a and 6a) this hypoth-
esis should be dismissed. The obtained values of Ddmax for 6-
CH are small (about 0.2 ppm) and comparable to those for
the 4,8-CH protons, which are unlikely to participate in the
anion binding. In contrast, the formation of a hydrogen
bond between 2-CH and anion in the five-membered deriva-
tives (7 and 8) is highly likely. The values of Ddmax for 2-CH
signal are about 1 ppm and much higher than those ob-
served for the other protons (see Table S2, Supporting Infor-
mation).
The determined binding constants are listed in Table 1;
for comparison we also present values obtained for the pyr-
840
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 838 – 846

Azulene-Based Anion Receptors
FULL PAPER
role-based bisamides 4a (R1 =nBu, R2 =H) and 4b (R1 =
Ph, R2 =H),[19] which possess an additional hydrogen bond
donor—pyrrole NH—but are poorly preorganized in the un-
favourable anti–anti conformation. The 3,4-diphenylpyrrole
derivatives 3 reported by Gale[12] interact with anions more
strongly than the unsubstituted ligands 4; as this fact cannot
be clearly explained at this moment, we chose ligands 4 for
comparison, as they are structurally more similar to com-
pounds 5–8. The azulene-based ligands show a typical pref-
erence for oxo anions over halides, and the interaction with
bromide anion was too weak to be measured in all the cases
studied. The amide 5a substituted in the seven-membered
ring binds the benzoate only twice as strongly as the chlo-
ride anion, so a change in relative selectivity is observed.
This may indicate that the geometry of its binding site is
better suited for the chloride anion than those based on a
five-membered ring as in 4 or 7.
values of Ka in acetone and DMSO. Because of the lower
charge density of bromide anion, the complexes with this
anion are less stable than those with chloride. The seven-
membered ligands 5a and 6a bind chloride more strongly
than their five-membered ring analogues (7a, 8a), which is
in agreement with the results obtained in DMSO. The con-
version of amide groups into thioamides increases affinity
towards both halides, but the selectivity for chloride over
bromide is improved. The binding constant for complexation
of chloride with 6a in acetone is remarkably high, but we
cannot explain why the interaction of the ligand 6a with
chloride is so efficiently enhanced in this medium. The be-
haviour of the aromatic protons in positions 2 and 6 was
similar in both solvents, while the values of Ka and Ddmax for
2-CH in 7 and 8 are comparable with those for NH signals,
so the possibility of CH···anion interaction is also quite high
in acetone.
The neighbourhood of the formal cyclopentadienyl anion
in the five-membered derivatives 7 does not obstruct inter-
action with anions. In contrast, the amide 7a binds oxo
anions more strongly than its seven-membered analogue 5a.
The five-membered phenyl amide 7b is an even better re-
ceptor than the pyrrole-based ligand 4b (R1 =Ph, R2 =H),
even though the latter compound possesses an additional
hydrogen bond donor in the form of the pyrrole NH, though
it has to be mentioned that, in the case of butylamides 7a
and 4a (R1 =nBu, R2 =H), the pyrrole-based ligand 4a
binds the anion more strongly than the azulene-based 7a.
These results support the thesis of the presence of the ad-
vantageous 2-CH···anion interaction in the five-membered
derivatives. As expected, the amide 7d derived from p-nitro-
aniline is the best receptor studied, due to the higher acidity
of its NH protons.
Conversion of amides into thioamides is expected to
boost anion binding[25] as a result of the more acidic charac-
ter of the latter compound class.[26] However, this beneficial
effect can be counterbalanced by undesirable structural
preferences of the thioamide groups.[19] In the case of the
azulene-based ligands studied here, transformation of the
amide groups into the thioamide groups has an ambiguous
character (Table 1). Replacement of the carbonyl group
with a thiocarbonyl moiety increased the values of the bind-
ing constants for the seven-membered ligand 6a, whereas it
did not significantly influence the interactions with oxo
anions in the case of 8a.
Although the values of Ka obtained for interaction with
chloride anion in DSMO are small, they are reliable and re-
producible, and since they were measured in the same con-
ditions, they can be used for comparison. However, the low
values of Ka mean that under typical conditions (concentra-
tions about 10ꢀ2 m) dissociation predominates over complex-
ation. We thus decided to measure the affinity towards hal-
ides in a less demanding solvent, acetone, that is structurally
similar to DMSO and still a more competitive medium than
acetonitrile or CH2Cl2. Using acetone as the solvent, we
were able to determine binding constants for both bromide
and chloride (Table 1). We observed similar trends in the
In advanced receptors, guest binding is achieved through
multiple interactions that originate from various binding
sites incorporated into the host structure. The observed se-
lectivity and binding strength cannot be ascribed as addition
or multiplication of binding constants determined for the
building blocks used for the ligand construction. However,
the geometrical parameters and conformational preferences
of the structural subunits determine the host structure and
arrangement of its binding sites. As a result, the structural
properties of building blocks are crucial for rational receptor
design and more important than their binding properties.
For this reason, we performed a comprehensive structural
analysis of azulene-based building blocks using different
methods of structural analysis.
We succeeded in preparing diffraction-grade crystals of
the three amide-based ligands 5a, 7a and 7b; the structure
of the seven-membered ligand 5a had been described previ-
ously[14] and is very similar to that of its five-membered ana-
logue 7a (Figure 2a, b; for colour versions of all enclosed
figures see the Supporting Information file). In both cases,
the ligand molecules adopt a syn–syn conformation, and the
carbonyl groups deviate from the plane of the azulene and
point in opposite directions (the torsion angles are between
21 and 368). To describe amide groupsꢁ orientations we use
syn and anti descriptors: the syn orientation denotes a con-
formation in which the carbonyl oxygen is close to the aro-
matic backbone and the amide NH is directed “outside”,
while in the anti conformation the positions of nitrogen and
oxygen are reversed (in other words, in a syn-syn conforma-
tion, the amide hydrogens point in the same direction as the
“middle” CH (6-CH in 5 or 2-CH in 7)). The structures of
5a and 7a showed the amide groups engaged in hydrogen
bonds with neighbouring molecules in such a manner that
two lines of hydrogen bonds are formed (the N–O distances
are about 2.9 Š; Figure 2).
The crystal structure of the phenyl amide 7b also resem-
bles those described above (Figure 2c). The amide groups
are in the syn–syn orientation, thus they are tilted and point
in opposite directions so that the carbonyl groups are locat-
ed on the different sides of the azulene ring. Two lines of
Chem. Eur. J. 2008, 14, 838 – 846
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
841

J. Jurczak et al.
preference for such a conforma-
tion is a great asset in a build-
ing block, as its binding site is
already preorganized for inter-
action with an anionic guest. To
determine the ligand structure
in solution, we measured the
2D NOSY spectra of com-
pounds 5a and 7a in DMSO
(see Figure S3, Supporting In-
formation). In both cases, the
NOE effect indicates that the
carbonyl groups adopt both syn
and anti orientations (NOE ef-
fects between the amide NH
and 4-CH and 6-CH protons
for ligand 5a, and between NH
and 2-CH and 4-CH for 7a).
This means that the amide
groups are either in the syn-anti
conformation or, more likely,
they can switch in solution be-
tween syn and anti orientations.
Figure 2. Different views (top, bottom) of crystal structures of model ligands showing the arrangements of mol-
ecules engaged in hydrogen bond networks (the butyl and phenyl groups are omitted for clarity). a) 5a, b) 7a,
c) 7b.
A further insight into the conformation preferences of li-
gands came from molecular modelling. We performed DFT
B3LYP calculations for the bismethyl amides (5c and 7c,
R=Me) and thioamides (6c and 8c, R=Me) in the gas
phase to optimize the conformer structures and the energy
distributions (the methyl bisamides were chosen in order to
simplify calculations). Since the carbonyl groups deviate
from the azulene plane, it is necessary to consider whether
they occupy the same or opposite sides of the ring. We de-
noted the former orientation “++” and the latter “+ꢀ”.
Both orientations have distinct energies. In all cases, the
“+ꢀ” conformers possess lower energy, probably due to the
opposite dipole orientations of the carbonyl groups
(Table 2).
For the seven-membered amide 5c, the syn–anti confor-
mations have the lowest energy, followed by the syn–
syn(+ꢀ) and then the syn–syn(++), which has an energy
value close to those of the anti–anti conformations. In the
case of its thioamide analogue 6c, the preferred conforma-
tion is also syn–anti(+ꢀ), but the syn–syn(++) conforma-
tion (the one with convergent binding sites) has an energy
higher than that of the anti–anti(+ꢀ) form. The five-mem-
Figure 3. Crystal structure of 7b showing water molecule engaged in hy-
drogen bonds.
hydrogen bonds are formed, linking adjoining molecules
(the N–O distance is 2.9 Š). The main difference in this
structure is the presence of water in the independent part.
The water molecule is positioned in such a manner that the
oxygen is in the azulene plane and the hydrogens are almost
perpendicular to the ring (Figure 3). These hydrogen atoms
form hydrogen bonds with the carbonyl groups, bridging
every third molecule of the ligand (the O–O distances are
about 2.9 Š). Although the three aromatic CHs (o-CH in
phenyl arms and 2-CH of azulene) point towards the water
oxygen atoms, the C–O distances are longer than 3.9 Š,
which dismisses any possibility of CH–anion interactions.
It is worth emphasizing that all three structures are very
similar. Firstly, the azulene subunits of neighbouring mole-
cules are alternately placed, probably to compensate their
dipole moments. Secondly, two lines of hydrogen bonds are
formed between the amide groups. Finally, the amide groups
are in the syn-syn orientation. The last feature is important
from the point of view of anion binding.
Table 2. Relative energies [kJmolꢀ1] for the conformers of the amides 5c
and 7c and the thioamides 6c and 8c.[a]
5c
6c
7c
8c
Eanti–anti+ꢀ
Eanti–anti++
Esyn–anti+ꢀ
Esyn–anti++
Esyn–syn+ꢀ
Esyn–syn++
8.38.1
10.4
0
1.5
3.7
22.2
13.6
0
4.36.2 .9
3.0
10.0
9.2
24.8
6.1
3
12.7
1.5
0
0.8
0
3.3
7.7
To bind anions with convergent hydrogen bonds, the aro-
matic bisamides must adopt the syn-syn conformation. A
[a] Calculations in the gas phase with use of DFT B3LYP/6-311+G
2pd)//B3LYP/6-31+G(d,p) method and basis sets.
ACHTREUNG
AHCTREUNG
842
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 838 – 846

Azulene-Based Anion Receptors
FULL PAPER
bered derivatives—amide 7c and the thioamide 8c—prefer
the advantageous syn-syn conformations and in the case of
the amide 7c, the difference between syn–syn and anti–anti
orientations is greater than 20 kJmolꢀ1. On the basis of
these results, it can be assumed that the stronger anion bind-
ing by the five-membered derivatives (type 7) originates
from their better preorganization in the syn–syn conforma-
tion. However, it has to be remembered that, in solution, we
observed both syn and anti orientations of the amide groups
for both isomers.
Analysis of the distributions of formal charges explains
why the seven-membered derivatives do not take advantage
of the azulene dipole moment. Analysis of Mulliken charges
reveals that the introduction of two carbonyl groups into the
5- and 7-positions in compound 5c induces the formation of
a negative charge on 6-C (Table 3). As a result, the charges
bonds formed by arenes with electron-withdrawing substitu-
ents,[23] which may suggest the presence of the CH···anion in-
teraction for the ligands of type 7. These results correlate
1
with those of the H NMR titration experiments: the 2-CH
proton signals of 7a and 8a were shifted more strongly
downfield than the 6-CH of the seven-membered analogues
5a and 6a. At the same time, the calculated lengths of the
NH hydrogen bonds to chloride are smaller for the seven-
membered 5c and 6c than for 7c and 8c (Table 4). There is
also a parallel to the titration experiments, as this anion was
bound more strongly by the seven-membered derivatives
than by the five-membered ones (cf. Tables 1 and 4).
We obtained crystal structure of the complex of ligand 7b
with tetrapropylammonium chloride (TPA-Cl). Unfortunate-
ly, the crystals are twined by pseudomerohedry. A possible
twinning element is the composition plane (010), and twin
obliquity (the angle between
the [010]* and [010] directions)
Table 3. The formal charges on selected atoms for bisamides 5c, 7c and azulene 12 estimated by the different
methods.[a]
is equal to 2.968. Because of the
low symmetry of the crystals
(triclinic system) and the differ-
ently overlapping reflections it
was not possible to apply proce-
dures for the twin crystals
during the structure refinement.
Therefore the final discrepancy
factors are quite large.
Atom
M[b]
NPA
ESP
0.20
ꢀ0.07
ꢀ0.39
0.30
M
NPA
ESP
0.07
0.02
ꢀ0.45
0.28
5c ss+ꢀ
7c ss+ꢀ
12
6-C
6-CH
N
NH
2-C
2-CH
N
NH
6-C
6-CH
2-C
ꢀ0.42
0.12
ꢀ0.17
0.24
5c ss++
7c ss++
ꢀ0.48
0.12
ꢀ0.17
0.24
ꢀ0.31
ꢀ0.68
ꢀ0.32
ꢀ0.68
0.30
0.43
0.29
0.42
ꢀ0.24
0.12
ꢀ0.18
0.29
0.03
0.06
ꢀ0.37
0.12
ꢀ0.18
ꢀ0.05
0.230.11
ꢀ0.31
ꢀ0.67
ꢀ0.51
ꢀ0.30
ꢀ0.67
ꢀ0.50
0.29
0.43
0.34
0.28
0.42
0.31
The independent part con-
sists of two ligand complexes,
the structures of which are simi-
lar, though one ligand has disor-
der in one of its phenyl side
chains (Figure 5a). The ligand
ꢀ0.01
ꢀ0.20
ꢀ0.06
0.130.25
0.11
0.10
ꢀ0.14
ꢀ0.21
0.12
2-CH
0.130.25
[a] DFT B3LYP/6-31+G(d,p) calculations in gas phase; ss+ꢀ and ss++ denote the conformations syn-syn+ꢀ and
E
syn–syn++, respectively. [b] M: Mulliken charges.
molecules are in the syn-syn
present on 6-CH and 2-CH hydrogen atoms are negative for
both five- and seven-membered derivatives (5c and 7c, re-
spectively). Moreover, the electron distributions around the
binding clefts are almost identical for both isomers (Fig-
ure S4). Similar observations can be made by two other ap-
proaches: Natural Population Analysis (NPA) and Electro-
static Potential Fit (ESP) Table 3. These results may suggest
that the ligands of both types are able to form hydrogen
bonds of comparable strength, and that the observed differ-
ences in anion affinity are caused by different geometries of
binding sites and conformational preferences of the ligands.
To throw some light on the binding pattern, we also per-
formed molecular modelling of the ligandsꢁ complexes with
chloride anion. For all model ligands 5c–8c, we obtained
the structures in which the chloride is bound by two hydro-
gen bonds formed by amide/thioamide NHs and the anion is
positioned above the azulene plane (Figure 4, Figure S5).
Deviations from coplanarity and distances between the
anion and the central carbonyl atom (6-C for 5c, 6c or 2-C
for 7c, 8c) are smaller in the case of the five-membered de-
rivatives (7c, 8c) (Table 4). The values of these parameters
for ligand 7c—3.4 Š, 1648—are in the CH hydrogen bond
range; moreover, the C–Cl distance corresponds to the
conformations, and bind chlorides through the amide groups
(N–Cl distances are about 3.5 Š). In both complexes the car-
bonyl groups deviate from the azulene planes (the torsion
angles are 7.5 and 178 in one ligand, and 11 and 148 in the
other), so the chloride anion is located above the ligand
plane (Figure 5b). Although the amide groups are twisted
with different angles, the chloride is positioned symmetrical-
ly with respect to the azulene. The crystal structure of
7b·TPA-Cl is in very good agreement with the results of mo-
lecular modelling for the complex of 7c and Clꢀ (cf. Fig-
ure 4b and 5b). The arrangements of ligands and chloride
anion in the two complexes are very similar, while the dis-
Table 4. The calculated distances between hydrogen bond donors and
chloride anion, and the angle between anion and azulene plane for
model ligandsꢁ complexes.[a]
dN–Cl [Š]
dC–Cl [Š][b]
aC-H-Cl[b] [8]
amide 5c
amide 7c
thioamide 6c
thioamide 8c
3.19
3.49
3.16
3.29
3.66
3.38
3.59
3.34
136
164
114
118
[a] Calculations in gas phase by the DFT B3LYP/6-31+G(d,p) method.
R
[b] The carbon (C) refers to 6-C for 5c and 6c or 2-C for 7c and 8c
Chem. Eur. J. 2008, 14, 838 – 846
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
843

J. Jurczak et al.
the phenyl side chains and the anion, which may indicate
the presence of additional aromatic CH···anion interactions
and explain the efficiency of the receptor 7b.
To describe the sizes and geometries of the binding clefts
of the azulene-based building blocks, we defined two param-
eters: the angle “a” assigned to the angle between the
amide/thioamide groups and the middle atom of a ligand,
and the distance “b” between the hydrogen bond donors
(Scheme 1). Since the amide groups are not coplanar in the
syn–syn conformations, here we report the values obtained
from different measurements: that based on X-ray analysis,
and three others based on molecular modelling, for the syn–
syn(++) and syn–syn(+ꢀ) conformations and for the com-
plexes with chloride anions (Table 5).
Table 5. Geometric parameters of the binding clefts of the azulene-based
building blocks.[a]
Data
5
6
7
8
X-ray structure[b] b (a)
4.0 (1178) 5.6 (1458)
4.9 (1168) 5.00 (1228) 5.6 (1428) 5.5 (1458)
4.7 (1108) 4.73(111 8) 5.5 (1388) 5.4 (1358)
[c]
Figure 4. Calculated structures of chloride anion complexes with ligands
by the DFT B3LYP/6-31+G(d,p) method. a) 5c b) 7c.
conf. syn–syn+ꢀ
conf. syn–syn++
Clꢀ complex
b (a)
b (a)
[c]
AHCTREUNG
b (a)[c] 4.8 (1118) 4.6 (1098) 5.3(135 8) 5.1 (1278)
[a] Parameter b [Š]: the distance between (thio)amide Ns, a [8]: angle N-
C-N; see Scheme 1. [b] X-ray measurements at 100 K, for butyl bisamides
(d,p) calculations for methyl bisa-
A
ACHTREUNG
5a and 7a. [c] DFT B3LYP/6-311+G
mides 5c–8c.
ACHTREUNG
The parameters obtained for the complexes with chloride
anions seem to be the most relevant because they describe
the geometry of a binding site that is involved in the host–
guest interaction. At the same time, the calculated values of
b and a for the 7c·Cl complex are similar to those found in
the crystal structure of 7b·TPA-Cl (b=5.3Š, a=1348). On
the basis of the collected parameters, it is apparent that the
five-membered ring derivatives have binding clefts much
wider than those of their seven-membered analogues. Such
differences in geometry can lead to dramatic changes in se-
lectivity and binding affinity, as has been shown for hybrid
systems containing different building blocks.[8,9]
Conclusion
Figure 5. Crystal structure of 7b·TPA-Cl. a) Independent part (the disor-
der in the TPA cation removed for clarity); b) different views of one of
the complexes, with the phenyl groups omitted.
Bisamides based on azulene are attractive building blocks
for anion receptors and they extend the pool of available
structural subunits for ligand design. The five-membered
ring azulene-1,3-dicarboxylic derivatives offer binding clefts
with geometries similar to those of the pyrrole-based bisa-
mides. This building block prefers the syn–syn conformation
of its amide groups; this feature should lead to well preor-
ganized hosts containing azulene subunits. The aromatic
amides of azulene-1,3-dicarboxylic acid are especially prom-
ising, as the model ligand 7b binds anions even more strong-
ly than the corresponding pyrrole-based ligand 4b. There is
also some evidence that the efficiency of these ligands is im-
proved by the CH···anion interactions. Moreover, because of
the ready accessibility of azulene-1,3-dicarboxylic acid di-
tances between 2-CH, amide groups and chloride anion in
the solid state also match the calculated values (distances
both in the modelled structure and in the solid state are
about 3.4 and 3.5 Š, respectively). The main difference in
the crystal structure is that the carbonyl groups are twisted
with a larger angle, and as result the anion is located
“higher” above the ligand plane (the C-H-Cl angles for the
two ligands in the independent part are 156 and 1508).
The crystal structure of 7b·TPA-Cl also supports the
thesis that 2-CH is involved in hydrogen bond with anions;
moreover, there is also a close contact between m-CH of
844
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 838 – 846

Azulene-Based Anion Receptors
FULL PAPER
washed with HCl (2m). The remaining solid was added to the CH2Cl2
phase and dissolved in ethyl acetate (200 mL), washed with HCl (2m, 2”
30 mL) and water (30 mL) and dried over MgSO4. The solvent was re-
moved in vacuo, and the crude product was recrystallized from hot di-
chloroethane with a small amount of pentane, yielding the amide 7b
(0.7 g, 78%) as violet crystals. M.p. 215–2168C; 1H NMR (200 MHz,
DMSO): d=10.29 (s, 2H; NH), 9.71 (d, J1 =9.8 Hz, 2H; 4,8-CH), 9.10 (s,
1H; 2-CH), 8.12 (t, J1 =9.6 Hz, 1H; 6-CH), 7.89–7.77 (m, 6H; o-CH+5,7-
CH), 7.38 (dd, J1 =J1 =7.8 Hz, 4H; m-CH), 7.13ppm (t, J1 =7.5 Hz, 2H;
p-CH); 13C NMR (50 MHz, DMSO): d=164.3, 142.1, 142.0, 140.1, 139.5,
138.4, 129.8, 129.1, 123.7, 120.4, 120.2 ppm; HR ESI: m/z: calcd for
C24H19N2O2: 367.1441; found: 367.1442 [M+H]+; elemental analysis (%)
calcd for 2C24H18N2O2·H2O: C 76.78, H 5.10, N 7.46; found: C 76.62, H
5.48, N 7.53.
chloride (15), the synthesis of the five-membered ring azu-
lene derivatives should be simple and convenient.
The azulene-5,7-dicarboxylic acid derivatives possess
narrow binding sites, the geometries of which originate from
the uncommon seven-membered ring. Such an arrangement
of the amide groups seems to be quite complementary to
the chloride anion. However, the preparation of receptors
containing 5,7-substituted azulene units is more challenging,
which might obstruct possible application of this building
block in the more elaborate systems.
Interaction of the ligands with anions does not significant-
ly affect the azulene chromophore. At this moment, it is un-
certain whether this lack of optical response is caused by
weak anion binding by these simple model ligands or wheth-
er the charge perturbation in the amide groups upon hydro-
gen bond formation is unable to trigger the colour changes
in the chromophore.
Azulene-1,3-dicarboxylic acid bis(4-nitrophenyl)amide (7d): A suspen-
sion of the acid dichloride 15 (1 g, 4 mmol) and p-nitroaniline (3.2 g,
23mmol) in dry CH 2Cl2 (60 mL) was heated at reflux under argon for
24 h. The solid was filtered off and thoroughly washed with HCl (2m),
CH2Cl2 and MeOH. The remaining solid was flash chromatographed
over silica gel with THF as the eluent. Removal of the solvent gave
amide 7d (1.3g, 72%) as dark red crystals. M.p. t.t. >3408C; 1H NMR
(200 MHz, DMSO): d=10.88 (s, 2H; NH), 9.73(d, J1 =10 Hz, 2H; 4,8-
CH), 9.19 (s, 1H; 2-CH), 8.30 (d, J1 =9.2 Hz, 4H; o-CH), 8.21 (t, J1 =
9.7 Hz, 1H; 6-CH), 8.13(d, J1 =9.2 Hz, 4H; m-CH), 7.93ppm (dd, J1 =
J2 =9.9 Hz, 2H; 5,7-CH); 13C NMR (50 MHz, DMSO): d=164.6, 146.4,
142.9, 142.6, 139.8, 139.1, 131.2, 125.5, 119.8, 119.3 ppm; elemental analy-
sis (%) calcd for C24H18N2O2: C 63.16, H 3.53, N 12.28; found: C 63.04, H
3.47, N 12.10.
Our comprehensive studies of azulene-based bisamides,
particularly their structural properties, should help in prepa-
ration of advanced and efficient receptors for anions.
Experimental Section
Azulene-1,3-dicarbothioic acid bis-butylamide (8a): Amide 7a (0.4 g,
1.2 mmol) and Lawessonꢁs reagent (1.3g, 3.2 mmol) were suspended in
dry THF (60 mL) and the mixture was heated at reflux under argon over-
night. The solvent was evaporated in vacuo, and the solid residue was pu-
rified by column chromatography on silica gel (60 g) with CH2Cl2 (1.7 L)
and CH2Cl2/MeOH 100:1 (0.5 L) as eluents. The product was recrystal-
lized from a hexane/ethyl acetate mixture 7:3(20 mL), yielding the thioa-
mide 8a (0.26 g, 60%) as brown-green crystals: m.p. 146–1478C;
1H NMR (200 MHz, CDCl3): d=9.24 (d, J1 =9.8 Hz, 2H; 4,8-CH), 8.01
(t, J1 =5.1 Hz, 2H; NH), 7.81 (s, 1H; 2-CH), 7.74 (t, J1 =9.7 Hz, 1H; 6-
CH), 7.36 (dd, J1 =J1 =9.8 Hz, 2H; 5,7-CH), 3.76 (dt, J1 =7.2 Hz, J2 =
5.6 Hz, 4H; CH2NH), 1.77 (m, 4H; CH2), 1.47 (m, 4H; CH2), 1.02 ppm
(t, J1 =7.3Hz, 6H; C H3); 13C NMR (50 MHz, CDCl3): d=192.3, 140.8,
139.7, 138.5, 133.0, 128.0, 127.6, 45.9, 30.3, 20.4, 13.8 ppm; HR EI: m/z:
calcd for. C20H26N2S2: 358.15374; found: 358.15324 [M]+; elemental anal-
ysis (%) calcd for C20H26N2S2: C 66.99, H 7.31, N 7.81, S 17.89; found: C
66.97, H 7.25, N 7.72, S 17.69.
Details concerning the determination of binding constants are provided
in the Supporting Information, together with structural data. The synthe-
ses of the amide 5a and the thioamide 6a have already been published.[14]
Azulene-1,3-dicarboxylic acid (14) was prepared by the literature proce-
dure.[18]
CCDC-267792, -267793, -633021, 633022 and -642743 contain the sup-
plementary crystallographic data for this paper. These data can be ob-
tained free of charge from The Cambridge Crystallographic Data Centre
via www.ccdc.cam.ac.uk/data_request/cif.
Azulene-1,3-dicarboxylic acid bis-butylamide (7a): Azulene-1,3-dicarbox-
ylic acid (14, 2 g, 9.2 mmol) was suspended in dry CH2Cl2 (80 mL), and
thionyl chloride (5.5 mL, 74 mmol) was then added, followed by two
drops of DMF, and the reaction mixture was heated under reflux under
argon for 12 h. Additional thionyl chloride (2 mL) was added, and heat-
ing was continued for the next 4 h. The solvent and unreacted SOCl2
were removed in vacuo. The crude dichloride 15 was recrystallized from
hot CH2Cl2, yielding red crystals (1.8 g, 78%) that were used without fur-
ther purification.
A suspension of the dichloride 15 (0.33 g, 1.3 mmol) in dry CH2Cl2 was
cooled to 08C under argon, a butylamine solution (0.8 mL, 8 mmol) in
CH2Cl2 (1 mL) was added dropwise, and the reaction mixture was stirred
overnight. The solvent was removed in vacuo, and the solid residue was
purified by column chromatography on silica gel (30 g) with use of a gra-
dient of CH2Cl2 and MeOH [100:1 (100 mL), 100:2 (100 mL), 100:3
(50 mL), 100:4 (100 mL)]. The amide 7a was recrystallized from hot di-
chloroethane, yielding 0.3g (71%) of amide 7a in the form of violet crys-
tals. M.p. 150–1518C; 1H NMR (200 MHz, CDCl3): d=9.32 (d, J1 =
9.8 Hz, 2H; 4,8-CH), 8.00 (s, 1H; 2-CH), 7.72 (t, J1 =9.7 Hz, 1H; 6-CH),
7.37 (dd, J1 =J1 =9.8 Hz, 2H; 5,7-CH), 6.70 (t, J1 =5.5 Hz, 2H; NH), 3.40
(dt, J1 =6.0 Hz, J2 =6.4 Hz, 4H; CH2NH), 1.57 (m, 4H; CH2), 1.40 (m,
4H; CH2), 0.94 ppm (t, J1 =7.1 Hz, 6H; CH3); 13C NMR (50 MHz,
CDCl3): d=166.0, 140.9, 140.2, 138.8, 134.0, 128.1, 120.0, 39.5, 31.9, 20.25,
13.81 ppm; HR ESI: m/z: calcd for C20H27N2O2: 327.2067; found:
327.2088 [M+H]+; elemental analysis (%) calcd for C20H26N2O2: C 73.59,
H 8.03, N 8.58; found: C 73.67, H 8.04, N 8.66.
[1] M. D. Lankshear, P. D. Beer, Coord. Chem. Rev. 2006, 250, 3142; T.
Gunnlaugsson, M. Glynn, G. M. Tocci, P. E. Kruger, F. M. Pfeffer,
Coord. Chem. Rev. 2006, 250, 3094; N. Gimeno, R. Vilar, Coord.
Chem. Rev. 2006, 250, 3161; P. A. Gale, R. Quesada, Coord. Chem.
Rev. 2006, 250, 3219; P. A. Gale, Coord. Chem. Rev. 2003, 240, 191;
P. A. Gale, Coord. Chem. Rev. 2001, 213, 79.
[2] J. L. Sessler, P. A. Gale, W. S. Cho, Anion Receptor Chemistry, The
Royal Society of Chemistry, Cambridge, 2006.
[3] H. Luecke, F. A. Quiocho, Nature 1990, 347, 402; J. J. He, F. A.
Quiocho, Science 1991, 251, 1479; J. W. Pflugrath, F. A. Quiocho,
Nature 1985, 314, 257
[4] J. L. Sessler, S. Camiolo, P. A. Gale, Coord. Chem. Rev. 2003, 240,
17; K. H. Choi, A. D. Hamilton, Coord. Chem. Rev. 2003, 240, 101.
[5] S. O. Kang, M. A. Hossain, K. Bowman-James, Coord. Chem. Rev.
2006, 250, 3038; S. O. Kang, R. A. Begum, K. Bowman-James,
Angew. Chem. 2006, 118, 8048; Angew. Chem. Int. Ed. 2006, 45,
7882; P. A. Gale, Acc. Chem. Res. 2006, 39, 465; C. R. Bondy, S. J.
Loeb, Coord. Chem. Rev. 2003, 240, 77.
[6] K. Bowman-James, Acc. Chem. Res. 2005, 38, 671.
[7] F. P. Schmidtchen, Coord. Chem. Rev. 2006, 250, 2918.
[8] M. J. Chmielewski, J. Jurczak, Tetrahedron Lett. 2005, 46, 3085.
Azulene-1,3-dicarboxylic acid bis-phenylamide (7b): Dichloride 15, pre-
pared as described above, was suspended in dry CH2Cl2 (0.8 g, 3.1 mmol
in 60 mL), aniline was added dropwise (1.7 mL, 18.6 mmol), and the mix-
ture was stirred overnight. The precipitate was filtered off and thoroughly
Chem. Eur. J. 2008, 14, 838 – 846
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
845

J. Jurczak et al.
[9] J. L. Sessler, E. Katayev, G. D. Pantos, P. Scherbakov, M. D. Reshe-
tova, V. N. Khrustalev, V. M. Lynch, Y. A. Ustynyuk, J. Am. Chem.
Soc. 2005, 127, 11442; J. L. Sessler, D. Q. An, W. S. Cho, V. Lynch,
D. W. Yoon, S. J. Hong, C. H. Lee, J. Org. Chem. 2005, 70, 1511.
[10] K. Kavallieratos, S. R. de Gala, D. J. Austin, R. H. Crabtree, J. Am.
Chem. Soc. 1997, 119, 2325.
[18] L. J. Mathias, C. G. Overberger, J. Org. Chem. 1980, 45, 1701.
´
[19] T. Zielinski, J. Jurczak, Tetrahedron 2005, 61, 4081.
[20] M. P. Cava, M. I. Levinson, Tetrahedron 1985, 41, 5061.
[21] V. Amendola, D. Esteban-Gomez, L. Fabbrizzi, M. Licchelli, Acc.
Chem. Res. 2006, 39, 343; D. Esteban-Gomez, L. Fabbrizzi, M. Lie-
chelli, J. Org. Chem. 2005, 70, 5717; M. Boiocchi, L. Del Boca, D.
Esteban-Gomez, L. Fabbrizzi, M. Licchelli, E. Monzani, Chem. Eur.
J. 2005, 11, 3097.
[11] K. Kavallieratos, C. M. Bertao, R. H. Crabtree, J. Org. Chem. 1999,
64, 1675.
[12] P. A. Gale, S. Camiolo, G. J. Tizzard, C. P. Chapman, M. E. Light,
S. J. Coles, M. B. Hursthouse, J. Org. Chem. 2001, 66, 7849.
[13] C. Dolain, V. Maurizot, I. Huc, Angew. Chem. 2003, 115, 2844;
Angew. Chem. Int. Ed. 2003, 42, 2738; K. S. Jeong, Y. L. Cho, J. U.
Song, H. Y. Chang, M. G. Choi, J. Am. Chem. Soc. 1998, 120, 10982.
[22] S. Kubik, C. Reyheller, S. Stuwe, J. Inclusion Phenom. Macrocyclic
Chem. 2005, 52, 137.
[23] V. S. Bryantsev, B. P. Hay, Org. Lett. 2005, 7, 5031.
[24] V. S. Bryantsev, B. P. Hay, J. Am. Chem. Soc. 2005, 127, 8282; O. B.
Berryman, V. S. Bryantsev, D. P. Stay, D. W. Johnson, B. P. Hay, J.
Am. Chem. Soc. 2007, 129, 48.
´
[14] T. Zielinski, M. Ke¸dziorek, J. Jurczak, Tetrahedron Lett. 2005, 46,
6231.
[25] M. A. Hossain, S. O. Kang, J. M. Llinares, D. Powell, K. Bowman-
James, Inorg. Chem. 2003, 42, 5043; Y. Inoue, T. Kanbara, T. Yama-
moto, Tetrahedron Lett. 2003, 44, 5167.
[26] H. J. Lee, Y. S. Choi, K. B. Lee, J. Park, C. J. Yoon, J. Phys. Chem. A
2002, 106, 7010.
[15] H. Salman, Y. Abraham, S. Tal, S. Meltzman, M. Kapon, N. Tessler,
S. Speiser, Y. Eichen, Eur. J. Org. Chem. 2005, 2207.
[16] H. J. Tobler, A. Bauder, H. H. Günthard, J. Mol. Spectrosc. 1965, 18,
239.
[17] C. Jutz, E. Schweiger, H.-G. Lçbering, A. Kraatz, W. Kosbahn,
Chem. Ber. 1974, 107, 2956.
Received: April 17, 2007
Published online: September 20, 2007
846
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 838 – 846