Preorganized anion traps for exploiting anion-π interactions: An experimental and computational study

Article abstract of DOI:10.1002/chem.201302598

1,3-Bis(pentafluorophenyl-imino)isoindoline (A^F) and 3,6-di-tert-butyl-1,8-bis(pentafluorophenyl)-9H-carbazole (B^F) have been designed as preorganized anion receptors that exploit anion-π interactions, and their ability to bind chlor

Full text of DOI:10.1002/chem.201302598

DOI: 10.1002/chem.201302598
Preorganized Anion Traps for Exploiting Anion–p Interactions:
An Experimental and Computational Study
Anne Bretschneider,[a] Diego M. Andrada,[b] Sebastian Dechert,[a] Steffen Meyer,[a]
Ricardo A. Mata,*[b] and Franc Meyer*[a]
Abstract:
1,3-Bis(pentafluorophenyl-
solution because of E,Z/Z,Z isomerism
of the host. In the case of the more
rigid receptor BF, Job plots evidence
1:1 complex formation with ClÀ and
BrÀ, and association constants up to
960mÀ1 have been determined depend-
ing on the solvent. Crystal structures of
BF and BF:DMSO visualize the distinct
preorganization of the host for anion–p
interactions. The reference compounds
1,3-bis(2-pyrimidylimino)isoindoline
(AN) and 3,6-di-tert-butyl-1,8-diphenyl-
9H-carbazole (BH), which lack the per-
fluorinated flaps, do not show any indi-
cation of anion binding under the same
conditions. A detailed computational
analysis of the receptors AF and BF and
their host–guest complexes with ClÀ or
BrÀ was carried out to quantify the in-
teractions in play. Local correlation
methods were applied, allowing for
a decomposition of the ring–anion in-
teractions. The latter were found to
contribute significantly to the stabiliza-
tion of these complexes (about half of
the total energy). Compounds AF and
BF represent rare examples of neutral
receptors that are well preorganized
for exploiting anion–p interactions, and
rare examples of receptors for which
the individual contributions to the
binding energy have been quantified.
imino)isoindoline (AF) and 3,6-di-tert-
butyl-1,8-bis(pentafluorophenyl)-9H-
carbazole (BF) have been designed as
preorganized anion receptors that ex-
ploit anion–p interactions, and their
ability to bind chloride and bromide in
various solvents has been evaluated.
Both receptors AF and BF are neutral
but provide a central NH hydrogen
bond that directs the halide anion into
a preorganized clamp of the two elec-
tron-deficient appended arenes. Crystal
structures of host–guest complexes
of AF with DMSO, ClÀ, or BrÀ
(AF:DMSO, AF:ClÀ, and AF2 :BrÀ)
reveal that in all cases the guest is lo-
cated in the cleft between the per-
fluorinated flaps, but NMR spectrosco-
py shows a more complex situation in
Keywords: anion–p interactions
·
density functional calculations · hy-
drogen bonds · receptors · solution
studies
Introduction
tions involving negatively charged residues and polarizable
aryl groups in host–guest systems[4] a series of computational
studies in 2002 supported the existence of attractive forces
between electron-deficient arenes and anions centered
above their p-cloud;[5] the term anion–p interaction was
then coined.[5d] Shortly thereafter anion–p interactions were
explicitly mentioned in crystallographic work for the first
time, evidencing the location of chloride above a 1,3,5-tria-
zine ring just as predicted by theory.[6] This marked the be-
ginning of systematic experimental studies towards the de-
tection, description, and targeted use of anion–p interactions
in the solid state and in solution, mostly focusing on N-het-
erocycles such as 1,3,5-triazine or suitably substituted arenes
such as pentafluorophenyl groups.[7] Anion–p interactions
are now beneficially exploited in fields such as anion sens-
ing,[8] anion transport through membranes,[9] or supramolec-
ular assembly,[10] and they are even considered relevant for
anion transport in biological systems.[11]
Anion–p interactions, that is, the attractive interactions be-
tween anions and electron-deficient p systems, have sparked
significant interest in recent years.[1] Somewhat counterintui-
tive at first sight and thus controversially discussed for some
time,[2] anion–p interactions are now considered as an im-
portant addition to the bouquet of established non-covalent
interactions such as hydrogen bonds or CH–p and cation–p
interactions, which form the basis of modern supramolecular
chemistry.[3] After early reports on weak attractive interac-
[a] A. Bretschneider, Dr. S. Dechert, S. Meyer, Prof. Dr. F. Meyer
Institut fꢀr Anorganische Chemie
Georg-August-Universitꢁt Gçttingen
Tammannstrasse 4, 37077 Gçttingen (Germany)
Fax : (+49)551-3933063
E-mail: franc.meyer@chemie.uni-goettingen.de
[b] Dr. D. M. Andrada, Prof. Dr. R. A. Mata
Institut fꢀr Physikalische Chemie
The number of reports that evidence anion–p interactions
in the solid state is indeed increasing rapidly. However, ex-
perimental data that support and quantify the attractive in-
teraction between anions and electron-deficient neutral
arenes in the solution phase are still rather limited.[12] Ac-
cording to most estimates the free energy of binding
Georg-August-Universitꢁt Gçttingen
Tammannstrasse 6, 37077 Gçttingen (Germany)
E-mail: rmata@gwdg.de
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201302598.
16988
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 16988 – 17000

FULL PAPER
(ÀDG0) of anion–p interactions, which mainly results from
electrostatic and anion-induced polarization contributions,[13]
is likely rather low. A recent account concluded that the
free energy of binding between halide anions and substitut-
ed phenyl groups is usually less than 1 kcalmolÀ1 in organic
solvents,[12c] though the interaction may be energetically
much more favorable in solvents with a large dielectric con-
stant.[12b] In fact, most of the reported synthetic receptors
combine anion–p interactions with other forces such as hy-
drogen bonding or salt bridges, and the anion–p interaction
serves to augment anion binding in solution.[14] Thus, the ex-
traction of thermodynamic data for the individual anion–p
contribution to the interaction is complicated.[15]
Cooperativity of anion–p and other interactions such as
hydrogen bonding may bring about reasonable binding con-
stants for the hosting of anions by neutral receptors in solu-
tion, a situation that is still rather rare for systems that
solely exploit anion–p contacts. Following this approach,
just recently constants for chloride binding in the order of
3ꢃ103 mÀ1 in CD3CN solution as well as the first crystallo-
graphic example of an anion–p interaction between an un-
charged pentafluorophenyl derivative and a halide anion
have been communicated.[15] However, most reported recep-
tors that use directing hydrogen-bonding groups or charge
assistance for anion–p interactions are conformationally
quite flexible. This leads to positional diversity of the anion
above the arene ring,[16] to competing intra- and intermolec-
ular interactions[17] as well as to facile superseding of the
anion–p contact by, for example, CH–anion hydrogen bond-
ing or interactions between anion and solvent. Herein, we
report two new neutral halide receptors AF and BF that trig-
established with their non-fluorinated derivates AN and BH
(see below).
Results and Discussion
Synthesis and structural characterization of the receptors:
Because of the highly electron-withdrawing nature of the
pentafluorophenyl groups, established procedures for the
synthesis of 1,3-bis(arylimino)isoindolines and 1,8-bisACHTUNGTRENNUNG(aryl)-
9H-carbazoles had to be modified in order to obtain the
target receptors. 1,3-Bis(pentafluorophenylimino)-isoindo-
line (AF) and the related 1,3-bis(2-pyrimidylimino)-isoindo-
line (AN) were prepared by heating phthalocyanid and the
corresponding amines (pentafluoroaniline or 2-aminopyrimi-
dine, respectively) in nBuOH to reflux, using CaCl2 as a cat-
alyst (Scheme 1).[18] Due to the electron-deficient arenes,
both amines are highly deactivated and thus yields were rel-
atively low even after reaction times of up to three weeks.
Additionally working under exclusion of water is strongly
recommended as this prevents hydrolysis, which further min-
imizes the yield.
À
ger, through a directing N H hydrogen bond, the binding of
the anion into a preorganized clamp of two electron-defi-
cient arenes exhibiting anion–p interactions. The design of
these receptors is reminiscent of two mussel shell valves
with the NH drawing the prey anion into the chamber.
Scheme 1. Synthesis of receptors AF and AN.
The synthesis of 3,6-di-tert-butyl-1,8-bis(perfluorophenyl)-
9H-carbazole (receptor BF) and the non-fluorinated parent
compound 3,6-di-tert-butyl-1,8-diphenyl-9H-carbazole (BH)
is depicted in Scheme 2. Following literature procedures
Friedel–Crafts alkylation of the carbazole was performed,
thus introducing tBu groups into the backbone. Reaction of
compound 1 with bromine then gave compound 2 in quanti-
tative yield.[19] Because no suitable cross-coupling protocol
was found to provide the target compounds directly from
compound 2, an additional step was included in which the
bromo substituents were exchanged against boronic esters.
This was achieved through in situ protection of the lithiated
NH group with CO2 followed by further addition of bis(pi-
nacolato)diboron to give compound 3. Finally, Suzuki–
Miyaura cross-coupling conditions were applied to yield BF
and BH in almost quantitative yields.[20]
Whereas BF is a particularly rigid system that enforces
anion–p interactions once the anion is drawn into the open
pocket between the pentafluorophenyl flaps, receptor AF
offers more conformational and configurational flexibility.
The study of the anion binding with AF and BF employing
both experimental and computational methods synergistical-
ly provides significant new information on the relevance of
anion–p interactions and renders guidelines for the develop-
ment of future new types of selective anion receptors that
exploit such anion–p attractive contacts. Comparison is also
X-ray diffraction analysis of single crystals of receptor BF,
obtained by reverse vapor diffusion of a solution of the re-
ceptor in chloroform into toluene, confirmed the molecular
structure (Figure 1) and highlighted the preorganized anion
binding site with the NH group “sting” directed into the
space between the two pentafluorophenyl flaps, which are
Chem. Eur. J. 2013, 19, 16988 – 17000
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16989

R. A. Mata, F. Meyer et al.
severely tilted with respect to the carbazole backbone
(angles between the C6F5 and the carbazole planes: 64/418).
A first indication for the ability of both receptors to bind
guests within their chambers flanked by the pentafluoro-
phenyl flaps was given by crystal structure analyses of their
adducts with DMSO, namely BF:DMSO (Figure 1bottom)
and AF:DMSO (Figure 2), both obtained by slow evapora-
tion of solutions of the compounds in DMSO. Most impor-
Scheme 2. Synthesis of receptors BF and BH.
Figure 2. Molecular structure of the AF:DMSO adduct. All hydrogen
atoms except the NH proton are omitted for clarity.
tantly, in BF:DMSO the orientation of the pentafluorophen-
yl rings with respect to the carbazole backbone is barely
changed (angles: 54/548) compared to the free receptor, re-
flecting favorable preorganization. This suggests that any en-
ergetic penalty for adapting the shape of the receptor upon
guest inclusion is low if the guest has the proper size, and
hence it may suggest some favorable size selectivity.
Determination of the anion binding capabilities: The ability
of the new receptors to bind anions was evaluated by deter-
mining, through NMR spectroscopy, the association con-
stants (Ka) of the receptor with different halides. To obtain
Ka values, 1H and 19F NMR titration experiments with
(Bu4N)Cl and (Bu4N)Br as anion sources were carried out
in different deuterated solvents. The stoichiometry of the
host–guest complexes in solution was determined by Jobꢄs
method.[21]
Type-A receptors: Preliminary studies with receptor AF
clearly demonstrated its ability to interact with halide
anions, reflected by significant chemical shift changes upon
addition of, for example, (Bu4N)Br (Figure 3). However,
when looking at the system in more detail, problems arose
to determine the stoichiometry as well as the stability con-
1
stant. The H and 19F NMR spectra of AF show the presence
of two isomers (Z,Z or E,Z) in solution in a ratio of roughly
1:1, with only slight dependency on the solvent used or the
temperature (Scheme 3).
Figure 1. Molecular structure of receptor BF (top) and the adduct
BF:DMSO (bottom). All hydrogen atoms except the NH proton are
omitted for clarity.
16990
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 16988 – 17000

Preorganized Anion Traps for Exploiting Anion–p Interactions
FULL PAPER
Figure 3. 1H (left) and 19F NMR spectra (right) of AF measured in CDCl3
before (top) and after (bottom) the addition of approximately ten equiv-
alents (Bu4N)Br. Signals labeled “a” and “b” are assigned to the E,Z and
Z,Z isomers, respectively (compare Scheme 3).
Figure 4. Molecular structures of (Me3PhN)[AF:Cl] (top) and
(Ph4P)[AF2:Br] (bottom). To visualize the position of the anions relative
to the C6F5 flaps, the top view of each arene plane is depicted next to the
structures. Cations and hydrogen atoms except for the NH proton are
omitted for clarity.
(Ph4P)Br gave crystalline material that was analyzed by X-
ray diffraction. The molecular structures of the anionic
host–guest complexes of the salts (Me3PhN)
[AF:Cl] and
ACHTUNGTRENNUNG
(Ph4P)[AF2:Br] are shown in Figure 4.
The obtained structures of AF:ClÀ and A2F:BrÀ both show
the halide located between the perfluorated flaps of the re-
ceptor AF and hydrogen bonded to its central NH group
(d(N···Cl)=3.18 ꢅ in AF:ClÀ, d(N···Br)=3.41/3.36 ꢅ in
AF2:BrÀ) as anticipated beforehand. With chloride, AF forms
the expected 1:1 complex, whereas two AF receptor mole-
cules wrap around the bromide in AF2 :BrÀ, likely because of
the larger size of bromide. In order to evaluate the two
structures with respect to their anion–p interactions, the lo-
cation of the anions relative to the C6F5 planes and their dis-
tances to either the center (d(X–centroid)) or the plane
(d(X–plane)) of the arene were considered. In case of
AF:ClÀ the chloride is located roughly above the center of
the arene rings (Figure 4top) with rather short distances
d(Cl–centroid)=3.28–3.35 ꢅ and d(Cl–plane)=3.28–3.30 ꢅ.
These values lie well within the typical range of anion–p in-
teractions.[16a] For AF2 :BrÀ the offset from the normal to the
center of the arene ring is more pronounced and the bro-
Scheme 3. Isomeric equilibrium between the E,Z (a) and Z,Z (b) forms
of AF in solution and their binding of halide anions XÀ.
Addition of halide salts led to considerable shifts for the
NH proton signals of both isomers as well as for the
19F NMR resonances, more so for the Z,Z isomer than for
the E,Z isomer. In addition, the Z,Z versus E,Z equilibrium
is shifted towards the Z,Z form (Scheme 3, Figure 3; roughly
a 1.5:1 Z,Z to E,Z ratio after addition of 10 equiv (Bu4N)Br
in CDCl3). Although a significant excess of halide salt was
added to the solution, no complete conversion to the Z,Z-
isomer was achieved. Thus, although receptor AF shows
a promising potential to bind anions, the complex equilibria
in solution make further investigations and analysis of this
system exceedingly cumbersome.
À
mide in most cases is located above a C C bond, hence the
interaction is best described as being of the h2-type
mode.[16a] Still, the distances are rather short and are found
in the ranges d(Br–centroid)=3.53–3.66 ꢅ and d(Br–
plane)=3.27–3.46 ꢅ, suggesting favorable anion–p interac-
tions. The four C6F5 flaps in AF2 :BrÀ wrap around the central
bromide in a roughly tetrahedral arrangement, though the
angle N···Br···N deviates from linearity (1288).
The related receptor AN, on the contrary, gave a very
1
Slow diffusion of hexane into a solution of the receptor
simple H NMR spectrum showing only one set of signals
AF in chloroform with an excess of either (Me3PhN)Cl or
with a deshielded NH proton, and no further isomers
Chem. Eur. J. 2013, 19, 16988 – 17000
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16991

R. A. Mata, F. Meyer et al.
Figure 6. 1H (left) and 19F NMR (right) spectra of receptor BF in
[D8]THF before (top) and after (bottom) addition of approximately
three equivalents of (Bu4N)Br.
Figure 5. Molecular structure of AN and its 1H NMR spectrum measured
in CDCl3. Hydrogen atoms except for the NH proton are omitted in the
structure.
(Figure 5). In this case the addition of either (Bu4N)Cl or
(Bu4N)Br or other halide salts had no effect on the NMR
chemical shifts, suggesting that AN does not bind any anions.
A rationalization is provided by the crystal structure, which
clearly shows two strong intramolecular NH···N interactions
(d(H···N)=2.09/2.19 ꢅ). This bifurcated hydrogen bond not
only prevents E/Z isomerism, but seems to be so strong that
neither solvent nor anions have any significant effect.
Type-B receptors: In the case of receptor BF, because of the
rigidity of the carbazole-based scaffold, isomerism cannot
take place and the receptor is highly preorganized. Indeed,
the neutral BF showed a pronounced ability of binding
1
halide anions as reflected by characteristic changes in the H
and 19F NMR spectra upon addition of either (Bu4N)Cl or
(Bu4N)Br (Figure 6). Unfortunately, all attempts to crystal-
lize the anion-containing host–guest complexes BF:XÀ were
unsuccessful.
Figure 7. Job plot (top) and titration curve (bottom) of receptor BF and
&
*
(Bu4N)Br in [D8]THF ( =NMR signal of the NH group, =NMR
signal of the p-F).
Job plots for receptor BF with ClÀ and BrÀ showed a 1:1
stoichiometry in all solvent systems used in this work,
namely CD2Cl2, CD3CN, C6D6, and [D8]THF (see Fig-
ure 7top, and the Supporting Information). Titration experi-
ments for each combination of anion and solvent were per-
formed, monitoring the shifts of the p-F and NH signal as
a function of the amount of anions added (Figure 7bottom).
The obtained titration curves were then fitted with non-
linear and linear regression methods to extract the binding
constants (see the Supporting Information for details). The
resulting Ka values derived from Scatchard plots are listed
in Table 1.
Table 1. Association constants Ka [mÀ1] for the 1:1 complexes of receptor
BF with ClÀ or BrÀ in different solvents.
CD2Cl2
CD3CN
C6D6
[D8]THF
BF:ClÀ
BF:BrÀ
4.1Æ0.4
2.8Æ0.2
12.0Æ0.7
6.6Æ0.4
(9.6Æ4.6)ꢃ102
(5.0Æ2.5)ꢃ102
(4.4Æ2.9)ꢃ102
(2.8Æ1.0)ꢃ102
In all solvents the binding constants are higher for ClÀ
than for BrÀ, roughly by a factor of two. Although this gen-
eral trend is expected,[15] the difference in the Ka values for
ClÀ and BrÀ may also reflect the better fit of the smaller ClÀ
16992
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 16988 – 17000

Preorganized Anion Traps for Exploiting Anion–p Interactions
FULL PAPER
into the highly preorganized cleft between the two arenes of
the receptor. Association constants in CD2Cl2 and CD3CN
are quite small and the halide binding abilities of BF are
rather weak (Ka =12.0mÀ1 for BF:ClÀ in CD3CN, corre-
sponding to DG=À6.1 kJmolÀ1), which may possibly reflect
competing interactions of the solvent with the NH unit of
the receptor as well as good solvation of the anions in the
polar solvent CD3CN. On the contrary, in C6D6 or [D8]THF,
where the halide salts (Bu4N)X are poorly soluble, the inter-
action of the anions with the receptor is more favored (Ka =
960mÀ1 for BF:ClÀ in C6D6, corresponding to DG=
À16.7 kJmolÀ1). It is interesting to note that a surprisingly
high association constant in the polar solvent CD3CN was
found for N,N’-(1,2-phenylene)-bis(pentafluorobenzamide)
and chloride (Ka ꢀ3500mÀ1),[15] likely because of the pres-
(X=Cl, Br) values are 3.016 and 3.230 ꢅ, respectively,
which are somewhat shorter than in the X-ray structures.
As mentioned above, the geometry of the receptor is only
slightly affected by the coordination. The distance between
both ring flaps (measured as the distance between both cen-
ters of mass) are 6.654, 6.166, and 6.641 ꢅ, for AF, AF:ClÀ
and AF:BrÀ, respectively. Even more, the tilted angles of the
flaps with respect to the isoindolin plane change from 50.8/
48.68 for AF to 52.9/48.88 for the complex with either anion.
Therefore, only in the case of the chloride do the rings
move closer, possibly to enhance the interaction between
the rings and the halide anion.
We have studied the energetics for the formation of host–
guest complexes by computing the interaction energy
(DEint), the deformation energy of the receptor (DEdef), and
the binding energy (DEbind =DEdef+DEint). The deformation
energy is the energy required to distort the free receptor
into the geometry observed in the complex. In this regard,
we have improved the energy obtained by carrying out
single-point calculations at the DF-LMP2 level of theor-
y,[23a–c] with the cc-pVTZ basis set for the hydrogen atoms
and aug-cc-pVTZ for all remaining elements.[24] The basis
set will be referred to as AVTZ for convenience. Because
density-fitting approximations have been used throughout,
we will also drop the “DF” prefix. The values are given in
Table 2.
À
ence of two N H···Cl hydrogen bonds. To assess the individ-
À
ual contributions of N H···X hydrogen-bonding and the
anion–p interaction for the synergetic halide binding by the
preorganized receptors AF and BF, detailed computational
studies have been performed (see the following section).
Receptor BH with phenyl flaps was synthesized to experi-
mentally evaluate the effect of the electron-deficient C6F5
rings and their anion–p interactions. Titration of BH with
solutions of (Bu4N)Cl or (Bu4N)Br indeed showed no shifts
of the NMR signals, neither for the NH nor the phenyl flaps
(see the Supporting Information). Obviously, the attraction
between the halide anion and the p-clouds of the phenyl
flaps is too weak (or is even repulsive) and not competitive
with the solvent. Electron-deficient arene rings are needed
to enhance the interaction.
Table 2. Binding (DEbind), interaction (DEint) and deformation (DEdef
)
energy values (in [kJmolÀ1]) for the receptors AF and BF with chloride
and bromide ions.
B3LYP-D3
DEdef
LMP2
DEdef
Computational results: In order to gain a deeper insight into
each system, we have performed a series of computational
structure calculations. The structures of the receptor system
AF both in its free form as well as interacting with chloride
and bromide ions, AF:XÀ (X=ClÀ, BrÀ), were optimized at
the B3LYP-D3/def2-TZVPP level of theory,[22] all of them in
the Z,Z conformation. The obtained structures of the
anion–receptor adducts are shown in Figure S1 in the Sup-
porting Information. Superposition plots of the structures of
AF:ClÀ determined by X-ray crystallography and computa-
tionally are provided in Figure S2 in the Supporting Infor-
mation, revealing a very good agreement except for the ro-
tational position of the C6F5 rings.
DEbind
DEint
DEbind
DEint
AF:ClÀ
AF:BrÀ
BF:ClÀ
BF:BrÀ
AF2:ClÀ
AF2:BrÀ
À183.2
À152.3
À178.2
À148.6
À336.6
À287.9
13.7
10.2
9.7
7.7
8.1/7.5
7.1/5.8
À196.9
À162.5
À187.9
À156.3
À352.2
À300.8
À171.6
À152.8
À160.1
À145.9
À338.6
À293.1
6.7
5.0
12.4
8.3
4.7/6.1
2.6/3.2
À178.3
À157.8
À172.5
À154.2
À349.4
À298.9
The LMP2/AVTZ and B3LYP-D3/def2-TZVPP energies
are in relatively good agreement, showing the same trends
in interaction strengths. The binding energies obtained are
in the range expected for receptors with electron-deficient
aromatic rings.[25] Comparing the halides, the binding ener-
gies with receptor AF are somewhat larger in the case of
ClÀ, by about 30.9 and 18.8 kJmolÀ1 for B3LYP-D3 and
LMP2, respectively. The deformation energies are small, as
expected in view of the only slight changes to the conforma-
tion. The small structural changes between the free receptor
and the receptor–anion systems as well as the computed
DEdef values confirm the compounds as suitably preorgan-
ized clamps for anion binding. The values in Table 2, howev-
er, give little information about the existence of anion–p in-
teractions, and their weight in the total interaction. To inves-
tigate this effect, we have carried out calculations in model
Only minor changes are observed on almost all atoms
when comparing the computed structures of the free recep-
tor with those of the host–guest complexes AF:XÀ (XÀ =ClÀ,
BrÀ). The only exception is for the NH upon hydrogen bind-
ing to the anion, which moves out of the isoindole ring
À
plane, directed towards the halide ion. Furthermore, the N
H bond length is slightly lengthened, from 1.004 to 1.064
and 1.049 ꢅ, for AF:ClÀ and AF:BrÀ, respectively. The ef-
fects of the hydrogen bond are visible in both the distance
and the bending angle. To better illustrate this latter effect,
superposition plots of the structure of receptor AF with the
adducts AF:ClÀ and AF:BrÀ are shown in Figures S3 and S4
in the Supporting Information, respectively. The d(N···X)
Chem. Eur. J. 2013, 19, 16988 – 17000
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16993

R. A. Mata, F. Meyer et al.
systems, examining in greater detail the interaction between
an anion and the fluorinated rings.
This is expected, because MP2 tends to overestimate the dis-
persion interactions.[31] The curves for the halides are rather
similar, with comparable well depths (in numbers À85.3 and
À71.6 kJmolÀ1 at the LCCSD(T0)/AVTZ level of theory),
but a shorter equilibrium distance in the case of ClÀ com-
pared to BrÀ (3.3 vs. 3.5 ꢅ). These distances are in the range
of those reported in the literature for 1:1 anion–p complex-
es,[26d,32] and they are in very good agreement with values
found experimentally for AF:ClÀ.
The effect of correlation is given by the difference of the
calculated curves to the HF curve. It is evident that the sys-
tems interact even at an uncorrelated level, which is caused
by electrostatic effects. For instance, when the anion is
a chloride, the value at the equilibrium distance predicted
by HF (3.5 ꢅ) is À61.9 kJmolÀ1 whereas for bromide
(3.8 ꢅ) it is À53.1 kJmolÀ1. The correlation energy contribu-
tion to the interaction predicted at the LMP2 level is about
45% for chloride and 50% for bromide. The correlation
energy values computed for each anion are À42.4 and
À43.6 kJmolÀ1 for chloride and bromide, respectively. The
values are quite similar as a result of a balance between the
shorter interaction distances for chloride on the one hand
and the larger polarizability of bromide on the other hand.
In order to better understand the role of correlation effects,
As a model system for the interaction of the flaps, we
have used two pentafluorobenzene rings. The attractive in-
teraction of the halide with electron-deficient aromatic rings
has been a subject of several previous computational stud-
ies.[5c,13,26] Among them, it has been demonstrated that the
strength of the anion–p interaction and its contributions to
the interaction energy, namely electrostatic, induction, and
dispersion, sharply depend on the quadrupole moment and
the molecular polarizability values of the aromatic com-
pound.[5c,d,27] On this ground, we opted for a pentafluoro-
substituted benzene to mimic the former properties of the
flaps of the receptor,[28] as the dipole moment can strongly
influence the interaction with the anion. We have also kept
in our calculations the orientation of the hydrogen bond as
close as possible to what should be observed in the recep-
tor.[16b,29] Because we are interested in the study of a receptor
class (not a single compound), the relative orientation of the
rings is a factor which has to be taken into account, given it
may vary depending on the compound and upon coordina-
tion.[1d] We have considered two cases. In a first series of cal-
culations, the pentafluorobenzene rings are placed parallel,
with the anion in the middle (similar to a sandwich com-
plex). In this case, the halide is
located above the rings cent-
roids. The potential energy plot
is generated by a symmetric dis-
placement of the rings, away
from the halide (Figure 8). Fur-
thermore, we have considered
a bent orientation at 1508 for
the angle ringcentroid-X-ringcentroid
,
closely reproducing the orienta-
tion of compound AF:XÀ, and
also the tilted angles of 58/588
from the parallel case. The dis-
placements are made along the
vector joining the halide and
the geometric center of the ring
(see Figure 8 for the represen-
tation). In this case, the halide
is offset relative to the rings
centroids, with the projection
points changing with the vary-
ing distance. The potential
energy curves for the chloride
and bromide ions in the “paral-
lel model” are given in Fig-
ures 8A and B, respectively. In
both cases, the reference is
given by the LCCSD(T0)/
AVTZ values.[23d] The SCS-
LMP2/AVTZ[30] curves
are
almost coincident to the latter,
whereas LMP2/AVTZ slightly
Figure 8. Potential energy curves of the parallel model with A) ClÀ and B) BrÀ ions and of the bent model
with C) ClÀ and D) BrÀ ions. The energy values are given as a function of the distance between the center of
overestimates the interaction. the ring and the halide anion.
16994
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 16988 – 17000

Preorganized Anion Traps for Exploiting Anion–p Interactions
FULL PAPER
we have carried out a decomposition of the LMP2 energy.
The local character of occupied and virtual orbitals in the
local correlation treatments offers the possibility to dissect
intramolecular effects (double excitations from occupied or-
bitals of one unit into virtual orbitals of the same unit) from
intermolecular ones (excitations involving orbitals from
both units).[33] Additionally, the intermolecular correlation
effects can be decomposed according to different excitation
classes: dispersion effects originate from simultaneous mon-
oexcitations at each unit, exchange–dispersion is similar but
the monoexcitations are from the occupied space of one
unit to the virtual space of the other and vice versa; ionic
contributions come from monoexcitations from the occupied
space of both units into the virtual space of only one of
them. The different classes have been discussed in detail by
Schꢀtz et al.[33] In Figures 8A and B, together with the
energy curves, each intermolecular contribution term is also
displayed as a function of the distance between the center
of the ring and the halide anion, corresponding to anion–
ring interactions. These curves show a continuous favorable
interaction with decreasing distances. As has been expected,
the main contributions to the interaction from the correla-
tion energy are linked to dispersion (58–63%) and ionic ex-
citations (36–37%) whereas the exchange–dispersion effect
is negligible. These observations are in accordance with pre-
vious reports on anion–p interactions.[5d,13b,26h,34] It is impor-
tant to stress that the dispersion and ionic effects significant-
ly contribute to a shorter con-
of the molecule instead of penetrating deep into the cavity.
Albrecht et al. and Hay et al. (among others) have ad-
dressed this issue in some experimental and computational
systematic studies on differently substituted electron-defi-
cient arenes. Their results revealed the flexibility in the posi-
tion of the anion over the rings. These positions can be con-
trolled by a hydrogen-bond donor directing substituent, such
[3c,12a,15,16,29,36]
À
À
À
À
as C H and N H bonds.
The N H···X interac-
tion in AF:ClÀ keeps the chloride ion close to the rim of the
arene, as evidenced by the offset distances (Figure 8). More-
over, the relative rigidity of the receptor flaps hinders the
perfect fit into the host cavity. The Wiberg bond order of
À
the linking C N bonds has been computed within the natu-
ral bond order (NBO)[37] framework, at the B3LYP/def2-
TZVPP level of theory, giving values of around 1.1 au,
which indicate a weak double-bond character (see Table S5
in the Supporting Information for more information). The
anion–p interactions nevertheless are enhanced by bringing
the two rings together. From the curve in Figure 8C a gain
of about 10 kJmolÀ1 can be estimated. In the case of BrÀ, no
such displacement takes place because the free state has an
optimal distance for the contact.
Insight into the nature of the binding energy was obtained
by evaluating the weight of the principal interactions in the
binding of the host–guest complex AF:XÀ, namely the hy-
drogen bond, dispersion, and ionic correlation energies. The
values are gathered in Table 3. An approach to the nature
tact, in both cases contracting
Table 3. Partitioning of the LMP2 interaction energies (in [kJmolÀ1]) and the natural population analysis
(NPA) charge transfer (qCT
)
for the anion–receptor interactions in all AF and BF anion adducts. The differ-
the equilibrium distance by
about 0.3–0.4 ꢅ. This is far
from being a minor effect, and
it cannot be attributed to classi-
cal electrostatic interactions.
However, this is in disagree-
ment with another theoretical
study, which connected the sta-
bilization of such systems to
a substituent–anion electrostatic
[a,b]
ent values per entry refer to the contribution of each separated ring.
Compound DEint
DEdisp
DEionic
DEdisp+DEionic
qCT
AF:ClÀ
À178.3 À14.8/À15.8
À91.4 À14.9/À15.3
À157.8 À15.8/À16.8
À81.9 À14.3/À15.3
À172.5 À16.0/À18.9
À93.6 À15.1/À15.0
À154.2 À17.0/À20.8
À87.3 À16.1/À16.3
À7.5/À8.6
À9.1/À8.8
À8.2/À9.1
À8.3/À9.2
À8.6/À9.8
À8.8/À8.4
À9.6/À11.5
À9.3/À9.2
À22.3/À24.4
À24.0/À24.1
À24.0/À25.9
À22.6/À24.6
À24.7/À28.8
À24.0/À23.4
À26.6/À32.3
À25.4/À25.4
0.156
AFm:ClÀ[c]
AF:BrÀ
0.131
0.121
0.100
AFm:BrÀ[c]
BF:ClÀ
BFm:ClÀ[c]
BF:BrÀ
BFm:BrÀ [c]
AF2:ClÀ
interaction.[35]
Such
energy
À349.4 À10.9/À16.1/À11.2/À15.1 À4.9/À7.5/À5.07/À6.6 À15.8/À23.7/À16.3/À21.7 0.183
AF2:BrÀ
À298.9 À16.0/À15.9/À17.6/À13.9 À8.4/À8.2/À9.3/À6.5 À24.4/À24.1/À26.9/À20.4 0.141
terms are included solely in the
HF contribution.
The results obtained for the
bent model show relatively sim-
ilar profiles (Figures 8C and
D). The minimum distances are practically coincident with
those derived from the parallel model, namely 3.2–3.4 ꢅ.
This indicates that the orientation of the rings plays a rela-
tively small role and that one can extrapolate some of the
findings in the parallel model to the compound structures, at
[a] NBO calculations were computed with B3LYP/def2-TZVPP. [b] Charge transfer is defined as qCT
qX(complex)ÀqX(isolated). [c] AFm and BFm refer to the flap model, see the main text.
=
À
À
of the N H···Cl interaction was reached by computing the
À
charge transfer (qCT) from the halide to the N H bond. The
NBO method was used to estimate the extent of this leading
interaction;[37] NBO calculations were carried out at the
B3LYP/def2-TZVPP level of theory.
least for 1508 of the bent angle ringcentroid-X-ringcentroid
.
Regarding the decomposition of the LMP2 energies, only
dispersion and ionic correlation interactions can be separat-
ed from the total interaction. These values are also given in
Table 3 for a better comparison, together with the sum of
the two previously mentioned contributions. In the case of
the complete complexes AF:ClÀ and AF:BrÀ, the DEdisp and
the DEionic values were computed by dissecting the energy
These results can help us to better understand the struc-
tures of receptor AF with the different anions. The curves in
Figures 8A and C show that the optimal distance for the
ClÀ···p interaction would be 3.2 ꢅ regardless of the disposi-
tion of both rings (the bent angle) at the LMP2 level of
theory. Strikingly, the anions stay out of the isoindole plane
Chem. Eur. J. 2013, 19, 16988 – 17000
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16995

R. A. Mata, F. Meyer et al.
terms between the orbitals in the rings and the anions
(Pipek-Mezey-localized). As it can be seen, the dispersion
interactions contribute with 17 and 21%, respectively, to the
binding energy for AF:ClÀ and AF:BrÀ. In the case of the
ionic interaction energy the contribution is even lower,
being roughly 10% of the total energy for both complexes.
It is worth noting that the total correlation energy values for
both anions are close, that is, 46.7 kJmolÀ1 for ClÀ and
49.9 kJmolÀ1 for BrÀ, pointing out that the differences in
the binding energy for each halide are already contained in
the reference energy. It is remarkable that the contributing
terms of the LMP2 energies computed here are coincident
with those computed for the parallel and bent curves, sup-
porting the good representation acquired by the model
used. Although electrostatic interactions dominate the coor-
dination of the halides to the compounds under study,[1d,2b,7b]
a significant part of the interaction is due to dispersion. Fur-
thermore, the bent structure of the rings is a direct conse-
quence of correlation effects between the diffuse electron
cloud of the anion and the ring system, shifting the interac-
tion potential of about 0.3–0.4 ꢅ to smaller contact distan-
ces.
Figure 9. Comparison of the optimized geometries for the AF2 :XÀ adduct
(in red) and the two overlapped relaxed AF monomers (in blue) in the
case of XÀ =chloride (top) and for XÀ = bromide (bottom). For further
views see the Supporting Information.
On the other hand, the values of the interaction NH···XÀ
seem to be strong contributors to the binding process. The
values shown here are within the range expected, given the
electron-withdrawing character of the perfluoroarene sub-
stituents of the isoindole fragment.[32,36c,38] Thus, in order to
been attributed to the extra interaction between arene
rings.[25] In the present case, this effect could be due to the
fact that each AF molecule hinders the other in its interac-
tion with the anion. This effect is higher for chloride com-
plexes because each receptor ring has to move closer to the
halide, as is reflected by the structural changes upon dimer
formation. For instance, the hydrogen bonds are longer in
the 2:1 complex than in the 1:1 complex by 0.070 ꢅ for ClÀ
and by 0.126 ꢅ for BrÀ, and the angles of the flap (C-N-C)
are widened by about 7 and 28 for ClÀ and BrÀ, respectively.
The radii of the chloride and bromide ions seem to be some-
what small for the pocket formed between the two host mol-
ecules; consequently they interact strongly with two of the
arenes and weakly with the remaining ones. The data in
Table 3 clearly show the different strength of the anion–p
interaction with the four flaps, and also that they are on
average slightly lower for the dimers than for the monomers.
Bulkier receptor molecules might completely avoid dimer
formation.
À
get an idea of the importance of N H bonds, we have built
a simplified model based on the good representations of the
pentafluorobenzene, keeping the position of the flaps as
they are in the host–guest complex but erasing the isoindole
moiety (or the carbazole ring); these model structures are
referred to in Table 3 as AFm:XÀ (or BFm:XÀ). The binding
energy obtained in this way revealed that the anion–p inter-
action contributes at least 53% of the total binding energy
in AF:XÀ and BF:XÀ, whereas the remainder is due to the
NH···XÀ hydrogen bonds. The energies obtained reveal
a pronounced difference of the interaction depending on the
nature of the halide, that is, 86.9 and 75.9 kJmolÀ1 for chlo-
ride and bromide ions, respectively. In the same way the
amount of charge transfer (CT) is higher for ClÀ than BrÀ,
which is in good agreement with previous reports.[32] The CT
is in the order of 0.1 electrons, which is mainly due to the
NH···XÀ bond.
As in the case of complexes AF:XÀ, the structures of com-
pounds BF:XÀ have been optimized and their interaction en-
ergies are compared in Table 2. The B3LYP-D3/def2-
TZVPP geometries are given in Figure 10 (and in Figure S8
in the Supporting Information).
Upon halide binding these structures show a larger differ-
ence in the ring distances (see Table S4 in the Supporting
Information). When compared to compound AF, the rings
move closer together and the difference between the two
halides is accentuated. As can be seen in Figure 10 the main
difference between the relaxed geometry of receptor BF and
the one in the BF:XÀ complexes is due to the tilted angle of
the flaps, the change is from 43.8/48.9 to 63.6/71.9 and 62.6/
72.08 for chloride and bromide (in that order). The better fit
It was also possible to obtain stable minima geometries
for the AF2 :XÀ complexes. Figure 9 shows superposition plots
comparing the computed AF2 :XÀ complexes (for XÀ =ClÀ
and BrÀ) with the free receptor structure (see Figures S5–S7
in the Supporting Information for further details).
The interaction energies for the complexes AF2 :XÀ are
given in Table 2. From a straightforward comparison be-
tween the interaction energies in the case of the monomers
and the dimers, one can observe that there is no additivity
effect on the binding energy. Even more, the total dimer in-
teraction is lower than the sum of two monomers by about
29.8 and 16.7 kJmolÀ1 for ClÀ and BrÀ, respectively
(B3LYP-D3). Non-additivity effects on 3:1 complexes have
16996
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 16988 – 17000

Preorganized Anion Traps for Exploiting Anion–p Interactions
FULL PAPER
guest complexes—the latter are rare examples of crystallo-
graphically characterized anion–p complexes with un-
charged receptors. The anion affinity is a result of favorable
anion–p interactions with the pentafluorophenyl groups in
complexes AF:XÀ and BF:XÀ, supported by the more direc-
tional hydrogen bond to the NH group in the central skele-
ton, and consequently the non-fluorinated derivatives AN
and BH are not capable of hosting any halide ions under the
conditions used in the present work.
In the case of AF2 :BrÀ, the anion is embraced by two host
molecules, possibly due to the larger size of the anion and
a slightly larger opening of the flanking rings. The relative
position to the C6F5 planes is within typical ranges for
anion–p interactions. In AF2 :BrÀ the anion is somewhat dis-
placed to the edge of the ring, forming an h2-type interac-
tion. Binding constants have also been measured for com-
pound BF, comparing the affinity to chloride and bromide
ions. The expected trends are observed, with chloride re-
vealing a stronger binding and with lower affinities for in-
creasing solvent polarity.
Detailed electronic structure calculations have been car-
ried out on the compounds in order to obtain further insight
into the specific anion interactions. Through the use of
a model consisting of two pentafluorobenzene rings, we
were able to decompose the interaction between the recep-
tors and the anions. Comparison of the interaction energies
in the full host–guest complex and the model shows that
direct anion interactions with the flanking electron-deficient
arenes can contribute as much as half of the total effect.
The adequacy of our model has been confirmed by analyz-
ing different conformations and comparing the correlation
energy contributions from localized orbitals found in the
arene flaps and the two isolated rings. Through the use of
local orbital spaces, we were also able to dissect the effect
of electronic correlation in the structure of the host–guest
complexes. Dispersion interactions contribute as much as
20% to the stabilization energy, with a slightly larger contri-
bution in the case of bromide. All of these results underline
the importance of the electron-deficient arenes in promoting
the anion binding. Low deformation energies upon complex-
ation show that the synthesized molecules are in fact close
to an ideal conformation to trap the halides within their
cleft. This observation is also supported by comparing the
optimal anion–ring distances in the potential energy profiles
and the distances measured both in the structures deter-
mined crystallographically and optimized computationally.
Overall this study shows that efficient anion binding can
be achieved in well preorganized receptors that exploit the
synergetic effect of directional hydrogen-bonding and
anion–p interactions. The latter contribute significantly to
the overall binding energies, confirming that anion–p inter-
actions are useful instruments in the design of uncharged
anion receptors that do not rely on salt bridges and require
Figure 10. Superposition plots with a least-square fit of the computed
structures for the free BF receptor and its complexes with chloride (top)
and bromide (bottom). The structure in blue is the free receptor and the
structures in red or gray are the structures of the adducts. The optimiza-
tions were performed at the B3LYP-D3/def2-TZVPP level of theory. For
further views see the Supporting Information.
into the host pocket is a consequence of a certain freedom
À
in the C C bond linker rotation; the computed Wiberg
bond orders reveal a single-bond character as expected
(Table S5 in the Supporting Information). However, as
shown in Table 2, this has no reflection in the interaction en-
ergies, which are lower than the energies obtained with the
receptor AF. The deformation energies are very similar. A
better interpretation can be drawn from the decompositions
of the energies given in Table 3. At the first glance it be-
comes obvious that the better fit means a small increase in
the anion–p interaction for receptor BF. Comparison of the
modeled flaps leads to an increase of this interaction of
about 2.2 and 5.4 kJmolÀ1 with ClÀ and BrÀ, respectively. In
contrast, the hydrogen-bond interaction is strongly reduced
in compound BF, and seems to be the mandatory interaction.
Thus, the charge transfer is lower for both halide anions.
This result follows the trend dictated by the acidity of the
heterocycle.
Conclusion
We have presented a combined experimental and computa-
tional study of two novel neutral anion receptors, that is,
compounds AF and BF, which are highly preorganized for ac-
commodating halide ions in a cleft between two pentafluor-
ophenyl rings. The propensity of the receptors to bind chlo-
ride and bromide was confirmed in solution through NMR
spectroscopy and in the crystal structures of several host–
À
only few complementary interactions, such as a single N
H···XÀ hydrogen bond in the present case.
Chem. Eur. J. 2013, 19, 16988 – 17000
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16997

R. A. Mata, F. Meyer et al.
of K2CO3 (2m, 50 mL) were added and the mixture was degassed 3–5
times. Subsequently bromopentafluorobenzene (2 mL, 16.0 mmol) was
added under a nitrogen atmosphere and the mixture was stirred at 908C
for 16 h. The solvent was removed and the crude product was re-dis-
solved in dichloromethane (50 mL). After filtration over a short silica
column the solvent was removed and the product was obtained as
a white solid (2.14 g, 85%). Kugelrohr distillation of the crude product
yielded white crystals, which were then used for further experiments.
1H NMR (300 MHz, CDCl3): d=8.24 (d, J=1.7 Hz, 2H), 7.45 (s, 2H),
7.40 (s, 1H; NH), 1.48 ppm (s, 18H); 13C NMR (75 MHz, CDCl3): d=
143.49, 136.44, 126.78, 124.60, 118.29, 108.64, 34.96, 32.05 ppm; 19F NMR
(282 MHz, CDCl3): d=À139.80 (dd, J=22.7, 7.8 Hz, 4F), À154.31 (t, J=
21.0 Hz, 2F), À161.02 ppm (td, J=22.3, 7.6 Hz, 4F); MS (EI): m/z (%):
611 (46) [M]+, 596 (100).
3,6-Di-tert-butyl-1,8-diphenyl-9H-carbazole (BH):[19,20] The synthesis was
done as described for receptor BF, but using four equivalents of bromo-
benzene instead. After removal of the solvent the solid was dissolved in
dichloromethane/EtOH and left to crystallize at 88C. The product was
obtained as pale yellow crystals (0.17 g, 70%, starting with 0.30 g
(0.56 mmol) of 3). 1H NMR (300 MHz, CDCl3): d=8.35 (s, 1H; NH),
8.14 (d, J=1.8 Hz, 2H), 7.75–7.63 (m, 4H), 7.60–7.46 (m, 6H), 7.41 (ddd,
J=7.3, 3.9, 1.3 Hz, 2H), 1.53 ppm (s, 18H); 13C NMR (75 MHz, CDCl3):
d=141.97, 138.61, 134.73, 128.18, 127.16, 126.30, 123.24, 123.00, 122.86,
114.60, 33.79, 31.05 ppm; MS (EI): m/z (%): 431 (79) [M]+, 416 (100).
Experimental Section
Physical measurements: NMR spectra were recorded on an Advance III
300 MHz spectrometer (Bruker) by using the indicated deuterated sol-
vent as internal standard. EI-MS spectra were recorded on a Finnigan
MAT 8200 spectrometer and ESI-MS spectra were recorded on an Ap-
plied Biosystems API 2000 spectrometer. Experimental procedures and
data analysis for the Job plots and titration experiments are provided in
the Supporting Information.
X-ray crystallography: X-ray data were collected on a STOE IPDS II dif-
fractometer with an area detector (graphite monochromated MoKa radia-
tion, l=0.71073 ꢅ) by using w scans at 133 K (Table S6 in the Supporting
Information). The structures were solved by direct methods and refined
on F2 by using all reflections with SHELX-97.[39] Most non-hydrogen
atoms were refined anisotropically. Most hydrogen atoms were placed in
calculated positions and assigned to an isotropic displacement parameter
of 1.2/1.5 Ueq (C or N). The hydrogen atoms H1 and H2 in AF2 :BrÀ and
AN were refined without any restraints or constraints; a fixed isotropic
displacement parameter of 0.08 ꢅ2 for the nitrogen-bound hydrogen
atoms was applied in case of BF:DMSO, AF:ClÀ, and BF. Disordered
DMSO solvent molecules are present in AF:DMSO (occupancy factors:
0.816(3)/0.184(3)) and BF:DMSO (fixed occupancy of 0.5 and 0.2818(17)/
0.2182(17)), and a disordered toluene molecule in BF (fixed occupancy of
0.5). One of the tBu groups in BF is disordered as well (occupancy fac-
tors: 0.747(4)/0.253(4)). EADP constraints (AF:DMSO, BF:DMSO),
DFIX, SIMU, DELU, and ISOR restraints (BF) were applied to model
the disorder. Face-indexed absorption corrections were performed nu-
merically with the program X-RED.[40]
Computational details: All geometry optimizations were carried out by
using the hybrid density functional B3LYP-D3[22a–c] with the basis set
def2-TZVPP.[22d] The RIJCOSX method was applied to speed up the cal-
culations.[41] The stationary points were located with the quasi-Newton al-
gorithm by using redundant internal coordinates. Hessians were comput-
ed to determine the nature of stationary points. These theoretical calcula-
tions were performed with the ORCA program package.[42]
CCDC-964440, -964441, -964442, -964443 -964444 and -964445 contain
the supplementary crystallographic data for this paper. These data can be
obtained free of charge from The Cambridge Crystallographic Data
Centre via www.ccdc.cam.ac.uk/data_request/cif
We performed single-point calculations on the B3LYP-D3-optimized
structures with second-order local Møller–Plesset perturbation theory
(LMP2)[23a–c] by employing the MOLPRO 2012.1[43] software program
package. Density fitting (DF) approximations have been used in this
local method.[23c] The aug-cc-pVTZ basis set was used for carbon, nitro-
gen, fluorine, chlorine, and bromine atoms whereas for hydrogen atoms
the cc-pVTZ basis set was used.[24] In the density fitting calculations re-
ported in this paper, we used the aug-cc-pVTZ/JKJIT and aug-cc-pVTZ/
MP2FIT auxiliary fitting basis sets[44] in the DF-HF and DF-LMP2 calcu-
lations, respectively.
Materials and synthetic procedures: All reactions were carried out under
a nitrogen atmosphere. n-Butanol was dried over sodium to remove
excess of water. Starting materials and solvents were purchased either
from abcr, Sigma Aldrich, or Acros.
1,3-Bis(pentafluorophenylimino)isoindoline
(AF):[18a]
Phthalonitrile
(1.28 g, 10 mmol), pentafluoroaniline (3.66 g, 40 mmol) and CaCl2 (0.05 g,
0.5 mmol) were heated to reflux in n-butanol (5 mL) for twenty days.
After removal of the solvent, hexane was added to the residue. Filtration
of the mixture gave a yellow solution. After removal of the solvent, the
residue was purified through column chromatography (ether/hexane 1:1)
yielding a pale yellow solid (1.08 g, 23%). 1H NMR (300 MHz, CDCl3):
d=8.13 (d, J=7.7 Hz, 1H; Ha), 8.05–8.01 (m, 2H; Hb), 7.82–7.75 (m, 2H;
Hb), 7.74 (td, J=7.6, 0.7 Hz, 1H; Ha), 7.62 (s, 1H; NHa), 7.57 (td, J=7.7,
1.0 Hz, 1H; Ha), 7.08 (d, J=7.8 Hz, 1H; Ha), 6.92 ppm (s, 1H; NHb);
19F NMR (282 MHz, CDCl3): d=À149.26 (dd, J=23.0, 5.8 Hz, 2F; Fa),
À149.40 (dd, J=22.5, 5.9 Hz, 4F; Fb), À151.13–À151.36 (m, 2F; Fa),
À160.15 (t, J=21.5 Hz, 2F; Fb), À160.49 (t, J=21.4 Hz, 1F; Fa),
À161.48–À161.76 (m, 3F; Fa, Fb), À161.88 (td, J=22.1, 6.4 Hz, 2F; Fa),
À162.15 ppm (td, J=21.4, 5.4 Hz, 2F; Fa); MS (EI): m/z (%): 477 (100)
[M]+, 295 (47), 438 (32), 458 (89).
1,3-Bis(2-pyrimidylimino)isoindoline (AN):[18] Phthalonitrile (0.64 g,
5 mmol), 2-aminopyrimidine (1.90 g, 20 mmol), and CaCl2 (0.05 g,
0.5 mmol) were dissolved in n-butanol (5 mL) and heated to reflux for
21 d. After evaporation of the solvent first byproducts were removed
through column chromatography (dichloromethane/methanol 19:1). To
further purify the product, Kugelrohr distillation was utilized. The prod-
uct was obtained as a yellow solid (0.1 g, 7%). 1H NMR (300 MHz,
CDCl3): d=13.47 (s, 1H; NH), 8.84 (d, J=4.8 Hz, 4H), 8.26–8.16 (m,
2H), 7.75–7.65 (m, 2H), 7.12 ppm (t, J=4.8 Hz, 2H); MS (EI): m/z (%):
301 (100) [M]+, 79 (27), 207 (38), 222 (33).
The LMP2 calculations were carried out by using Pipek-Mezey-localized
orbitals.[45] The domains were determined with the use of natural popula-
tion analysis (NPA) criteria, with TNPA =0.03.[46]
The NPA charge analysis[37a,c–e] and the reported Wiberg bond indices
were computed by using the GENNBO 5.9 program.[47]
Acknowledgements
Financial support by the Georg-August-University is gratefully acknowl-
edged. D.M.A. gratefully acknowledges a postdoctoral fellowship from
ARCOIRIS Erasmus Mundus Action 2.
[1] a) B. L. Schottel, H. T. Chifotides, K. R. Dunbar, Chem. Soc. Rev.
2008, 37, 68–83; b) O. B. Berryman, D. W. Johnson, Chem.
Commun. 2009, 3143–3153; c) A. Robertazzi, F. Krull, E. W. Knapp,
P. Gamez, CrystEngComm 2011, 13, 3293–3300; d) A. Frontera, P.
Gamez, M. Mascal, T. J. Mooibroek, J. Reedijk, Angew. Chem. 2011,
123, 9736–9756; Angew. Chem. Int. Ed. 2011, 50, 9564–9583.
[2] a) B. P. Hay, R. Custelcean, Cryst. Growth Des. 2009, 9, 2539–2545;
b) T. J. Mooibroek, C. A. Black, P. Gamez, J. Reedijk, Cryst. Growth
Des. 2008, 8, 1082–1093.
3,6-Di-tert-butyl-1,8-bis(perfluorophenyl)-9H-carbazole (BF): The boro-
nated precursor was synthesized following literature procedures.[19–20] 3,6-
Di-tert-butyl-1,8-bis(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-9H-car-
bazole 3 (2.18 g, 4.1 mmol) and [Pd(PPh3)4] (94.8 mg, 0.8 mmol) were dis-
solved in toluene (500 mL). Ethanol (200 mL) and an aqueous solution
[3] a) E. A. Meyer, R. K. Castellano, F. Diederich, Angew. Chem. 2003,
115, 1244–1287; Angew. Chem. Int. Ed. 2003, 42, 1210–1250;
b) E. A. Meyer, R. K. Castellano, F. Diederich, Angew. Chem. 2003,
16998
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 16988 – 17000

Preorganized Anion Traps for Exploiting Anion–p Interactions
FULL PAPER
115, 4254; Angew. Chem. Int. Ed. 2003, 42, 4120; c) B. P. Hay, V. S.
Bryantsev, Chem. Commun. 2008, 2417–2428.
[4] a) H.-J. Schneider, F. Werner, T. Blatter, J. Phys. Org. Chem. 1993,
6, 590–594; b) H. J. Schneider, Angew. Chem. 1991, 103, 1419–1439;
Angew. Chem. Int. Ed. Engl. 1991, 30, 1417–1436; c) K. Hiraoka, S.
Mizuse, S. Yamabe, J. Phys. Chem. 1987, 91, 5294–5297.
[5] a) M. Mascal, A. Armstrong, M. D. Bartberger, J. Am. Chem. Soc.
2002, 124, 6274–6276; b) I. Alkorta, I. Rozas, J. Elguero, J. Am.
Chem. Soc. 2002, 124, 8593–8598; c) D. QuiÇonero, C. Garau, A.
Frontera, P. Ballester, A. Costa, P. M. Deyꢆ, Chem. Phys. Lett. 2002,
359, 486–492; d) D. QuiÇonero, C. Garau, C. Rotger, A. Frontera,
P. Ballester, A. Costa, P. M. Deyꢆ, Angew. Chem. 2002, 114, 3539–
3542; Angew. Chem. Int. Ed. 2002, 41, 3389–3392.
[23] a) P. Pulay, Chem. Phys. Lett. 1983, 100, 151–154; b) S. Sæbø, P.
Pulay, Chem. Phys. Lett. 1985, 113, 13–18; c) H.-J. Werner, F. R.
Manby, P. J. Knowles, J. Chem Phys. 2003, 118, 8149–8160; d) H.-J.
Werner, M. Schꢀtz, J. Chem. Phys. 2011, 135, 144166.
[24] a) T. H. Dunning Jr., J. Chem. Phys. 1989, 90, 1007–1023; b) D. E.
Woon, T. H. Dunning Jr., J. Chem. Phys. 1993, 98, 1358–1371.
[25] C. Garau, D. QuiÇonero, A. Frontera, P. Ballester, A. Costa, P. M.
Deyꢆ, J. Phys. Chem. A 2005, 109, 9341–9345.
[26] a) C. Garau, A. Frontera, D. QuiÇonero, P. Ballester, A. Costa, P. M.
Deyꢆ, Chem. Phys. Lett. 2003, 382, 534–540; b) C. Garau, D. QuiÇo-
nero, A. Frontera, P. Ballester, A. Costa, P. M. Deyꢆ, New J. Chem.
2003, 27, 211–214; c) C. Garau, A. Frontera, P. Ballester, D. QuiÇo-
nero, A. Costa, P. M. Deyꢆ, Eur. J. Org. Chem. 2005, 179–183; d) D.
QuiÇonero, C. Garau, A. Frontera, P. Ballester, A. Costa, P. M.
Deyꢆ, J. Phys. Chem. A 2005, 109, 4632–4637; e) A. Frontera, D.
QuiÇonero, C. Garau, A. Costa, P. Ballester, P. M. Deyꢆ, J. Phys.
Chem. A 2006, 110, 9307–9309; f) D. QuiÇonero, A. Frontera, C.
Garau, P. Ballester, A. Costa, P. M. Deyꢆ, ChemPhysChem 2006, 7,
2487–2491; g) A. Frontera, D. QuiÇonero, A. Costa, P. Ballester,
P. M. Deyꢆ, New J. Chem. 2007, 31, 556–560; h) C. Garau, D. Qui-
Çonero, A. Frontera, D. Escudero, P. Ballester, A. Costa, P. M.
Deyꢆ, Chem. Phys. Lett. 2007, 438, 104–108; i) D. QuiÇonero, A.
Frontera, D. Escudero, P. Ballester, A. Costa, P. M. Deyꢆ, Chem-
PhysChem 2007, 8, 1182–1187; j) D. Kim, P. Tarakeshwar, K. S.
Kim, J. Phys. Chem. A 2004, 108, 1250–1258.
[6] S. Demeshko, S. Dechert, F. Meyer, J. Am. Chem. Soc. 2004, 126,
4508–4509.
[7] a) Y. S. Rosokha, S. V. Lindeman, S. V. Rosokha, J. K. Kochi,
Angew. Chem. 2004, 116, 4750–4752; Angew. Chem. Int. Ed. 2004,
43, 4650–4652; b) P. De Hoog, P. Gamez, I. Mutikainen, U. Turpei-
nen, J. Reedijk, Angew. Chem. 2004, 116, 5939–5941; Angew. Chem.
Int. Ed. 2004, 43, 5815–5817; c) A. Frontera, F. Saczewski, M. Gda-
niec, E. Dziemidowicz-Borys, A. Kurland, P. M. Deyꢆ, D. QuiÇo-
nero, C. Garau, Chem. Eur. J. 2005, 11, 6560–6567; d) M. Mascal, I.
Yakovlev, E. B. Nikitin, J. C. Fettinger, Angew. Chem. 2007, 119,
8938–8940; Angew. Chem. Int. Ed. 2007, 46, 8782–8784.
[8] S. Guha, S. Saha, J. Am. Chem. Soc. 2010, 132, 17674–17677.
[9] a) J. Mareda, S. Matile, Chem. Eur. J. 2009, 15, 28–37; b) A. Vargas
Jentzsch, D. Emery, J. Mareda, P. Metrangolo, G. Resnati, S. Matile,
Angew. Chem. 2011, 123, 11879–11882; Angew. Chem. Int. Ed.
2011, 50, 11675–11678; c) A. Vargas Jentzsch, A. Hennig, J.
Mareda, S. Matile, Acc. Chem. Res. 2013, DOI: 10.1021/ar400014r.
[10] a) D. QuiÇonero, A. Frontera, P. M. Deyꢆ in Anion Coordination
Chemistry, Wiley-VCH, Weinheim, 2011, pp. 321–361; b) H. T. Chi-
fotides, K. R. Dunbar, Acc. Chem. Res. 2013, 46, 894–906.
[11] a) C. Estarellas, A. Frontera, D. QuiÇonero, P. M. Deyꢆ, Chem.
Asian J. 2011, 6, 2316–2318; b) C. Estarellas, A. Frontera, D. QuiÇo-
nero, P. M. Dey¼, Angew. Chem. 2011, 123, 435–438; Angew. Chem.
Int. Ed. 2011, 50, 415–418.
[12] a) G. Gil-Ramꢇrez, E. C. Escudero-Adꢈn, J. Benet-Buchholz, P. Bal-
lester, Angew. Chem. 2008, 120, 4182–4186; Angew. Chem. Int. Ed.
2008, 47, 4114–4118; b) P. Arranz-Mascarꢉs, C. Bazzicalupi, A.
Bianchi, C. Giorgi, M.-L. Godino-Salido, M.-D. Gutiꢊrrez-Valero, R.
Lopez-Garzꢉn, M. Savastano, J. Am. Chem. Soc. 2013, 135, 102–
105; c) P. Ballester, Acc. Chem. Res. 2013, 46, 874–884.
[13] a) C. Garau, A. Frontera, D. QuiÇonero, P. Ballester, A. Costa, P. M.
Deyꢆ, ChemPhysChem 2003, 4, 1344–1348; b) C. Garau, A. Fron-
tera, D. QuiÇonero, P. Ballester, A. Costa, P. M. Deyꢆ, J. Phys.
Chem. A 2004, 108, 9423–9427.
[14] O. B. Berryman, F. Hof, M. J. Hynes, D. W. Johnson, Chem.
Commun. 2006, 506–508.
[15] M. Giese, M. Albrecht, T. Krappitz, M. Peters, V. Gossen, G. Raabe,
A. Valkonen, K. Rissanen, Chem. Commun. 2012, 48, 9983–9985.
[16] a) M. Albrecht, C. Wessel, M. De Groot, K. Rissanen, A. Lꢀchow, J.
Am. Chem. Soc. 2008, 130, 4600–4601; b) M. Giese, M. Albrecht, K.
Wiemer, A. Valkonen, K. Rissanen, New J. Chem. 2012, 36, 1368–
1372.
[27] A. Clements, M. Lewis, J. Phys. Chem. A 2006, 110, 12705–12710.
[28] a) J. Hernꢈndez-Trujillo, A. Vela, J. Phys. Chem. 1996, 100, 6524–
6530; b) J. Hernꢈndez-Trujillo, M. Costas, A. Vela, J. Chem. Soc.
Faraday Trans. 1993, 89, 2441–2443; c) J. R. Trudell, Biophys. Chem.
1998, 73, 7–11; d) R. J. Doerksen, A. J. Thakkar, J. Phys. Chem. A
1999, 103, 10009–10014; e) A. Alparone, J. Fluorine Chem. 2012,
144, 94–101.
[29] M. Giese, M. Albrecht, C. Bannwarth, G. Raabe, A. Valkonen, K.
Rissanen, Chem. Commun. 2011, 47, 8542–8544.
[30] S. Grimme, J. Chem. Phys. 2003, 118, 9095–9102.
[31] a) J. G. Hill, J. A. Platts, H. J. Werner, Phys. Chem. Chem. Phys.
2006, 8, 4072–4078; b) S. Grimme, Chem. Eur. J. 2004, 10, 3423–
3429; c) M. Piacenza, S. Grimme, J. Am. Chem. Soc. 2005, 127,
14841–14848; d) M. Piacenza, S. Grimme, ChemPhysChem 2005, 6,
1554–1558; e) T. Takatani, C. D. Sherrill, Phys. Chem. Chem. Phys.
2007, 9, 6106–6114; f) T. Van Mourik, A. K. Wilson, T. H. Dun-
ning Jr., Mol. Phys. 1999, 96, 529–547.
[32] O. B. Berryman, V. S. Bryantsev, D. P. Stay, D. W. Johnson, B. P. Hay,
J. Am. Chem. Soc. 2007, 129, 48–58.
[33] M. Schꢀtz, G. Rauhut, H. J. Werner, J. Phys. Chem. A 1998, 102,
5997–6003.
[34] C. Garau, A. Frontera, D. QuiÇonero, P. Ballester, A. Costa, P. M.
Deyꢆ, Chem. Phys. Lett. 2004, 392, 85–89.
[35] S. E. Wheeler, K. N. Houk, J. Phys. Chem. A 2010, 114, 8658–8664.
[36] a) M. Albrecht, M. Mꢀller, O. Mergel, K. Rissanen, A. Valkonen,
Chem. Eur. J. 2010, 16, 5062–5069; b) M. Mꢀller, M. Albrecht, J.
Sackmann, A. Hoffmann, F. Dierkes, A. Valkonen, K. Rissanen,
Dalton Trans. 2010, 39, 11329–11334; c) V. S. Bryantsev, B. P. Hay, J.
Am. Chem. Soc. 2005, 127, 8282–8283; d) B. P. Hay, T. K. Firman,
B. A. Moyer, J. Am. Chem. Soc. 2005, 127, 1810–1819; e) O. B. Ber-
ryman, A. C. Sather, B. P. Hay, J. S. Meisner, D. W. Johnson, J. Am.
Chem. Soc. 2008, 130, 10895–10897; f) L. Pedzisa, B. P. Hay, J. Org.
Chem. 2009, 74, 2554–2560; g) B. P. Hay, Chem. Soc. Rev. 2010, 39,
3700–3708; h) M. Schnopp, S. Ernst, G. Haberhauer, Eur. J. Org.
Chem. 2009, 213–222; i) D. W. Yoon, D. E. Gross, V. M. Lynch, J. L.
Sessler, B. P. Hay, C. H. Lee, Angew. Chem. 2008, 120, 5116–5120;
Angew. Chem. Int. Ed. 2008, 47, 5038–5042.
[17] M. Giese, M. Albrecht, K. Wiemer, G. Kubik, A. Valkonen, K. Ris-
sanen, Eur. J. Inorg. Chem. 2012, 2995–2999.
[18] a) W. O. Siegl, J. Org. Chem. 1977, 42, 1872–1878; b) D. M. Baird,
K. Y. Shih, Polyhedron 1991, 10, 229–235.
[19] V. C. Gibson, S. K. Spitzmesser, A. J. P. White, D. J. Williams, Dalton
Trans. 2003, 2718–2727.
[20] L. Arnold, H. Norouzi-Arasi, M. Wagner, V. Enkelmann, K. Mꢀllen,
Chem. Commun. 2011, 47, 970–972.
[37] a) J. P. Foster, F. Weinhold, J. Am. Chem. Soc. 1980, 102, 7211–
7218; b) A. E. Reed, F. Weinhold, J. Chem. Phys. 1983, 78, 4066–
4073; c) A. E. Reed, F. Weinhold, J. Chem. Phys. 1985, 83, 1736–
1740; d) A. E. Reed, R. B. Weinstock, F. Weinhold, J. Chem. Phys.
1985, 83, 735–746; e) A. E. Reed, L. A. Curtiss, F. Weinhold, Chem.
Rev. 1988, 88, 899–926.
[21] K. A. Connors, Binding Constants, Wiley, New York, 1987.
[22] a) A. D. Becke, Phys. Rev. A 1988, 38, 3098–3100; b) C. Lee, W.
Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785–789; c) S. Grimme, J.
Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 2010, 132, 154104; d) F.
Weigend, R. Ahlrichs, Phys. Chem. Chem. Phys. 2005, 7, 3297–3305.
[38] V. S. Bryantsev, B. P. Hay, Org. Lett. 2005, 7, 5031–5034.
Chem. Eur. J. 2013, 19, 16988 – 17000
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16999

R. A. Mata, F. Meyer et al.
[39] G. Sheldrick, Acta Crystallogr. Sect. A 2008, 64, 112–122.
[40] X-RED, STOE & CIE, Darmstadt (Germany), 2002.
[41] a) R. Ahlrichs, Phys. Chem. Chem. Phys. 2004, 6, 5119–5121; b) F.
Neese, F. Wennmohs, A. Hansen, U. Becker, Chem. Phys. 2009, 356,
98–109; c) F. Neese, J. Comput. Chem. 2003, 24, 1740–1747.
[42] F. Neese, Comput. Mol. Sci. 2012, 2, 73–78.
[43] H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby, M. Schꢀtz, P.
Celani, T. Korona, R. Lindh, A. Mitrushenkov, G. Rauhut, K. R.
Shamasundar, T. B. Adler, R. D. Amos, A. Bernhardsson, A. Bern-
ing, D. L. Cooper, M. J. O. Deegan, A. J. Dobbyn, F. Eckert, E.
Goll, C. Hampel, A. Hesselmann, G. Hetzer, T. Hrenar, G. Jansen,
C. Kçppl, Y. Liu, A. W. Lloyd, R. A. Mata, A. J. May, S. J. McNicho-
las, W. Meyer, M. E. Mura, A. Nicklass, D. P. OꢄNeill, P. Palmieri, D.
Peng, K. Pflꢀger, R. Pitzer, M. Reiher, T. Shiozaki, H. Stoll, A. J.
Stone, R. Tarroni, T. Thorsteinsson, M. Wang, http://www.molpro.-
net, 2012.
[44] a) F. Weigend, Phys. Chem. Chem. Phys. 2002, 4, 4285–4291; b) F.
Weigend, A. Kçhn, C. Hꢁttig, J. Chem. Phys. 2002, 116, 3175.
[45] J. Pipek, P. G. Mezey, J. Chem. Phys. 1989, 90, 4916–4926.
[46] R. A. Mata, H.-J. Werner, Mol. Phys. 2007, 125, 2753–2761.
[47] E. D. Glendening, J. K. Badenhoop, A. E. Reed, J. E. Carpenter,
J. A. Bohmann, C. M. Morales, F. Weinhold, http://www.chem.wisc.
edu/~nbo5, Theoretical Chemistry Institute, University of Wiscon-
sin, Madison, WI, 2009.
Received: July 4, 2013
Published online: November 6, 2013
17000
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 16988 – 17000