Host perturbation in a β-hydroquinone clathrate studied by combined X-ray/neutron charge-density analysis: Implications for molecular inclusion in supramolecular entities

Article abstract of DOI:10.1002/chem.201400129

X-ray/neutron (X/N) diffraction data measured at very low temperature (15 K) in conjunction with ab initio theoretical calculations were used to model the crystal charge density (CD) of the host-guest complex of hydroquinone (HQ) and acetonitrile. Due to pseudosymmetry, information about the ordering of the acetonitrile molecules within the HQ cavities is present only in almost extinct, very weak diffraction data, which cannot be measured with sufficient accuracy even by using the brightest X-ray and neutron sources available, and the CD model of the guest molecule was ultimately based on theoretical calculations. On the other hand, the CD of the HQ host structure is well determined by the experimental data. The neutron diffraction data provide hydrogen anisotropic thermal parameters and positions, which are important to obtain a reliable CD for this light-atom-only crystal. Atomic displacement parameters obtained independently from the X-ray and neutron diffraction data show excellent agreement with a |ΔU| value of 0.00058 A² indicating outstanding data quality. The CD and especially the derived electrostatic properties clearly reveal increased polarization of the HQ molecules in the host-guest complex compared with the HQ molecules in the empty HQ apohost crystal structure. It was found that the origin of the increased polarization is inclusion of the acetonitrile molecule, whereas the change in geometry of the HQ host structure following inclusion of the guest has very little effect on the electrostatic potential. The fact that guest inclusion has a profound effect on the electrostatic potential suggests that nonpolarizable force fields may be unsuitable for molecular dynamics simulations of host-guest interaction (e.g., in protein-drug complexes), at least for polar molecules. Electron-density analysis by low-temperature X-ray/neutron diffraction and ab initio calculations revealed that inclusion of acetonitrile in a hydroquinone clathrate leads to substantial polarization of the host structure (see figure). Thus, doubt is cast on the accuracy of nonpolarizable force fields used in molecular dynamics simulations.

Full text of DOI:10.1002/chem.201400129

DOI: 10.1002/chem.201400129
Full Paper
&
Clathrates
Host Perturbation in a b-Hydroquinone Clathrate Studied by
Combined X-ray/Neutron Charge-Density Analysis: Implications
for Molecular Inclusion in Supramolecular Entities
Henrik F. Clausen,[a] Mads R. V. Jørgensen,[a] Simone Cenedese,[b, e] Mette S. Schmøkel,[a]
Mogens Christensen,[a] Yu-Sheng Chen,[c] George Koutsantonis,[d] Jacob Overgaard,*[a]
Mark A. Spackman,*[d] and Bo B. Iversen*[a]
Abstract: X-ray/neutron (X/N) diffraction data measured at
very low temperature (15 K) in conjunction with ab initio
theoretical calculations were used to model the crystal
charge density (CD) of the host–guest complex of hydroqui-
none (HQ) and acetonitrile. Due to pseudosymmetry, infor-
mation about the ordering of the acetonitrile molecules
within the HQ cavities is present only in almost extinct, very
weak diffraction data, which cannot be measured with suffi-
cient accuracy even by using the brightest X-ray and neu-
tron sources available, and the CD model of the guest mole-
cule was ultimately based on theoretical calculations. On the
other hand, the CD of the HQ host structure is well deter-
mined by the experimental data. The neutron diffraction
data provide hydrogen anisotropic thermal parameters and
positions, which are important to obtain a reliable CD for
this light-atom-only crystal. Atomic displacement parameters
obtained independently from the X-ray and neutron diffrac-
tion data show excellent agreement with a jDUj value of
0.00058 ꢀ2 indicating outstanding data quality. The CD and
especially the derived electrostatic properties clearly reveal
increased polarization of the HQ molecules in the host–
guest complex compared with the HQ molecules in the
empty HQ apohost crystal structure. It was found that the
origin of the increased polarization is inclusion of the aceto-
nitrile molecule, whereas the change in geometry of the HQ
host structure following inclusion of the guest has very little
effect on the electrostatic potential. The fact that guest
inclusion has a profound effect on the electrostatic potential
suggests that nonpolarizable force fields may be unsuitable
for molecular dynamics simulations of host–guest interaction
(e.g., in protein–drug complexes), at least for polar
molecules.
Introduction
opment of a huge number of advanced materials (e.g., for gas
storage, sensors, nonlinear optics),[1–3] it is the origin of self-as-
sembly processes in nanoscience,[4,5] and it is the key driver of
protein and drug interactions in biochemistry.[6] Since noncova-
lent interactions provide the basic control of a vast number of
processes, they are of fundamental importance to quantitative-
ly understanding the chemical interactions between guest
molecules and host structures. The complexity of supramolec-
ular systems, however, limits the possibilities of gaining a fun-
damental understanding. Indeed, it is fair to say that, although
chemists today can synthesize supramolecular structures with
amazing complexity, often even a basic understanding of the
interactions leading to such systems is lacking.[7] Information at
the electron-density (ED) level is a potential way to provide un-
derstanding of noncovalent interactions, and molecular crystals
are excellent supramolecular model entities facilitating their
study.[8]
Supramolecular chemistry is the chemistry of noncovalent in-
teractions, and it has very broad implications in natural sci-
ence. It is for instance the foundation for the design and devel-
[a] Dr. H. F. Clausen, Dr. M. R. V. Jørgensen, Dr. M. S. Schmøkel,
Dr. M. Christensen, Dr. J. Overgaard, Prof. Dr. B. B. Iversen
Center for Materials Crystallography
Department of Chemistry and iNANO, Aarhus University (Denmark)
E-mail: jacobo@chem.au.dk
bo@chem.au.dk
[b] Dr. S. Cenedese
Istituto di Scienze e Tecnologie Molecolari del CNR (CNR-ISTM) and
Dipartimento di Chimica Fisica ed Elettrochimica
Universitꢀ di Milano (Italy)
[c] Dr. Y.-S. Chen
ChemMatCARS, Advanced Photon Source, University of Chicago (USA)
[d] G. Koutsantonis, Prof. M. A. Spackman
School of Chemistry and Biochemistry
University of Western Australia (Australia)
E-mail: mark.spackman@uwa.edu.au
The porous host–guest systems obtained by crystallization
of hydroquinone (HQ) from small-molecule solvents are simple
supramolecular entities that have been known for many
years.[9,10] HQ exists in three polymorphs (a, b, and g forms)
under ambient conditions. In the b modification relatively rigid
spherical cages of nearly 4 ꢀ inner diameter are formed by six
[e] Dr. S. Cenedese
Current address: University of Toledo, Toledo, Ohio 43606 (USA)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201400129.
Chem. Eur. J. 2014, 20, 8089 – 8098
8089
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
porous host materials. Such simulations are ultimately depen-
dent on the accuracy of the force fields used in solving the
equations of motion. In the case of HQ clathrates recent inter-
est has concerned the phenomenon of flip-flop hydrogen
bonding, in which individual hydrogen atoms can alternate be-
tween hydrogen bonding to two acceptors.[17] Flip-flop hydro-
gen bonding is also believed to be of significant importance in
many biological systems (e.g., hydrated saccharides, nucleic
acids, resveratrol, and cholic acid),[18–23] and understanding the
mechanism of reorientation of such networks of hydrogen
bonds is a fascinating challenge. In the HQ system the flip-flop
hydrogen bonding is an important factor affecting the access
of guest molecules to the voids. By comparing MD simulations
on the empty host and the host with ethanol inclusion in a sim-
ilar host–guest system (Dianin’s compound), it was found that
the guest molecule significantly affects the energy landscape
of hydrogen-bond reorientation.[17] However, in such simula-
tions one main approximation is that the force field is un-
changed with and without guest-molecule inclusion. Thus,
even though the host and the guest molecules in reality may
have significant chemical interactions leading to polarization of
the ED, this is not accounted for in the modeling of the dy-
namical system. This is a general limitation in numerous MD
studies, and it is not known precisely how this affects the pre-
dictive power of the MD method. In the present experimental
study we report the accurate CD of HQ:acetonitrile (1). By
comparing it with our previously determined CD of the HQ
apohost 2, we can experimentally quantify the polarization of
the host due to inclusion of a polar guest molecule, and there-
by evaluate whether its omission in MD simulations could be
important.
Figure 1. a) The asymmetric unit of b-type hydroquinone clathrate with mo-
lecular labeling. b) The cavity generated by six HQ molecules enclathrating
an acetonitrile molecule. c) Packing diagram illustrating the hydrogen-
bonded hexagons viewed along the c axis of the unit cell of 1. The thermal
ellipsoids are drawn at a 90% probability.
There is a rich and varied crystal chemistry of the HQ clath-
rates and, depending on the ordering of the enclathrated sol-
vent, the b-HQ clathrate is reported to crystallize in three dif-
ferent structure types.[24] In the type I clathrate the guest mole-
cules are completely disordered leading to the highest symme-
HQ molecules, and potential guest molecules can fit inside the
cavity (Figure 1). The general formula of these clathrates is
3C6H4(OH)2·xS, where x is the fractional occupancy (x=0–1) of
the guest molecule S, such as CO2, CH3OH, and CH3CN.[11–14]
Access to the central voids of the HQ clathrate is through
a neck formed by a ring of six hydroxyl hydrogen bonds. A
particular advantage of b-HQ clathrates as supramolecular enti-
ties is that the apohost can be formed without any enclathrat-
ed guests, and we have previously carried out a CD study on
this reference system.[15] On the basis of accurate low-tempera-
ture synchrotron X-ray diffraction data, the CD of the HQ apo-
host crystal structure was modeled by using the multipole
method.[16] Analysis of the CD provided a detailed description
of the chemical bonding and electrostatic properties of the
cavity, which in turn determine the inclusion properties of the
host system.[15] Due to its relatively limited complexity, the HQ
apohost is well suited for fundamental studies on noncovalent
intermolecular interactions in supramolecular entities.
ꢀ
try (space group R3). Removing the inversion center at the
cage center gives rise to the type II clathrate in the polar space
group R3, which has partially ordered solvent molecules. Type
III clathrates are obtained by reduction of lattice symmetry
from rhombohedral to trigonal leading to the polar space
group P3, which is caused by further solvent ordering. Acetoni-
trile enclathrated in the b-HQ apohost[11] is a type III clathrate
in which only the ordered enclathrated acetonitrile molecules
break the rhombohedral symmetry, leading to trigonal symme-
try. This means that the ꢀh+k+l¼ 3n Bragg reflections are
no longer extinct, but contain contributions that are related to
the relatively small difference between the density distribu-
tions of the ordered and disordered guest molecules as well as
the perturbations of the host network. These reflections are in-
herently extremely weak, but nevertheless of significance for
the ED of the pseudorhombohedral structure of 1. In the pres-
ent study, we invested much effort into precisely measuring
these weak reflections using an extremely intense synchrotron
X-ray beam combined with a very low experimental tempera-
ture. To experimentally determine the very fine details of the
ED, we combined the synchrotron X-ray diffraction data with
Supramolecular processes are dynamic, and molecular dy-
namics (MD) simulations have become a widely used method
to probe processes ranging from molecular motion in ion
channels of membrane proteins to uptake of gas molecules in
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8090
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
independent neutron diffraction data to obtain the unbiased
hydrogen positional and thermal parameters. The combined X-
ray/neutron (X/N) data provide an experimental CD that is not
biased by the charge shift of the covalently bonded hydrogen
atoms. However, we found that addition of theoretical calcula-
tions was necessary to complete the picture, since the noise in
the very weak reflections limits the modeling of the ED of the
solvent molecule.
Results and Discussion
The host structure consists of two interconnecting hydrogen-
bonded networks, which only have weak interactions between
them. The crystal structure of 1 contains three unique HQ mol-
ecules and a total of nine molecules in the unit cell, generating
three cavities, each occupied by one acetonitrile guest mole-
cule (see Figure 1c). Three symmetry-independent acetonitrile
molecules, which are situated on threefold axes, fit inside the
voids of the rhombohedral HQ framework with one guest mol-
ecule oriented in a direction opposite to that of the other two
(Figure 1a). The interactions between the HQ host framework
and the acetonitrile guest molecules are expected to be weak,
and the unit cell parameters increase as expected (ca. 66 ꢀ3)
on inclusion of the acetonitrile molecule in 1 (a=15.7807(5),
c=6.1682(2) ꢀ; V=1330.30(7) ꢀ3) compared with the apohost
2 (a=16.5249(2), c=5.3430(1) ꢀ; V=1263.55(4) ꢀ3). The posi-
tion of the solvent molecule in the cavity is mainly determined
by two competing interactions, namely, the guest–guest inter-
action and the host–guest interaction.[25] The guest–guest in-
teraction of polar solvent molecules is of particular interest, as
the clathrates undergo a structural phase transition, which is
presumably driven by the interaction between the dipole mo-
ments of the guest molecules.[26] For 1, the ordering occurs
above room temperature and hence it crystallizes in the or-
dered phase. The ordering reduces the symmetry of the b-HQ
Figure 2. a) Static deformation ED and b) Laplacian distribution of one
unique hydroquinone host molecule (experimental refinement) of 1. c) Static
deformation density and d) Laplacian distribution of a guest acetonitrile
molecule of 1 (fixed at theoretical multipole values). For the deformation
density the contour interval is 0.1 eꢀꢀ3 with solid contours being positive
and dashed negative. For the Laplacian the contour level is 2x ꢂ10y (x=0, 1,
2, 3; y=ꢀ2, ꢀ1, 0, 1, 2, 3), the red dashed lines indicate positive values, and
the solid blue lines negative values.
whereas the ED of the acetonitrile molecules was mainly deter-
mined from the theoretical data, that of the host structure was
based solely on experimental data, and this was used in the
analysis of the chemical bonding and electrostatic properties.
The chemical bonding of the host structure can be analyzed
by using the quantum theory of atoms in molecules,[29] and
topological properties at the bond critical points (BCPs) of se-
lected hydrogen bonds are summarized in Table 1. The pertur-
bation of the HQ host due to inclusion of acetonitrile is not re-
flected in the covalent intramolecular interactions. On the
other hand, the hydrogen bonds are significantly affected. The
intermolecular hydrogen bonds of the HQ molecules of 1 are
all fairly similar with a BCP ED in the range 0.24–0.27 eꢀꢀ3.
These values are significantly smaller than those observed for
the hydrogen bond in empty b-HQ.[15] The intermolecular hy-
drogen bonds of the HQ molecules are primarily closed-shell
interactions with positive Laplacian values. However, the slight-
ly negative total energy density is indicative of covalent contri-
butions to these hydrogen bonds. A crude estimation of the
energy of the hydrogen bonds based on the energy density
was suggested by Espinosa et al., and for 1 this resulted in
ꢀ
clathrate from R3 to P3. The heat capacity of 1 was measured
from 2–300 K (see Supporting Information). There is no sharp
structural transition due to ordering of the solvent molecules.
The transition temperature has been reported to be propor-
tional to the square of the dipole moment of the guest mole-
cules and, due to the high dipole moment of acetonitrile, the
transition is above room temperature.[27] In the range from 20
to 75 K the slope of the heat capacity changes smoothly due
to the rotational ordering of the methyl groups, which has
been investigated previously by Detken et al. using single-crys-
tal NMR and inelastic neutron scattering.[28] At the temperature
of the present neutron and synchrotron experiments (ꢁ15 K)
there should be no rotational disorder, as confirmed by the
structural analysis.
Electron density
E(HB)=0.5 V=ꢀ41 kJmolꢀ1 [30]
,
while the value found for the
Static deformation electron densities and Laplacian distribu-
tions in the plane of an HQ molecule and of an acetonitrile
molecule are shown in Figure 2. The bonding regions of the ar-
omatic system show single peak maxima, which is an indica-
tion of the accuracy of the derived density. We emphasize that,
empty HQ clathrate 2 was 50% higher.[15]
Bader atom charges are additive, and a summation of the
host-structure charges derived from the experimental ED re-
sults in an overall neutral network with zero charge. Given that
the multipole parameters of the acetonitrile guest atoms are
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8091
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
static interaction between the host and the guest, or
simply a result of the increased distance between HQ
molecules, as evidenced by the increased unit-cell
and hydrogen-bonding distances, is examined below
in the analysis of the electrostatic potential.
Table 1. Topological analysis of the hydrogen bonding in the host structure of 1. The
second row gives the values obtained for the HQ apohost in ref. [13]. R [ꢀ] is the
bond length, Rij [ꢀ] the sum of the distance from the first atom through the BCP to
the other atom, d1 [ꢀ] the distance from the O atom to the BCP, 1 is the electron den-
2
sity [eꢀꢀ3], r 1 is the Laplacian [eꢀꢀ5], and e the ellipticity of the bond. G, V, and H
are repectively the local kinetic, potential, and total energy densities estimated by
using the Abramov functional.[31] The random least-squares errors in the values de-
rived from the electron density are smaller than systematic errors between different
models, and therefore only two significant digits are listed.
Electrostatic properties
The ED and the overall charge state of the host
system are important, since the inclusion properties
of the host are directly determined by these. As is
evident from the above discussion, the atomic charg-
es are affected by the inclusion of acetonitrile in the
HQ host. However, to assess the effect of these
changes on the electrostatics, which strongly influen-
ces the interaction energies, the electrostatic poten-
tial (EP) on the molecular surfaces of the HQ mole-
cules was calculated (Figure 4). The five different iso-
surfaces originate from 1) the experimental ED of
2
Bond[a]
R
Rij
d1
1
r 1
e
G
V
H
O(1)···H(11A)
1.7512(5) 1.7774 1.1797 0.24 2.4 0.58 0.19 ꢀ0.21
0.0
ꢀ0.3 ꢀ0.1
1.6899(2) 1.6916 1.1217 0.34 2.2 0.03 0.2
O(1A)···H(11)iv
O(11)···H(1A)i
1.7512(5) 1.7647 1.1361 0.27 1.9 0.33 0.18 ꢀ0.23 ꢀ0.05
1.7607(5) 1.7776 1.1800 0.26 2.1 0.57 0.18 ꢀ0.22 ꢀ0.04
O(14)···H(14A)iii 1.7543(5) 1.7700 1.1423 0.27 1.9 0.36 0.18 ꢀ0.22 ꢀ0.04
O(11A)···H(1)i
1.7606(5) 1.7800 1.1508 0.25 1.9 0.41 0.17 ꢀ0.21 ꢀ0.04
O(14A)···H(14)ii 1.7544(5) 1.7763 1.1797 0.25 2.3 0.60 0.19 ꢀ0.21 ꢀ0.03
[a] i: ꢀx+y+1, ꢀxꢀ1, z+1; ii: ꢀyꢀ1, xꢀy, z+1; iii: ꢀyꢀ1, xꢀyꢀ1, zꢀ1; iv: x, y, zꢀ1.
fixed to the values obtained from modeling the theoretical
data, the observation of charge neutrality of the host frame-
work is expected, but it is still an important observation illus-
trating the reliability of the integration of charge. The Bader
charges obtained for 1 are shown in Figure 3, together with
charges for 2 for comparison.[15] The aromatic hydrogen atoms
of the HQ molecules of 1 are more positively charged than
those of the empty host structure, and the transferred elec-
trons are primarily moved onto the hydroxyl-bonded carbon
atom of the aromatic ring. This indicates that, on inclusion of
the acetonitrile guest molecule, the ED of the host structure is
significantly perturbed. To ensure that this conspicuous result
is not due to the use of neutron parameters, an X-ray-only
model of 1 was produced in a similar manner to our previous
study on 2.[15] The obtained atomic Bader charges (shown in
green in Figure 3) deviate only slightly from the values ob-
tained for the combined X/N model of 1, and thus add confi-
dence to the conclusion that the electron redistribution of HQ
occurs as a consequence of the inclusion of the acetonitrile
molecules in the cavities. Whether this is a result of an electro-
1 (1exp), 2) the theoretical ED of 1 (1theo), 3) the experimental ED
of 2 (2exp), 4) the theoretical ED of 2 (2theo), and 5) the theoreti-
cal ED from a modification of 1 (1theo,mod) in which the HQ mol-
ecules were in the same geometry as 1, but the acetonitrile
solvent molecules were omitted.
Three main conclusions come from this comparison: Firstly,
it is clear that the experimental and theoretical EP on the mo-
lecular surfaces are similar for both 1 (Figure 4a, b) and 2 (Fig-
ure 4c, d), except for the electropositive (red) area on top of
the OH group in 1theo (Figure 4b), which is not seen for 1exp
(Figure 4a). This indicates that polarization of the HQ mole-
cules in the crystal structure is indeed accessible from experi-
ment by using a multipole model. Secondly, the differences
observed in the ED between 1 and 2 are clearly expressed in
the EP, since for 1 the aromatic hydrogen atoms are significant-
ly more positive (electrophilic; yellow/red), whereas the aro-
matic rings and the carbonyl groups are more nucleophilic
(blue) compared with 2, as also inferred from the atomic
charges. However, the most important point remains the
origin of the polarization of the host molecules. There are at
least two viable alternative explanations for the polarization:
1) that the electrostatic moments of the acetonitrile molecule
directly polarize the HQ molecules; 2) that the larger separa-
tion between HQ molecules changes the intermolecular con-
tacts to such an extent that the electrostatics of HQ are influ-
enced. Figure 4e shows the EP of an artificially expanded
empty HQ crystal structure (HQ geometry as in 1 but without
acetonitrile inclusion), and from its remarkable similarity with
Figure 4d it is immediately clear that enlargement of the
cavity has virtually no influence on the EP evaluated at the HQ
isosurface. On the other hand, the inclusion of acetonitrile,
which may be regarded as the difference between Figure 4b
and e, has a profound effect on the EP on the molecular sur-
face. We are fully aware of the existing negative results on ex-
perimental and theoretical descriptions of interaction densi-
ties,[32,33] but recognize that the differences between the two
compounds studied here, 1 and 2, are based on two different
crystal structures. The electrostatic effects of incorporating
Figure 3. Topological charges of the Bader atoms of 1 (blue), the b-HQ apo-
host 2 (red), and 1 based on X-ray refinement only, without including the
neutron parameters (green). The arrows show the movements of electrons
due to the inclusion of the acetonitrile molecules.
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8092
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
acetonitrile molecules is considerably weaker in
1 than in, for example, the chromium-wheel com-
pound. The polarization of the host structure on in-
clusion of acetonitrile changes the electrostatic po-
tential in the center of the cavity an order of magni-
tude from 0.014 eꢀꢀ1 in 2 to 0.13 eꢀꢀ1 in 1, and this
directly demonstrates that host-structure polarization
has important energetic consequences. The differ-
ence between the EP in the cavities of HQ and in the
chromium-wheel compound results in different affini-
ty towards molecular groups. In the chromium
wheel, the electronegative cavity attracts the electro-
positive groups of the molecular guests, and hence
the electronegative group is expected to point out of
the cavity. This preferential alignment was indeed ex-
perimentally observed for a number of chromium-
wheel host–guest complexes. Thus, acetonitrile ori-
ents the positively polarized carbon atom into the
electronegative center of the cavity of the chromium
wheel, whereas the negatively polarized nitrogen
atom points towards the more electropositive surface
of the host.
In 1 the center of the cavity is electropositive, and
the potential increases along the cavity in the direc-
tion of the threefold axis, which suggests that elec-
tronegative regions of molecules situated in the
cavity would point in these directions, for example,
towards the hexagon of hydrogen-bonded HQ mole-
cules. This is in fact the case for the nitrogen atom of
the cyano group at one end of the acetonitrile mole-
cule, and this placement of electronegative groups
was also suggested by Zubkus et al.[37] In the perpen-
dicular direction away from the center of the cavity,
the electrostatic potential drops further, but due to
steric hindrance the guest molecules in 1 cannot
point in that direction. Therefore, the inclusion of
acetonitrile must be associated with an electrostatic
penalty, since it cannot attain electrostatic comple-
Figure 4. The electrostatic potential mapped from ꢀ0.10 (blue) to 0.10 eꢀꢀ1 (red) on the
electron-density isosurface of 0.01 eꢀꢀ3 visualized by using MoleCoolQt.[62] The surfaces
were calculated from multipole models based on structure factors from a) experiment,
HQ–acetonitrile (1exp); b) theory, HQ–acetonitrile (1theo); c) experiment, empty HQ (2exp);
d) theory, empty HQ (2theo); e) a theoretical modified version of 1 in which the acetoni-
trile molecules are omitted but the geometry of the HQ molecules is as in HQ–acetoni-
trile (1theo,mod).
guest molecules into the HQ apohost structure can be further
examined by plotting the EP in planes parallel to the a and
c axes along the threefold axis (Figure 5). These plots clearly
show the electropositive nature of the cavity.
mentarity at both ends, because tilting of the molecule is hin-
dered. If tilting were feasible the positively charged atom
could be placed away from the threefold axis to reduce the
electrostatic energy penalty, since the EP decreases on moving
perpendicular to the threefold axis. Since acetonitrile is among
the largest molecules that can be accommodated and form
a stable HQ host–guest compound, it is expected to be steri-
cally hindered. In the case of the smaller methanol guest, a tilt
of about 318 is observed in the HQ host,[38] and this presuma-
bly is due to the electrostatically more favorable tilted position.
However, HCN is aligned along the rotation axis, and hence
other contributions to the combined energy may play signifi-
cant roles in this particular case. Overall, the inclusion and ori-
entation of guest molecules in the HQ cavities most probably
depend on several factors and are a combination of steric,
electrostatic, and covalent contributions.
Theoretical modeling of hydrogen storage in metal–organic
frameworks (MOFs) has revealed that the host–guest interac-
tions are promoted by the local polarity of the host network,
which induces a dipole in the otherwise nonpolar H2 guest
molecules.[34] This shows that the electrostatic properties of the
host are of great importance in predicting the properties of
host–guest systems.[35] In the chromium-wheel host complex
[Cr8F8(tBuCO2)16] the EP was funnel-shaped with a three-dimen-
sional saddle point located at the center, which was found to
be highly electronegative with an electrostatic potential of
ꢀ1.24 eꢀꢀ1.[36] In 1 and in empty b-HQ (2), a saddle point is
also observed, but in this case the EP is positive. In 1 the EP
goes from about 0.30 eꢀꢀ1 near the hydrogen-bonded hexa-
gon of the host structure to about 0.13 eꢀꢀ1 at the center of
the cavity. Hence, the electrostatic interaction energy with the
The electric field (EF) of the host structure in 1 was calculat-
ed with all host molecules residing inside a box of 2ꢂ2ꢂ3 unit
cells excluding the contribution from the guest molecules (i.e.,
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8093
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 5. a) Schematic showing the plane used for the EP maps. It is parallel
to the a and c axes of the unit cell. b)–d) show the EP of the cavities holding
acetonitrile molecules AceI, AceII, and AceIII, respectively, plotted in
a 6.1ꢂ6.1 ꢀ2 section. The contour interval is in 0.01 eꢀꢀ1 steps with EP
ranging from 0.40 eꢀꢀ1 (near atom centers) to 0.15 eꢀꢀ1. Red contours are
drawn for every 0.05 eꢀꢀ1
.
due solely to the host molecules), and it is as expected fairly
symmetrical (Figure 6). The field is close to zero near the
center of the cavity (by symmetry it is exactly zero at the inver-
sion center), and in general, on moving along the threefold
axis away from the center the field increases. The EF along the
c axis of the unit cell 1 ꢀ from the center of the three cavities
ranges from 0.027 to 0.036 eꢀꢀ2, depending on the cavity, and
the fields point toward the center of the cavity. The numerical
values are significantly higher than those observed in the
empty b-HQ structure (0.0185 eꢀꢀ2). On moving perpendicular
to the threefold axis, the EF also increases, but it now points
away from the center of the cavity. The EF values 1 ꢀ away
from the cavity center in this direction range from 0.013 to
0.019 eꢀꢀ2, whereas the EF of the empty b-HQ was previously
found to be 0.0105 eꢀꢀ2. In both directions, parallel and per-
pendicular to the threefold axis, the EF of 1 is up to twice as
high as that reported for the empty b-HQ reference struc-
ture 2.
The observed differences between 1 and the empty host 2
show that inclusion of the acetonitrile guest molecules is ac-
companied by a polarization of the host structure that involves
movement of electrons from the aromatic hydrogen atoms
onto the carbon atoms of the benzene ring. This leads to
almost a doubling of the electric field and the EP in 1 at a dis-
tance of 1 ꢀ from the center of the cavity compared with 2.
The change in the ED of the host structure has general impli-
cations for theoretical calculations such as MD calculations
that implement force-field methods to obtain the potential
energy of a given system. Force fields are the foundation, for
Figure 6. The electric field due to just the host molecules in 1, plotted in
a 2ꢂ2 ꢀ plane parallel to the a and c axes of the unit cell and through the
center of the cavities of a) AceI, b) AceII, and c) AceIII. The centers of the
cavities are denoted by the green dots. Arrows indicate the direction of the
EF and their length its magnitude.
example, of many investigations of protein–ligand interactions
in drug design.[39] The molecules in such calculations are com-
posed of atom types that depend on the atomic number and
the type of chemical bonding.[40] The molecules are described
by a ball-and-spring model, in which atoms in a nonpolarizable
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8094
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
force field are considered to be static and thereby unable to
be perturbed by nearby ions or molecules. The observed per-
turbation of the host molecule in 1 compared with the empty
HQ structure of 2 would not be accounted for in typical MD
calculations with nonpolarizable force fields, and this must
lead to errors in the electrostatic energy, which contributes to
the noncovalent part of the potential energy being minimized.
For nonpolar systems, such as hydrocarbons, the electrostatic
energy term is not crucial, but for polar systems the long-
range electrostatic interaction is often the dominant term.[40]
Typical examples of nonpolarizable force fields are MM2, MM3,
and OPLS.[41,42] Another commonly used force field is the gen-
eral AMBER force field (GAFF), which provides parameters for
small organic molecules for simulations of drugs and small-
molecule ligands. Nemkevich et al. recently used GAFF in
a study of flip-flop hydrogen bonding in clathrates such as Dia-
nin’s compounds and HQ.[17] The barrier of reorientation of
a single hydroxyl group in the two host molecules was found
to be lowest for b-HQ, and hence the reorientation of the hy-
drogen bonding is more frequent. A nonpolarizable force field
was used for the study; however, the energy estimations of
the barriers must be influenced by the polar guest molecules,
as indicated in the present study, which shows that they have
the ability to polarize the host structure of b-HQ.
quantitatively describe the effects of the crystal field on elec-
trostatic properties. In particular, they showed that multipole
refinements of unperturbed and perturbed densities cannot
differentiate electrostatic energies of up to 60 kJmolꢀ1. Howev-
er, Bak et al were operating with identical geometries, and the
HQ distribution and the hydrogen-bonding network in 1 are
significantly different from those in the apohost. It is thus not
immediately obvious that the conclusions made by Bak et al.
have validity in this study. Using the electrostatic potential as
the probe reveals that the effect of guest inclusion is at least
an order of magnitude larger than the geometry change and is
accessible by using a multipole refinement, in contrast to the
results of Bak et al. Changes in atomic charges on the order of
0.15 e make a significant contribution to the electrostatic ener-
gies of the system.
Conclusion
State-of-the-art single-crystal synchrotron X-ray diffraction data
in combination with single-crystal neutron diffraction data
measured at very low temperature have been used to obtain
an experimental CD model of the HQ:acetonitrile clathrate 1.
Significant efforts were devoted to obtaining accurate Bragg
intensities for the weak reflections containing information
about the ED of the acetonitrile guest molecules, but these
weak data currently remain out of reach for experimentalists at
the level of precision required to perform successful ED model-
ing. Instead, an ED model was obtained through use of theo-
retical calculations on the guest molecule and constraints on
the symmetry of the host structure. The HQ:acetonitrile crystal
structure features three independent molecules of acetonitrile,
of which two point in a direction opposite to that of the third.
The ED of the HQ host molecule in 1 was found to be signif-
icantly different from that of the HQ apohost 2, and intramo-
lecular transfer of electrons from the hydrogen atoms to the
carbon atoms takes place. This induces a change in electrostat-
ic potential at the periphery of the HQ molecules. The ob-
served change in EP was shown to follow not from the geo-
metrical change that happens on inclusion, but from the polar-
izing interaction between host and guest. The fact that the HQ
molecule is polarized in 1 compared with the apohost struc-
ture of 2 due to incorporation of a guest molecule may have
a significant implication for molecular simulations with nonpo-
larizable force fields, since the interaction energies possibly
contain significant errors.
The present results also have important implications for
the so-called pseudoatom databases, such as ELMAM,[43]
ELMAM2,[44] Invariom,[45,46] UBDB,[47–49] and MOF building
blocks,[50,51] which are used for experimental structure refine-
ment with aspherical scattering factors as an alternative to the
standard independent-atom model. Bak et al. recently com-
pared the effectiveness of the different databases, and found
that the estimation of molecular electrostatic potential is limit-
ed to purely qualitative information due to the low precision
of the obtained EPs,[32] and this suggested that the electrostat-
ic interaction energies will also be poorly determined. One of
the principal ideas behind these databases is that the ED and
the derived properties of an atom should be transferable to
another molecule if the local environment is similar (i.e., the
neighboring atoms are identical). This transferability provides
a tool for using the fragments as building blocks for the addi-
tive generation of the ED of larger systems or to obtain EDs
from low-resolution data. However, the perturbation of the
charge observed in 1 compared with apohost 2 indicates that
such transferability may not be strictly valid, and therefore de-
rived properties such as EPs may be in significant error. One
approach to remedy this situation would be to include longer-
ranging interactions, including intermolecular interactions,
when categorizing atom types for the ED databases. This is,
however, not an option when building force fields in, for exam-
ple, protein-docking studies where the small molecules are
free to roam around looking for potentially energy-rich places
to interact. In studies of this kind, the energies that separate
one site from another are typically small, and hence the host
polarization revealed in the present study, albeit small, may be
an important contribution to the total energy. Bak et al. and
Dittrich et al.[33] recently concluded, on the basis of theoretical
studies on amino acids, that the multipole model is not able to
Experimental Section
Synthesis
b-HQ clathrates 3C6H4(OH)2·xS are simply prepared by evaporation
of the solvent from HQ solutions. For the synthesis of 1, several
batches were prepared to accommodate the different crystal sizes
needed for neutron and synchrotron X-ray experiments (see the
Supporting Information for further details).
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8095
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
rected for the imperfect detection by the CCD phosphor by using
a transmission factor of 0.6041 in the program SADABS.[54,55] This
program also corrects for slight misalignment of the crystal with
respect to the f axis, as well as absorption and other systematic
effects. The weak reflections were expected to be important for
the accurate modeling of the ED of the guest molecules. The sub-
group of weak reflections breaking the rhombohedral symmetry
ꢀh+k+l¼3n was taken only from the 3 s runs to avoid poorly de-
termined intensities from the measurements with shorter exposure
times. The remaining 131229 reflections were averaged and outli-
ers were rejected with the program SORTAV to give a total of
21872 unique reflections with Rint =4.40%.[53] Only reflections mea-
sured three times or more were used in the subsequent multipole
refinements. Selected experimental details are listed in Table 2.
Heat capacity
The heat capacity Cp of 1 was measured from 2 to 300 K on a Quan-
tum Design Physical Properties Measurement System (PPMS) by
using pellets pressed from finely ground powder. The data are
shown in the Supporting Information.
Neutron diffraction data collection
Single-crystal neutron diffraction data of 1 were collected at the
D19 instrument at the Institute Laue-Langevin, France, by using
monochromatic thermal neutrons with a wavelength of 0.9458 ꢀ.
A hexagonal prism-shaped crystal with dimensions of 1.5ꢂ1.5ꢂ
2.5 mm was wrapped in Al foil and mounted on the tip of an alu-
minum pin. The sample was placed in a two-stage closed cycle re-
frigerator, cooled to 15 K, and the data were subsequently collect-
ed with a large position-sensitive detector. A total of 43696 reflec-
tions were collected to a maximum resolution of 0.929 ꢀꢀ1. The
crystal faces were indexed as {100}, {ꢀ120} and {101}, and the data
were integrated and corrected for absorption by using locally de-
veloped programs and a calculated absorption coefficient of
Theoretical calculations
A wave function was calculated by using ab initio periodic DFT
with the B3LYP exchange and correlation functional[56,57] imple-
mented in the program CRYSTAL06.[58] A standard Pople 6-311G**
all-electron basis set was used in the calculations. The calculations
were performed with the present experimental geometry and the
geometry of the apohost from Clausen et al.[15] The SCF (self-
consistent field) energy convergence threshold was set to
DE<10ꢀ6 hartree. The topology of the density was analyzed with
the program TOPOND08.[59]
1.8 cmꢀ1 [52]
Outlier rejection and merging with the program
.
SORTAV gave a total of 5699 unique reflections with an internal
agreement factor of Rint =3.26% and an overall completeness of
94.9%.[53] Further experimental details are listed in Table 2.
Refinement of neutron data
Table 2. Crystallographic details and refinement residuals.
The guest molecules in 1 are ordered with two acetonitrile mole-
cules pointing up and one pointing down (Figure 1a). The model
that was used for describing the molecular structure included an
Neutron
Synchrotron X-ray
empirical formula
(C6H6O2)3CH3CN
15(2)
15.7845(2)
6.1736(1)
1332.08(3)
0.9458
(C6H6O2)3CH3CN
15(2)
15.7807(5)
6.1683(2)
1330.30(7)
0.41328
0.011
0.89/1.00
144185/21872
0.044
T [K]
a [ꢀ]
c [ꢀ]
V [ꢀ3]
l [ꢀ]
ꢀ
inversion center for the host atoms (adhering to P3), while the indi-
vidual guest molecules were allowed to differ (P3 symmetry). The
geometry and ADPs of the three unique acetonitrile molecules
were constrained to be identical. Details of this constrained refine-
ment are listed in Table 2.
m [mmꢀ1
Tmin/Tmax
]
0.18
0.6783/0.8028
43696/5699
0.0326
A Hirshfeld rigid-bond test was satisfactorily fulfilled for the struc-
ture[60,61] with a mean value of the difference of mean-square dis-
placement amplitudes (DMSDA) of DAꢀB=5.9 pm2. The maximum
DAꢀB value of the host was 10 pm2, whereas in the cyano group
of the solvent molecule the DMSDA value rose to 21 pm2.[62] The
positions and (scaled) ADPs of the hydrogen atoms were used in
the X/N multipole refinement (see below).
Nmeasured/Nunique
Rint
(sinq/l)max [ꢀꢀ1]
0.93
1.20
X/N multipole model
P3
4276 (2s level)
192
0.1134/0.0221
0.0610
ꢀ0.66/+0.71 (acetonitrile),
<0.17 (HQ)
0.7510
space group
Nobs
Npar/Nrestraints
Rall(F), R2s(F)
wR(F2)
P3
5699
157/12
0.0892/0.0511
0.0825
X-ray multipole refinement
residuals
ꢀ1.44/+2.12
The neutron model obtained from SHELXL (atomic positions and
ADPs for all atom types including hydrogen) was chosen as a start-
ing point for the multipole refinement. The model was imported
into the XD program,[63] and non-H atom parameters were refined
against the high-order part of the synchrotron data with sinq/l>
0.8 ꢀꢀ1. The constraints used for the neutron model, for example,
GoF
1.541
X-ray diffraction data collection
A single crystal of 1 was mounted on the Bruker D8 diffractometer center of symmetry on the coordinates and ADPs of the hydroqui-
at the ChemMatCARS beamline at the Advanced Photon Source in none host, and restraints on the geometry and ADPs of the aceto-
Chicago. The transparent crystal (0.040ꢂ0.040ꢂ0.050 mm) was nitrile guest molecules, were also used in the subsequent multi-
cooled to 15(2) K by using an open-flow liquid-helium cryostat. pole refinement. In addition, the bond lengths from the heavy
The data were collected at a wavelength of 0.41328 ꢀ with a maxi- atoms to the hydrogen atoms were fixed to the values obtained in
mum resolution of 1.20 ꢀꢀ1. A combination of exposure times of the refinement of the neutron data. To optimize the starting
0.5, 1, and 3 s was chosen to obtain accurate intensities of the model, nonzero multipole populations were retrieved from Invar-
weak reflections without compromising the intensities of the iom atoms by using the InvariomTool.[64] After initial refinement of
strong reflections due to the finite dynamic range of the Bruker these parameters, a full multipole model was introduced with
APEXII CCD detector. A total of 144185 reflections were collected gradually increasing complexity while maintaining the symmetry
and integrated with SAINT+.[54] Subsequently the data were cor- restrictions. The imposed constraints reduced the number of refin-
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8096
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
able parameters to those of half a hydroquinone molecule, while
only one unique acetonitrile was refined and the other two aceto-
nitrile molecules were constrained to have the same multipole
parameters.
ing of the theoretical structure factors. The geometries of all three
acetonitrile guest molecules were fixed to be identical. The final
model thus consisted of 192 parameters, which were refined
against 4276 observations [I>2s(I)]. The Hirshfeld rigid-bond test
was satisfactory fulfilled for the host (DAꢀBꢂ8 pm2, hDAꢀBi=
4.0 pm2), but for the acetonitrile molecules the values are highly af-
fected by noise, and this led to values of 21 and 47 pm2, respec-
tively. The residual density near the acetonitrile molecule showed
peaks ranging from ꢀ0.66 to +0.71 eꢀꢀ3 situated on the threefold
rotation axis, whereas the residual density is less than ꢃ0.17 eꢀꢀ3
near the host HQ molecules. Residual density maps are provided in
the Supporting Information.
It is possible that forcing identical electronic distributions on the
two different orientations of the acetonitrile molecules (up and
down) is inappropriate; however, as already discussed, information
about the differences between the three acetonitrile molecules is
restricted to the very weak intensities breaking the rhombohedral
symmetry (ꢀh+k+l¼3n), which makes it difficult to assess the
severity of this approximation. Attempts to distinguish the ED of
the solvent molecules resulted in unrealistically large charge trans-
fers between the atoms of the acetonitrile molecules, and hence
the monopole populations of the acetonitrile guest were fixed to
the Invariom values. Experimental noise tends to accumulate on
the high-symmetry axis,[65] which in this structure coincides with
the threefold axis of the acetonitrile molecules, and hence this
noise will significantly bias the refinement.
Acknowledgements
This work was supported by the Danish National Research
Foundation (Center for Materials Crystallography, DNRF-93),
the Danish Research Council for Nature and Universe (Dan-
scatt) and the Australian Research Council (DP0879405 and
DP130103304). ChemMatCARS Sector 15 is principally support-
ed by the National Science Foundation Department of Energy
under grant number NSF/CHE-0822838. Use of the Advanced
Photon Source was supported by the U. S. Department of
Energy, Office of Science, Office of Basic Energy Sciences,
under Contract No. DE-AC02-06CH11357. We would like to
acknowledge the instrument time obtained at D19 at the ILL.
After co-refinement of all parameters (structural, thermal, and elec-
tronic parameters for the host, electronic parameters and ADPs for
the guest molecules), the ADPs of the heavy atoms were compared
with the values obtained from refinement against the neutron
data. In general, the ADPs serve as a bin for uncorrected systematic
errors in the experiments, and hence inaccurate data tend to show
higher deviation between X-ray and neutron ADPs.[66] Comparing
the ADPs of the 33 heavy atoms in 1 by using the program UIJXN
leads to a mean deviation hDUi of 0.00058 ꢀ2,[67] which, for such
a large system, is an excellent correspondence compared with
other X/N studies.[68] The hydrogen nuclear positions and ADPs,
which are important in accurate ED modeling of hydrogen-bonded
systems such as 1,[69] were therefore expected to be of excellent
quality. Increasing the complexity of the multipole model of the
host structure, for example, by removing the center of symmetry,
removing constraints (electronic, positional, and thermal), or
adding additional radial expansion–contraction parameters did not
result in a significant improvement of the residual density or the
agreement factors. Despite the imposed constraints, the refine-
ment of the ED of the acetonitrile molecule still resulted in unreal-
istic polarization, that is, a density of acetonitrile that is very differ-
ent from other acetonitrile molecules studied in the literature.[70]
Keywords: clathrates
·
host–guest systems
·
neutron
diffraction · supramolecular chemistry · X-ray diffraction
[1] G. R. Desiraju, Nature 2001, 412, 397.
[2] J. Reinbold, K. Buhlmann, K. Cammann, A. Wierig, C. Wimmer, E. Weber,
Sens. Actuators B 1994, 18, 77.
[3] O. Kçnig, H. B. Burgi, T. Armbruster, J. Hulliger, T. Weber, J. Am. Chem.
Soc. 1997, 119, 10632.
[4] S. I. Stupp, Nano Lett. 2010, 10, 4783.
[5] S. Srivastava, N. A. Kotov, Soft Matter 2009, 5, 1146.
[6] A. M. Seddon, D. Casey, R. V. Law, A. Gee, R. H. Templer, O. Ces, Chem.
Soc. Rev. 2009, 38, 2509.
Due to the problematic modeling of the acetonitrile ED, another
approach was chosen for the final refinements. Structure factors
from a full theoretical calculation of 1 using the experimental ge-
ometry were used for multipole refinement in XD. Refinement de-
tails are included in the Supporting Information. The resulting mul-
tipole parameters were used as starting values in the subsequent
refinements against the experimental data. However, refinement of
these parameters was still not successful for the guest molecules,
probably due to the increased random noise along the symmetry
axis, and they were therefore fixed to the theoretical values. The
random noise that accumulates on the rotation axis and unfortu-
nately coincides with the position of the acetonitrile molecule also
hampers accurate determination of the scale factor. On the basis
of these observations we conclude that the weak subgroup of re-
flections breaking the rhombohedral symmetry ꢀh+k+l¼3n are
not determined accurately enough despite the great care that was
taken during the experiments (I/s=4.4 for these (1262) reflec-
tions). It was not possible to generate a model for disorder on the
acetonitrile sites that could lower the residuals.
[7] P. O. Brown, K. A. Udachin, G. D. Enright, J. A. Ripmeester, Chem. Eur. J.
2006, 12, 8240.
[8] G. R. Desiraju in Comprehensive Supramolecular Chemistry, Vol. 6 (Eds:
D. D. MacNicol, F. Toda, R. Bishop) Pergamon, Oxford (UK), 1996.
[9] W. A. Caspari, J. Chem. Soc. 1926, 129, 2944.
[10] W. A. Caspari, J. Chem. Soc. 1927, 1093.
[11] D. E. Palin, H. M. Powell, J. Chem. Soc. 1948, 815.
[12] D. E. Palin, H. M. Powell, J. Chem. Soc. 1948, 571.
[13] H. M. Powell, J. Chem. Soc. 1948, 61.
[14] S. C. Wallwork, H. M. Powell, J. Chem. Soc. 1956, 4855.
[15] H. F. Clausen, Y.-S. Chen, D. Jayatilaka, J. Overgaard, G. A. Koutsantonis,
M. A. Spackman, B. B. Iversen, J. Phys. Chem. A 2011, 115, 12962.
[16] N. K. Hansen, P. Coppens, Acta Crystallogr. Sect. A 1978, 34, 909.
[17] A. Nemkevich, M. A. Spackman, B. Corry, J. Am. Chem. Soc. 2011, 133,
18880.
[18] W. Saenger, C. Betzel, B. Hingerty, G. M. Brown, Nature 1982, 296, 581.
[19] C. Betzel, W. Saenger, B. E. Hingerty, G. M. Brown, J. Am. Chem. Soc.
1984, 106, 7545.
[20] V. V. Mossine, C. L. Barnes, T. P. Mawhinney, Carbohydr. Res. 2009, 344,
948.
[21] O. Kennard, Pure Appl. Chem. 1984, 56, 989.
[22] F. Caruso, J. Tanski, A. Villegas-Estrada, M. Rossi, J. Agric. Food Chem.
2004, 52, 7279.
[23] L. Lessinger, B. W. Low, J. Cryst. Spectrosc. 1993, 23, 85.
[24] T. C. W. Mak, J. Chem. Soc. Perkin Trans. 2 1982, 1435.
In summary, the final multipole model used simultaneous refine-
ment of the structural, thermal, and electronic parameters of the
host structure, whereas the electronic parameters of the acetoni-
trile molecules were fixed at the values obtained from the model-
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8097
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[25] T. L. Chan, T. C. W. Mak, J. Chem. Soc. Perkin Trans. 2 1983, 777.
[26] M. C. Rheinstadter, A. V. Kityk, A. Klopperpieper, K. Knorr, Phys. Rev. B
2002, 66, 064105.
[27] T. Matsuo, H. Suga, J. Incl. Phen. Macrocycl. Chem. 1984, 2, 49.
[28] A. Detken, P. Schiebel, M. R. Johnson, H. Zimmermann, U. Haeberlen, J.
Chem. Phys. 1998, 238, 301–314.
[50] A. Munter, Scattering Length Density Calculator, see http://
www.ncnr.nist.gov/resources/sldcalc.html.
[51] A. S. Chimpri, P. Macchi, Phys. Scr. 2013, 87, 048105.
[52] A. S. Chimpri, M. Gryl, L. H. R. Dos Santos, A. Krawczuk, P. Macchi, Cryst.
Growth Des. 2013, 13, 2995.
[53] R. H. Blessing, J. Appl. Crystallogr. 1997, 30, 421.
[29] R. F. W: Bader, Atoms in Molecules: A Quantum Theory, Oxford University
Press, Oxford, 1990.
[30] E. Espinosa, E. Molins, C. Lecomte, Chem. Phys. Lett. 1998, 285, 170.
[31] Y. A. Abramov, Acta Crystallogr. Sect. A 1997, 53, 264.
[32] J. M. Bak, Z. Czyznikowskab, P. M. Dominiak, Acta Crystallogr. Sect. A
2012, 68, 705–714.
[33] B. Dittrich, E. Sze, J. J. Holstein, C. B. Hꢃbschle, D. Jayatilaka, Acta Crys-
tallogr. Sect. A 2012, 68, 435–442.
[34] J. L. Belof, A. C. Stern, M. Eddaoudi, B. Space, J. Am. Chem. Soc. 2007,
129, 15202.
[35] J. A. Ripmeester, R. E. Hawkins, D. W. Davidson, J. Chem. Phys. 1979, 71,
1889.
[36] J. Overgaard, B. B. Iversen, S. P. Palii, G. A. Timco, N. V. Gerbeleu, F. K.
Larsen, Chem. Eur. J. 2002, 8, 2775.
[37] V. E. Zubkus, I. L. Shamovsky, E. E. Tornau, J. Chem. Phys. 1992, 97, 8617.
[38] J. M. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman, D. A. Case, J.
Comput. Chem. 2004, 25, 1157.
[39] F. Jensen, Introduction to Computational Chemistry, Wiley, Hoboken,
2006.
[54] G. M. Sheldrick, Acta Crystallogr. Sect. A 2008, 64, 112–122.
[55] G. Wu, B. L. Rodrigues, P. Coppens, J. Appl. Crystallogr. 2002, 35, 356.
[56] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
[57] C. T. Lee, W. T. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785.
[58] R. Dovesi, V. R. Saunders, C. Roetti, R. Orlando, C. M. Zicovich-Wilson, F.
Pascale, B. Civalleri, K. Doll, N. M. Harrison, I. J. Bush, P. D’Arco, M. Llu-
nell, CRYSTAL06 User’s Manual (University of Torino, Torino, 2006).
[59] C. Gatti, TOPOND98: an electron density topological program for
system periodic in N (N=0–3) dimensions, User’s Manual, CNR-ISTM,
Milano (Italy), 1999; C. Gatti, V. R. Saunders, C. Roetti, J. Chem. Phys.
1994, 101, 10686.
[60] F. L. Hirshfeld, Acta Crystallogr. Sect. A 1976, 32, 239.
[61] M. Harel, F. L. Hirshfeld, Acta Crystallogr. Sect. B 1975, 31, 162.
[62] V. Schomaker, K. N. Trueblood, Acta Crystallogr. Sect. B 1968, 24, 63.
[63] A. Volkov, P. Macchi, L. J. Farrugia, C. Gatti, P. Mallinson, T. Richter, T. Kor-
itsanszky, XD2006 - A Computer Program Package for Multipole Refine-
ment, Topological Analysis of Charge Densities and Evaluation of
Intermolecular Energies from Experimental and Theoretical Structure
Factors.
[40] N. L. Allinger, J. Am. Chem. Soc. 1977, 99, 8127.
[41] N. L. Allinger, Y. H. Yuh, J. H. Lii, J. Am. Chem. Soc. 1989, 111, 8551.
[42] W. Damm, A. Frontera, J. Tirado-Rives, W. L. Jorgensen, J. Comp. Chem.
1997, 18, 1955.
[64] C. B. Hꢃbschle, P. Luger, B. Dittrich, J. Appl. Crystallogr. 2007, 40, 623.
[65] B. Rees, Acta Crystallogr. Sect. A 1976, 32, 483.
[66] B. B. Iversen, F. K. Larsen, B. B. Figgis, P. A. Reynolds, A. J. Schultz, Acta
Crystallogr. Sect. B 1996, 52, 923.
[43] V. Pichon-Pesme, C. Lecomte, H. Lachekar, J. Phys. Chem. 1995, 99, 6242.
[44] S. Domagala, C. Jelsch, J. Appl. Crystallogr. 2008, 41, 1140.
[45] B. Dittrich, T. Koritsanszky, P. Luger, Angew. Chem. 2004, 116, 2773;
Angew. Chem. Int. Ed. 2004, 43, 2718.
[46] P. Luger, C. B. Hꢃbschle, M. A. Spackman, B. Dittrich, Acta Crystallogr.
Sect. D 2006, 62, 1325.
[67] R. H. Blessing, Acta Crystallogr. Sect. B 1995, 51, 816.
[68] W. Morgenroth, J. Overgaard, H. F. Clausen, H. Svendsen, M. R. V. Jørgen-
sen, F. K. Larsen, B. B. Iversen, J. Appl. Crystallogr. 2008, 41, 846.
[69] J. Overgaard, B. Schiott, F. K. Larsen, B. B. Iversen, Chem. Eur. J. 2001, 7,
3756.
[70] J. Overgaard, D. E. Hibbs, E. Rentschler, G. A. Timco, F. K. Larsen, Inorg.
Chem. 2003, 42, 7593.
[47] T. Koritsanszky, A. Volkov, P. Coppens, Acta Crystallogr. Sect. A 2002, 58,
464.
[48] X. Li, A. Volkov, K. Szalewicz, P. Coppens, Acta Crystallogr Sect. D 2006,
62, 639.
Received: January 12, 2014
[49] P. M. Dominiak, A. Volkov, X. Li, M. Messerschmidt, P. Coppens, J. Chem.
Theory Comput. 2007, 3, 232.
Published online on May 14, 2014
Chem. Eur. J. 2014, 20, 8089 – 8098
www.chemeurj.org
8098
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim