Structure, stability and guest affinity of tris(3-ureidobenzyl)amine capsules in solution

Article abstract of DOI:10.1002/chem.200601022

In non-competitive solvents, the tris(3-ureidobenzyl)amines 1a-c form dimeric assemblies in which guests such as CH₃CN, CH ₃NO₂, CH₂Cl₃, CH₃I, CH₂BrCl, CH₂Br₂, CHCl₃ and C ₆H₆ can be encapsulated. Variable temperature ¹ and ¹H,¹H-ROESY NMR spectroscopy, as well as pulsed-gradient spin-echo (PGSE) diffusion measurements were used to in-ivestigate the encapsulation within 1a·1a (1a: tris{3-[N′-(4- butylphenyl)-ureido]benzyl}amine). Kinetic parameters for the encapsulation of CH₃NO₂, CH₂Cl₂ and CH₃I, both in CDCl₃ and in [D₈]toluene have been obtained by using magnetisation transfer methods. These data are discussed together with the thermodynamic parameters. The affinity between guest and capsule seems to be dictated mainly by the electronic, size and shape complementarity between cavity and guest. A gating mechanism for guest exchange is proposed.

Full text of DOI:10.1002/chem.200601022

FULL PAPER
DOI: 10.1002/chem.200601022
Structure, Stability and Guest Affinity of Tris(3-ureidobenzyl)amine Capsules
in Solution
Mateo Alajarín,[a] Aurelia Pastor,*[a] Raffll-Angel Orenes,[a] Eloísa Martínez-Viviente,[b, c]
Heinz Rüegger,[b] and Paul S. Pregosin*[b]
Dedicated to Dr. JosØ Antonio Abad BaÇos on his retirement
Abstract: In non-competitive solvents,
the tris(3-ureidobenzyl)amines 1a–c
form dimeric assemblies in which
guests such as CH3CN, CH3NO2,
CH2Cl2, CH3I, CH2BrCl, CH2Br2,
CHCl3 and C6H6 can be encapsulated.
Variable temperature 1H and 1H,1H-
ROESY NMR spectroscopy, as well as
pulsed-gradient spin-echo (PGSE) dif-
fusion measurements were used to in-
vestigate the encapsulation within
1a·1a (1a: tris{3-[N’-(4-butylphenyl)-
ureido]benzyl}amine). Kinetic parame-
ters for the encapsulation of CH3NO2,
CH2Cl2 and CH3I, both in CDCl3 and
in [D8]toluene have been obtained by
using magnetisation transfer methods.
These data are discussed together with
the thermodynamic parameters. The af-
finity between guest and capsule seems
to be dictated mainly by the electronic,
size and shape complementarity be-
tween cavity and guest. A gating mech-
anism for guest exchange is proposed.
AHCTREUNG
AHCTREUNG
Keywords: diffusion
· encapsula-
tion · hydrogen bonds · molecular
recognition · self-assembly
Introduction
by a large molecular host, both held together by weak inter-
molecular forces.[1] The host/guest affinity depends on the
size, shape and chemical surfaces of the species involved.
For self-assembled capsules, the lifetime of the assemblies
can range from milliseconds to days.[2] This time span is long
enough for a variety of transformations to take place.[3]
Thus, the structure of the capsule and its selectivity towards
a certain guest are thermodynamically, rather than kinetical-
ly, determined. Surprisingly, relatively little quantitative in-
formation is known with respect to the kinetics and thermo-
dynamics of encapsulation processes.[4]
Encapsulation complexes are reversibly formed assemblies
in which a small molecular guest is completely surrounded
[a] Prof. M. Alajarín, Dr. A. Pastor, R.-A. Orenes
Departamento de Química Orgµnica
Facultad de Química, Universidad de Murcia
Campus de Espinardo, Murcia-30100 (Spain)
Fax : (+34)968-364-149
E-mail: aureliap@um.es
[b] Dr. E. Martínez-Viviente, Dr. H. Rüegger, Prof. P. S. Pregosin
Laboratory of Inorganic Chemistry
It was recently shown by some of us that tris(ureidoben-
zyl)amines are very useful modules in supramolecular
ETHZ, Hçnggerberg, 8093 Zürich (Schweiz)
Fax : (+41)1-632-1090
chemistry.[5]
The
tris(3-ureidobenzyl)amines
1a–c
E-mail: pregosin@inorg.chem.ethz.ch
(Scheme 1) associate by hydrogen bonding between the six
urea functions, forming a head-to-tail directional array of 12
hydrogen bonds.[5b] The structure and existence of the self-
assembled capsule 1a·1a·CH2Cl2 in the solid state was con-
firmed by means of X-ray analysis, while NMR spectroscopy
and electrospray ionisation mass spectrometry (ESIMS)
data supported the persistence of the dimeric aggregates 1·1
in non-competitive solvents.[5b] Preliminary investigations of
a solution of 1a·1a in C2D2Cl4 suggested the formation of
the capsules 1a·1a·CH2Cl2 and 1a·1a·CHCl3 upon addition
of small amounts of CH2Cl2 and CHCl3, respectively, to the
solution.[5b]
[c] Dr. E. Martínez-Viviente
Departamento de Química Inorgµnica
Facultad de Química, Universidad de Murcia
Campus de Espinardo, Murcia-30100 (Spain)
Supporting information for this article is available on the WWW
under http://www.chemistry.org or from the author. It contains
1H NMR spectra at 300 and 213 K of solutions of triurea 1a in
CDCl3, with and without addition of CH2Cl2; more detailed descrip-
tions of the determination of equilibrium constants, rate constants,
thermodynamic parameters and activation parameters; all data for
the Vanꢀt Hoff, Arrhenius and Eyring plots; VT H NMR spectra of
1a·1a in [D8]toluene; ROESY spectrum at 300 K of a solution of 1a
in [D6]DMSO; Figure 8 in colour.
1
Chem. Eur. J. 2007, 13, 1559 – 1569
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1559

Table 1. Changes in the 1H chemical shifts [ppm] of guest (G) and host
1a·1a (H; selected resonances) upon formation of the encapsulation
complex 1a·1a·G at 213 K in CDCl3 and the van der Waals volumes (V
in Š3) of each guest.
[a,c,d]
[a,c,e]
Guest
Dd(G)[a,b]
Dd(H)N ꢀ
1
Dd(H)C ꢀ
2
V (G)[f]
H
H
CH3CN
CH3NO2
CH2Cl2
CH3I
CH2BrCl
CH2Br2
CHCl3
ꢀ0.39
ꢀ0.40
ꢀ0.42
ꢀ0.33
ꢀ0.48
ꢀ0.52
ꢀ0.00
+0.44
+0.52
+0.31
+0.56
+0.26
+0.18
+0.00
+0.09
ꢀ0.05
+0.06
+0.16
+0.05
+0.06
+0.00
47
49
58
59
66
73
74
[a] From
1H NMR
(400 MHz)
spectra.
[b] dGencapꢀdGfree
.
1
ꢀ
[c] dHGꢀdHCDCl . [d] Resonances at dꢁ8 ppm, assigned to the N H pro-
Scheme 1. Tris(3-ureidobenzyl)amines 1a–c (left) dimerise in non-com-
petitive solvents to form molecular capsules in which guests can be ac-
commodated (right).
3
tons (see Scheme 1). [e] Resonances at 5.8ꢁdꢁ5.6 ppm, assigned to the
2
3
ꢀ
C
H protons (see Scheme 1). [f] Van der Waals volumes [Š ]. Calculat-
ed with Hyperchem 6.01.
One of the challenges in supramolecular chemistry and, in
particular, in the study of self-assembling capsules,[1] is their
adequate characterization in solution.[6] Recently, pulsed-
gradient spin-echo (PGSE) NMR diffusion measurements[7]
proved to be useful for demonstrating encapsulation and
studying the structure of hydrogen-bonded capsules in solu-
tion.[8] The diffusion coefficient, D, depends on the molecu-
lar shape and size (the larger the molecule, the lower the D
value). In solution, an encapsulated guest forms a kinetic
unit with the capsule. Consequently, both molecules move
together and their D values will be coincident even if they
have very different sizes. Encapsulation can thus be very ef-
ficiently detected by using PGSE experiments.
In this paper we show that a wide range of neutral mole-
cules, such as CH2ClBr, CH3CN, CH2Br2, C6H6, CH2Cl2,
CH3I and CH3NO2, can be accommodated inside the inter-
nal cavity of 1a·1a. PGSE[7] and magnetisation transfer
methods[9] were used to investigate the encapsulation of
CH3I, CH2Cl2 and CH3NO2 in two solvents of quite different
physical and chemical properties: CDCl3 and [D8]toluene.
Our kinetic and thermodynamic data contribute to the un-
derstanding of the encapsulation of guests inside self-assem-
bling hosts.
Results and Discussion
1
1H NMR spectra: 1D H NMR spectra of 1a·1a in CDCl3
(4–8 mm) with the addition of a 20- to 50-fold excess of
guest G (0.14–0.27m) were measured. The added guest parti-
ally displaces CDCl3 from inside 1a·1a,[5b] establishing an
equilibrium between the assemblies 1a·1a·CDCl3 and
1
1a·1a·G. Indeed, the H NMR spectra of these solutions at
room temperature show a broadening of the host signals
upon addition of the guest (see Supporting Information for
an example). At 213 K (Table 1 and Figure 1) the resonan-
ces of the host reveal two sets of signals that can be assigned
1
Figure 1. Sections of the H NMR spectra at 213 K for solutions of 1a·1a
in CDCl3 (2”10ꢀ3 m) before and after addition of a 20-fold excess of
guest (CH2Cl2, CH3I and CH3NO2). Selected peaks from 1a·1a·CDCl3
1
2
ꢀ
ꢀ
(N H, C H and one amine methylenic proton, see Scheme 1, and Fig-
ure S1 in the Supporting Information) are marked with circles,[10] the cor-
responding resonances from 1a·1a·G are indicated with triangles, the
guest within 1a·1a is distinguished by a trefoil and the free guest with a
star. The resonances from chloroform are overlapped with those of the
capsule. The spectra are truncated in the upper part.
[10]
to 1a·1a·CDCl3 and 1a·1a·G. New singlets, which can be
assigned to the encapsulated guest, appear at d=0.33–
0.52 ppm, shifted to lower frequency with respect to the res-
onances of the “free” guests. These new peaks integrate in a
1560
www.chemeurj.org
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 1559 – 1569

Capsules in Solution
FULL PAPER
1:1 ratio with respect to the set of signals attributed to
1a·1a·G.
1
ꢀ
For the new species 1a·1a·G the resonance from the N
H urea protons, originally at approximately 8 ppm,[11] shifts
to higher frequencies (see Figure 1 and Table 1). This shift
suggests that the net of hydrogen bonds within 1a·1a is
strengthened by the displacement of CDCl3 by any of the in-
vestigated guests, all of which have smaller van der Waals
volumes than CDCl3 (Table 1). Notably, the NMR chemical-
shift data suggest a subtle influence of the encapsulated
guest upon the structure of the capsule or, in other words,
the inside of the host “feels” the presence of the guest.
2
ꢀ
Thus, the resonances of the C H protons, which are direct-
ed towards the cavity, slightly change with the nature of the
guest (Figure 1 and Table 1).
1
Figure 2. Two different sections of the VT H NMR spectra (400 MHz) of
Contrary to the results shown above, the addition of small
amounts of benzene (V=84 Š3),[12] toluene (V=101 Š3)[12]
and C2H2Cl4 (V=106 Š3)[12] to solutions of 1a·1a in CDCl3
1a·1a in [D8]toluene (5”10ꢀ3 m) showing the resonances for the N
protons (9.20–8.80 ppm), the N H protons (8.20–7.80 ppm) and the C
H protons (6.10 ppm): a) 308, b) 288, c) 270, d) 250, e) 233 and f) 213 K.
H
1
ꢀ
2
2
ꢀ
ꢀ
1
produces no significant changes in the H NMR spectrum of
the host, either at room temperature or at low temperature.
It seems that the smaller CDCl3 fits better inside 1a·1a than
these added, larger molecules, and, consequently, the deuter-
ated solvent is not displaced. In two further experiments, an
excess of toluene was added to a solution of 1a·1a in
C2D2Cl4, and an excess of C2H2Cl4 was added to a solution
2
ꢀ
carded, as the resonance from the C H protons, at approxi-
mately 6.1 ppm, does not split (Figure 2). These protons are
directed towards the cavity (Scheme 1) and would be affect-
ed by the presence of a guest. We suggest that two slightly
different conformations of the empty capsule 1a·1a exist in
solution. At room temperature these two conformations
would be in fast exchange. Decreasing the temperature
would allow the two forms to be distinguished by means of
NMR spectroscopy, while simultaneously favouring the most
stable.
1
of 1a·1a in [D8]toluene. The H resonances from 1a·1a in
both solutions showed no changes upon addition of the
second species, either at room temperature or at low tem-
perature.[13] No new signals for either toluene or C2H2Cl4
could be observed. To eliminate the possibility of both sol-
vents having similar (and low) affinities for 1a·1a, in a sepa-
rate experiment 1a·1a was dissolved in a mixture of C2D2Cl4
and [D8]toluene. 1H NMR spectra were then recorded at
300 and 213 K. Instead of two sets of signals (for the hypo-
thetical 1a·1a·C2D2Cl4 and 1a·1a·[D8]toluene), only one set
was observed, which was assigned to the empty capsule.[14] It
We have performed encapsulation experiments with sev-
eral guests in solutions of 1a·1a in [D8]toluene.[15] Upon ad-
1
dition of the guest, the H NMR spectra at 213 K (Figure 3)
show the presence of a new set of signals, assigned to
1a·1a·G, while the resonances from the empty host, 1a·1a,
strongly decrease or even disappear. Averaged spectra are
observed at 298 K (not shown). In [D8]toluene the encapsu-
lated guests resonate at similar frequencies to those in
CDCl3 (compare Figures 1 and 3).[16] This similarity of chem-
ical shifts, in spite of the different properties of the two sol-
vents, can be explained by the shielding effect of the host.
Indeed, the free guests in [D8]toluene resonate at approxi-
mately 1 ppm or more below their chemical shifts in
CDCl3.[17]
Table 2 shows an almost general decrease of the chemical
shift of the N H protons of 1a·1a upon encapsulation of the
guest in [D8]toluene; an indication of the weakening of the
hydrogen bonds within the capsule. In the series G=
CH2Cl2, CH2BrCl, CH2Br2 and CHCl3, the larger the size of
3
seems that guests with volumes exceeding 101 Š cannot be
associated with the host.
As the resonances for 1a·1a in [D8]toluene are sharper
than those in C2D2Cl4, subsequent experiments were con-
ducted in the former solvent. The resonances from the urea
protons of 1a·1a in [D8]toluene are shifted to higher fre-
quencies (d=8.78, 7.84 ppm at 308 K) with respect to CDCl3
(d=8.12, 6.70 ppm at 303 K). The new values might indicate
a more strongly hydrogen-bonded ring of urea groups in
[D8]toluene than in CDCl3. Although the lower polarity of
[D8]toluene could be playing a role in this result, we suggest
that the absence of a guest inside the capsule is at least par-
tially responsible for the stronger hydrogen bonds of the
urea groups in toluene. Curiously, in this solvent the reso-
nances from the NH protons split in two at 250 K with Dd<
0.06 ppm at 400 MHz (see Figure 2 and the Supporting In-
formation for the complete spectra). At even lower temper-
atures, the intensity of one of the two sets of signals increas-
es at the expense of the other, until one of the sets almost
disappears at 213 K. The possibility that the two sets of reso-
nances would belong to empty and filled capsule was dis-
ꢀ
the guest, the greater the weakening of the hydrogen bonds.
The C H protons, directed towards the cavity, are signifi-
cantly affected by the presence of the guest.
2
ꢀ
ROESY spectrum: An H,1H ROESY spectrum of 1a·1a in
1
CDCl3 (1.1”10ꢀ2 m) in the presence of a 30-fold excess of
CH2Cl2 was measured at 223 K. Under these conditions a
mixture of the capsules 1a·1a·CDCl3 and 1a·1a·CH2Cl2
Chem. Eur. J. 2007, 13, 1559 – 1569
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1561

A. Pastor, P. S. Pregosin et al.
Figure 4. Intramolecular (solid arrows) and intermolecular (dashed
arrows) ROE interactions found within the dimeric capsules 1a·1a in the
1H,1H ROESY spectrum of a solution in CDCl3 (1.1”10ꢀ2 m), in the pres-
ence of a 30-fold excess of CH2Cl2, at 223 K.
mately 7.1 ppm and only one of the two non-equivalent
methylenic protons of the amine moiety (at approximately
2.6 ppm, Figure 5a). Consequently, these methylenic protons
1
Figure 3. Sections of the H NMR spectra at 213 K for solutions of 1a·1a
ꢀ
are assigned as being in a “pseudoaxial” position (C Hax
in [D8]toluene (2”10ꢀ3 m), before and after addition of a 20-fold excess
of guest (CH2Cl2, CH3I and CH3NO2). Selected peaks from 1a·1a (N H,
bond parallel to the C3 axis of the dimer). The “pseudoequa-
1
ꢀ
2
2
torial” methylenic protons at 3.5–3.6 ppm (Heq) show intra-
ꢀ
ꢀ
N
H and C H, see Scheme 1, and Figure S1 in the Supporting Informa-
2
ꢀ
tion) are marked with circles, the corresponding resonances from
1a·1a·G are indicated with triangles, the encapsulated guest is distin-
guished by a trefoil and the free guest with a star. The spectra are trun-
cated in the upper part.
molecular ROE cross peaks with the aromatic C H protons
at approximately 5.6–5.7 ppm (also in Figure 5a). Figure 5b
2
ꢀ
shows ROE cross peaks between the same C H protons
and
the
resonances
at
6.74 ppm (for 1a·1a·CDCl3)
and 7.18 ppm (for
1a·1a·CH2Cl2), which supports
Table 2. Changes in the 1H chemical shifts [ppm] of guest (G) and host 1a·1a (H; selected resonances) upon
formation of the encapsulation complex 1a·1a·G at 213 K in [D8]toluene and the van der Waals volumes (V in
Š3) of each guest.
the previous assignment of
[a,c,d]
[a,c,e]
[a,c,f]
Guest
Dd(G)[a,b]
Dd(H)N ꢀ
1
Dd(H)N ꢀ
2
Dd(H)C ꢀ
2
H
V (G)[g]
H
H
2
ꢀ
these peaks to the N H pro-
CH3CN
CH3NO2
CH2Cl2
CH3I[h]
CH2BrCl
CH2Br2
CHCl3
C6H6
+1.48
+1.71
+1.30
+0.95
+1.21
+1.16
ꢀ0.09
ꢀ0.02
ꢀ0.21
+0.05
ꢀ0.27
ꢀ0.34
ꢀ0.48
ꢀ0.23
ꢀ0.32
ꢀ0.26
ꢀ0.45
ꢀ0.46
ꢀ0.60
ꢀ0.70
ꢀ0.06
ꢀ0.18
ꢀ0.10
ꢀ0.01
ꢀ0.11
ꢀ0.10
ꢀ0.18
ꢀ0.19
47
49
58
59
66
73
74
84
tons. Additionally, the ROESY
spectrum also shows cross
peaks between the two mole-
cules of 1a in each capsule
(Figure 5a, c). These intermo-
lecular contacts confirm the
capsular structure of 1a·1a, be-
cause the aromatic protons of
the p-butylphenyl groups of
one molecule of 1a interact
[h]
[i]
[j]
–
–
[i]
[j]
–
–
[a] From 1H NMR (400 MHz) spectra. [b] dGencapꢀdGfree. [c] dHGꢀdHempty. [d] Resonances at dꢁ8.7 ppm, as-
1
2
ꢀ
ꢀ
signed to the N H protons (see Scheme 1). [e] Resonances at 8.2>d>7.3 ppm, assigned to the N H protons
2
ꢀ
(Scheme 1). [f] Resonances at 6.2>d>5.8 ppm, assigned to the C H protons (see Scheme 1). [g] Calculated
with Hyperchem 6.01. [h] At this concentration only the resonances of the filled capsule were observed. [i] The
resonance from the encapsulated guest overlaps with those due to the aromatic protons of 1a·1a and 1a·1a·G.
via ROE with the tribenzyl-
R
2
ꢀ
[j] The resonances from the N H protons in 1a·1a·G overlap with those due to the aromatic protons of 1a·1a
and 1a·1a·G.
A
molecule.[18] Undoubtedly, the
most relevant signal in the
exists, with the latter being the major species. Figure 4
shows the most significant intramolecular and intermolecu-
lar ROE contacts found within the dimeric capsules
1a·1a·G. Figure 5 shows the relevant sections of the
ROESY spectrum.
ROESY spectrum is shown in Figure 6: a cross peak be-
tween the resonance assigned to the encapsulated CH2Cl2
2
ꢀ
(at 4.95 ppm) and the C H protons of 1a·1a·CH2Cl2 (at ca.
5.70 ppm). This is consistent with the presence of a guest
molecule of CH2Cl2 inside the cavity of 1a·1a. No other
cross peaks between the encapsulated guest and the host
were found.
Intramolecular ROE contacts are found within each mole-
6
ꢀ
cule of 1a between the aromatic C H protons at approxi-
1562
www.chemeurj.org
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 1559 – 1569

Capsules in Solution
FULL PAPER
Figure 5. Three different sections of the 1H,1H ROESY spectrum of a solution of 1a·1a in CDCl3 (1.1”10ꢀ2 m) in the presence of a 30-fold excess of
CH2Cl2, measured at 223 K. Selected peaks from 1a·1a·CDCl3 are marked with circles, and the corresponding peaks from 1a·1a·CH2Cl2 are marked with
triangles. The ROE cross peaks due to intramolecular interactions (within one molecule of 1a) are in grey, while those due to intermolecular interactions
between the two molecules of 1a in each capsule are in black. The three sections have the same plotting cut-off levels.
Table 3. Diffusion coefficients,
D
(”10ꢀ10 m2 sꢀ1), and hydrodynamic
radii, rH [Š], for solutions of 1a·1a (2”10ꢀ3 m) plus a 20-fold excess of
guest (G) in CDCl3.[a]
300 K
1a·1a (average)[b]
G (average)[c]
[e]
[e,f]
Guest
(CDCl3)
CH2Cl2
CH3NO2
CH3I
D[d]
rH
D[d]
rH
G
4.63
4.63
4.61
4.65
9.0
9.0
9.0
8.9
–
–
26.6
25.7
25.4
1.6 (2.8)
1.6 (2.9)
1.6 (2.9)
213 K
[g]
1a·1a·CDCl3
1a·1a·G[g]
Encap G
Free G
[e]
[e]
[e]
[e]
Guest
D[d]
rH
D[d]
rH
D[d]
rH
D[d]
rH
A
0.885
0.894
9.4
9.3
9.4
9.3
–
–
–
–
–
–
0.889 9.4
0.881 9.5
0.892 9.3
0.880 9.5
0.881 9.5
0.894 9.3
6.08 1.4
5.17 1.6
5.43 1.5
1
Figure 6. Section of the H,1H ROESY spectrum of a solution of 1a·1a in
CH3NO2 0.884
CH3I 0.895
CDCl3 (1.1”10ꢀ2 m) in the presence of a 30-fold excess of CH2Cl2, at
[a] h(CHCl3, 300 K)=0.529”10ꢀ3 Kgsꢀ1 mꢀ1
; h(CHCl3, 213 K)=1.87”
223 K. The cross peak relating the resonances assigned to the encapsulat-
2
10ꢀ3 Kgsꢀ1 mꢀ1. The peak from the chloroform solvent is masked by the
resonances from 1a·1a. [b] The resonances from 1a·1a·CDCl3 and
1a·1a·G are not resolved at 300 K when G=CH2Cl2 and CH3NO2 (the D
value measured on the average resonances of 1a·1a is given). For G=
CH3I, the average of the D values measured on the separate resonances
ꢀ
ed CH2Cl2 and the C H protons of 1a·1a·CH2Cl2 (marked with a trian-
2
ꢀ
gle) is shown. The resonance of the C H protons of 1a·1a·CDCl3 is
marked with a circle. The plotting cut-off levels are the same as those in
Figure 5.
of 1a·1a·CH3I (D=4.64”10ꢀ3 m2 sꢀ1
) and 1a·1a·CDCl3 (D=4.66”
10ꢀ3 m2 sꢀ1) is given. [c] The resonances from the encapsulated and free
guest are not resolved. [d] Experimental error is less than ꢂ2%. [e] Stan-
dard deviation is approximately ꢂ0.1 Š. [f] The radii calculated by using
a factor 3.4 instead of 6 in the Stokes–Einstein equation are given in pa-
rentheses (see ref. [19]). [g] The average of the D values measured for
the three resonances marked in Figure 1, is given. These D values were
identical within the experimental error.
1
1H PGSE diffusion investigations: We carried out H PGSE
diffusion measurements on solutions of 1a·1a in CDCl3 plus
a 20-fold excess of G, where G=CH3NO2, CH2Cl2 and
CH3I, at 300 and 213 K. At the lower temperature, the in-
tensity decays were followed on the separate resonances
marked in Figure 1. The average peaks were used at 300 K.
The results are shown in Table 3, in which the hydrodynamic
radii have been calculated from the D values by using the
Stokes–Einstein equation.[19]
At 300 K, the D values measured for the average signal of
the capsule (from 1a·1a·CDCl3 and 1a·1a·G) are similar for
the three guests (D=4.63ꢂ0.02). The result is the same for
1a·1a·CDCl3 in the absence of added guest. The coincidence
is not surprising, as the presence of a guest is not expected
to affect the size of the resulting host complex. The calculat-
ed rH value is 9.0ꢂ0.1 Š, which is in good agreement with
the size estimated from the X-ray structure determination.[20]
The observed average resonance for the guests at 300 K
stems mainly from the free guest molecules (the contribu-
tion of 1a·1a·G, is very small, as G is added in a 20-fold
excess with respect to 1a·1a). Consequently, the measured
average D values for the guests at 300 K are, as expected,
much larger than those for 1a·1a, and the rH values are, cor-
respondingly, much smaller.[19] No significant difference in
size between the three guests can be detected by using this
method. At 213 K, the D values for the host 1a·1a in the
presence of CH3NO2, CH2Cl2 or CH3I are all identical
Chem. Eur. J. 2007, 13, 1559 – 1569
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1563

A. Pastor, P. S. Pregosin et al.
D
(”10ꢀ10 m2 sꢀ1), and hydrodynamic
Table 4. Diffusion coefficients,
within
the
experimental
error
(D=0.889ꢂ0.01”
radii, rH [Š], for solutions (2”10ꢀ3 m) of 1a·1a plus a 20-fold excess of
guest (G) in [D8]toluene.
10ꢀ10 m2 sꢀ1). This diffusion coefficient is much smaller than
the value found at 300 K, owing to the higher viscosity of
the solvent at the lower temperature. The calculated hydro-
dynamic radius compensates for the change in viscosity,[19]
and affords a value of 9.4ꢂ0.1 Š, which is in fairly good
agreement with the estimation at 300 K.[21] But the most in-
teresting diffusion data in Table 3 correspond to the mea-
300 K
1a·1a (average)[a]
G (average)[b]
Toluene
[d]
[d,e]
[d,f]
Guest
D[c]
rH
D[c]
rH
D[c]
rH
–
4.34
4.38
4.37
4.43
9.4
9.3
9.3
9.2
–
–
20.5 2.0 (3.4)
20.8 1.9 (3.4)
20.8 2.0 (3.4)
20.7 2.0 (3.4)
CH2Cl2
CH3NO2
CH3I
27.2
26.7
26.7
1.5 (2.8)
1.5 (2.9)
1.5 (2.8)
A
guests at 213 K. The D values for these signals (0.880, 0.881
and 0.894”10ꢀ10 m2 sꢀ1) are all identical, within experimental
error, and close to those of 1a·1a, clearly confirming that
the host and guest exist together in solution as a single unit.
For the resonances of the free guests the D values are much
larger.
213 K
1a·1a
1a·1a·G
Encap G
Free G
Toluene
[d]
[d]
[d]
[d]
[d]
Guest
D[c]
rH
D[c]
rH
D[c]
rH
D[c] rH
D[c] rH
–
0.533 9.8
–
–
–
–
–
–
2.88 1.8
CH2Cl2
CH3NO2
CH3I
0.538 9.8 0.538 9.8 0.541 9.7 4.39 1.2 2.93 1.8
[g]
–
–
–
–
0.533 9.8 0.535 9.8 4.14 1.3 2.92 1.8
0.533 9.8 0.539 9.7 3.94 1.3 2.91 1.8
[h]
Figure 7 shows a plot of the diffusion results at 213 K for
a solution of 1a·1a in CDCl3 plus a 20-fold excess of CH3I.
[a] h(toluene, 300 K)=0.537”10ꢀ3 Kgsꢀ1 mꢀ1; h(toluene, 213 K)=2.97”
10ꢀ3 Kgsꢀ1 mꢀ1. The resonances from the empty capsule 1a·1a and the
capsule with guest (1a·1a·G) are not resolved at 300 K. The D value
measured on the average signal of 1a·1a is given. [b] The resonances
from the encapsulated and free guest are not resolved. [c] Experimental
error is approximately ꢂ2%. [d] Standard deviation is approximately
ꢂ0.1 Š. [e] The radii calculated by using a factor 3.2 instead of 6 in the
Stokes–Einstein equation are shown in parentheses. [f] The radii calculat-
ed by using a factor 3.5 instead of 6 in the Stokes–Einstein equation are
shown in parentheses (see ref. [19]). [g] The signals for the empty capsule
1a·1a are too weak for measuring its D value. [h] No signals for the
empty capsule 1a·1a can be detected.
The coincidence of the
D
value for the capsules
1a·1a·CDCl3 and 1a·1a·CH3I and the encapsulated CH3I
can be clearly appreciated, while the free CH3I moves much
faster.
213 K, which afford the same D values as the host. This
result clearly shows again that each host–guest assembly
acts as a kinetic unit in solution. The calculated hydro-
dynamic radii are slightly larger in [D8]toluene than in
CDCl3.
Diffusion measurements on a sample of only the solvent
[D8]toluene (on the residual proton resonance) afforded D
values of 21.0”10ꢀ10 m2 sꢀ1 at 300 K and 2.97”10ꢀ10 m2 sꢀ1 at
213 K, similar to those given for this solvent in Table 4. Con-
sequently, the presence of 1a·1a, at least at our relatively
low concentrations, does not affect the diffusion of the sol-
vent. A similar comparison with CDCl3 was not possible,
due to overlap with resonances from 1a·1a.
Figure 7. Plot of ln(I/Io) versus arbitrary units proportional to the square
A
1
of the gradient amplitude, for a H PGSE diffusion measurement of a 2”
10ꢀ3 m solution of 1a·1a in CDCl3 plus a 20-fold excess of CH3I. The
*
) and
measurements on the resonances of the capsule in 1a·1a·CDCl3
(
Finally, we investigated a solution of 1a in [D6]DMSO
(2”10ꢀ3 m). In this strongly hydrogen-bonding solvent, 1a
exists as a monomeric, non-assembled species.[5b] The mea-
*
1a·1a·CH3I ( ), as well as on the resonance from the encapsulated CH3I
(&) afford very similar intensity decays, while on the resonance from the
~
free CH3I ( ) the intensity decay is much stronger, indicating a much
sured
D
value, 1.28”10ꢀ10 m2 sꢀ1, corresponds to rH =
larger D value. Diffusion parameters (see the Experimental Section): D=
65.5 ms, d=4.5 ms, number of scans=40.
9.0 Š,[22] which is, surprisingly, the same hydrodynamic
radius as found for the dimer 1a·1a in CDCl3. We have pre-
viously reported D values in [D6]DMSO for related com-
pounds,[5d] and found that solvation and/or hydrogen bond-
ing by [D6]DMSO produces an increase in the hydrodynam-
ic radius of the solute of approximately 1 Š with respect to
CDCl3. With this information, a “corrected” rH value for
monomeric 1a would be approximately 8 Š, which is in
better agreement with an estimation made with the
Chem3D modelling software (8.1 Š).
1H PGSE diffusion measurements have also been per-
formed on solutions of 1a·1a in [D8]toluene plus the same
guests (G=CH3I, CH2Cl2, and CH3NO2). The results are
shown in Table 4. As expected, the free guest and the sol-
vent move much faster than the host. As in Table 3, the
presence of the guests does not significantly affect the D
values for the host in [D8]toluene (4.39ꢂ0.05”10ꢀ10 m2 sꢀ1
at 300 K and 0.537ꢂ0.004”10ꢀ10 m2 sꢀ1 at 213 K). Yet again,
the most important information comes from the measure-
ments on the resonances from the encapsulated guest, at
Molecular modelling: We have attempted a simple molecu-
lar modelling investigation on our tris(3-ureidobenzyl)-
1564
www.chemeurj.org
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 1559 – 1569

Capsules in Solution
FULL PAPER
A
being similar to the value found in the empty dimer. As for
the hydrogen bonds, they are slightly weakened by the pres-
ence of the guest, with those involving N2 being more affect-
8.1 (AMBER* force field). For simplicity, the pendant n-
butyl chains of 1a have been replaced by methyl groups, as
in 1b (Scheme 1). We have focused on the empty dimer
1b·1b and the capsules 1b·1b·CH3NO2, 1b·1b·CH2Cl2,
1b·1b·CH3I and 1b·1b·CHCl3, in the presence of CDCl3 as
the solvent (Figure 8).
ed than those involving N1: d
(N2···O=C)=2.83–3.16 Š.
The encapsulation of the larger guests CH2Cl2, CH3I and
CHCl3 (Figure 8c–e) leads to an increase in the distance be-
tween the two amine nitrogen atoms (d(N···N)=10.14 Š,
G
d
ACHTREUNG
AHCTREUNG
10.26 Š and 10.76 Š, respectively) with respect to 1b·1b
and 1b·1b·CH3NO2. Similarly, the hydrogen bonds in
1b·1b·CH2Cl2, 1b·1b·CH3I, and 1b·1b·CHCl3 are signifi-
cantly weakened and, again, those involving N2 are more
strongly affected (for CH2Cl2: d
and d
(N2···O=C)=2.92–3.45 Š; for CH3I: d
2.76–2.82 Š and d
(N2···O=C)=2.85–3.62 Š; for CHCl3: d-
(N1···O=C)=2.74–2.85 Š and d(N2···O=C)=2.93–3.80 Š).[23]
In conclusion, the hydrogen bonds involving N1 are the
ACHTREUNG
A
ACHTREUNG
AHCTREUNG
A
ACHTREUNG
strongest and least affected by encapsulation. These obser-
vations support the assignment of the proton resonance at
1
ꢀ
higher frequency to the N H protons, which was based on
the ROESY spectrum of 1a·1a·CH2Cl2 (see above). In tolu-
1
ꢀ
ene (Table 2) this N H resonance is less sensitive to the
presence of an encapsulated guest (jDdj=0.02–0.48 ppm)
2
ꢀ
than the N H resonance (jDdj=0.26–0.70 ppm).
Equilibria: To gain insight into the mechanism of the encap-
sulation process, we have investigated the kinetics and ther-
modynamics of the equilibria according to Equations (1)
and (2) (in CDCl3 and [D8]toluene, respectively):
kD
1 a Á 1 a Á CDCl þ G
1 a Á 1 a Á G þ CDCl
ð1Þ
ð2Þ
G
H
3
3
kI
kD
1 a Á 1 a þ G
1 a Á 1 a Á G
H
G
kI
in which G=CH2Cl2, CH3I, CH3NO2, and kD and kI are the
rate constants for the forward and the reverse reactions, re-
spectively. CH2Cl2, CH3I and CH3NO2 were chosen as
guests, because they form the most stable complexes.
Figure 8. Energy-minimized structures (MacroModel 8.1, AMBER* force
field) of a) 1b·1b, b) 1b·1b·CH3NO2, c) 1b·1b·CH2Cl2, d) 1b·1b·CH3I
and e) 1b·1b·CHCl3. All hydrogen atoms and methyl groups of the p-
tolyl fragments have been omitted for clarity.
Thermodynamic data: At temperatures between approxi-
mately 228 and 273 K the equilibria shown in Equations (1)
and (2) are slow on the NMR time scale, affording separate
resonances for the involved species. The association con-
stants in this range of temperatures have been determined
from the integrals of the separate peaks (see the Supporting
Information). From a Vanꢀt Hoff plot (lnKa vs. 1/T) we have
calculated the thermodynamic parameters DH and DS, as-
suming that they are not temperature dependent. The re-
sults are shown in Table 5, together with the DG values at
298, 273 and 213 K.
The lowest energy conformation for 1b·1b was deter-
mined by Monte Carlo multiple minimisation. Each of the
four guests was then placed inside this structure and a new
calculation was performed to obtain the minimised struc-
tures of the filled capsules. The empty dimer 1b·1b shows a
perfect S6 symmetry with a distance between the two pivo-
tal, amine nitrogen atoms of 10.04 Š (Figure 8a). The hydro-
gen bonds involving the terminal nitrogen atom of the urea
functionality (N1 in Scheme 1) are stronger than those in-
The values found for DG reflect some of the facts previ-
1
volving N2:
d
G
d
G
ously observed in the H NMR spectra shown in Figures 1
2.91 Š (d=distance). The S6 symmetry is broken in
1b·1b·CH3NO2 (Figure 8b) although an averaged structure
is expected in solution. The presence of the guest CH3NO2
does not affect the separation between the two subunits; the
distance between the two amine nitrogen atoms (10.07 Š)
and 3: Thus, 1) DG for Equations (1) and (2) is always nega-
tive and, indeed, the encapsulation of the guest takes place,
1
as shown by the H NMR spectra and 2) the largest jDG213 j,
in both solvents, is found for CH3I, for which the degree of
encapsulation is seen to be larger (i.e., for the same concen-
Chem. Eur. J. 2007, 13, 1559 – 1569
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1565

A. Pastor, P. S. Pregosin et al.
Table 5. Thermodynamic parameters for the reactions shown in Equa-
tion (1) (CDCl3) and Equation (2) ([D8]toluene).
positive, suggesting that the liberation of CDCl3 completely
compensates for the loss of entropy produced by the encap-
sulation of the added guest. Only the positive value of DS
for the encapsulation of CH3NO2 in [D8]toluene is difficult
to understand, and might be related to more subtle thermo-
dynamic contributions, such as specific heat and/or volume
differences between the capsule and their separated compo-
nents, that is, dimer and guest. The trend in DS is CH3NO2 >
CH2Cl2 >CH3I in the two solvents investigated. The forma-
tion of 1a·1a·CH3NO2 might be entropically favoured due
to its somewhat smaller size, which would allow a greater
mobility of the guest within the capsule than for the larger
CH3I and CH2Cl2. These two guests, additionally, could dis-
tort the structure of the capsule more significantly than
CH3NO2.
[a]
[a]
[a]
Solvent
Guest
DH[a]
DS[b]
DG298
DG273
DG213
CDCl3
CDCl3
CDCl3
[D8]toluene CH3NO2
[D8]toluene CH2Cl2
[D8]toluene CH3I
CH3NO2
CH2Cl2
CH3I
0.0
ꢀ2.6
ꢀ6.8
ꢀ8.8
42.6 ꢀ12.7
28.8 ꢀ11.2
22.0 ꢀ13.3
14.9 ꢀ13.2
ꢀ11.6
ꢀ10.5
ꢀ12.8
ꢀ12.8
ꢀ9.5
ꢀ9.1
ꢀ8.7
ꢀ11.5
ꢀ11.9
ꢀ10.7
ꢀ15.2
ꢀ14.8 ꢀ19.3
ꢀ9.1
ꢀ21.4 ꢀ29.0 ꢀ12.8
ꢀ13.5
[a] In kJmolꢀ1. [b] In JKꢀ1 molꢀ1
.
tration of added guest the ratio 1a·1a·G/1a·1a·CDCl3 or
1a·1a·G/1a·1a is larger for CH3I). The encapsulation of the
three guests is enthalpically favoured, both starting from an
empty capsule [in [D8]toluene, Eq. (2)] or when a CDCl3–
guest exchange takes place [Eq. (1)]. The only exception is
the exchange of CDCl3 by CH3NO2, which is energetically
neutral (DH=0.0 kJmolꢀ1). Generally speaking, a “filled”
capsule is favoured over an empty host because of the at-
tractive capsule–guest interactions (in our case, mainly van
der Waals forces).[4a,24] The larger jDHj values for Equa-
tion (2) than for Equation (1) are in agreement with this
principle. These attractive interactions between guest and
host may compensate for the weakening of the hydrogen
bonds within 1a·1a upon encapsulation of the guest. In
CDCl3, the clearly favourable DH values for the exchange
of CDCl3 by CH2Cl2 and CH3I suggest a stronger affinity of
the capsule for these two guests than for CDCl3.
The data in Table 5 also indicate that the encapsulation of
the guest is enthalpically favoured in the order CH3I>
CH2Cl2 >CH3NO2, both in CDCl3 and [D8]toluene, in spite
of the different polar and hydrogen-bonding characteristics
of the two solvents. Thus, the guest–solvent interaction does
not seem to be playing an important role in the relative DH
values. The ability of the guest to hydrogen bond with the
capsule does not seem to be a determining factor either, as
CH3NO2 should then probably be the most favoured. Conse-
quently, the affinity between guest and capsule seems to be
dictated mainly by the electronic, size and shape characteris-
tics of guest and cavity.
The decreasing order in DS is the opposite to that for
jDHj, showing a compensation between the entropic and
enthalpic factors. A plot of DH versus DS (Figure 9) gives a
Figure 9. Plot of DH versus DS for the encapsulation of CH3I, CH2Cl2
and CH3NO2 inside 1a·1a in both solvents, CDCl3 and [D8]toluene.
good linear correlation (R=0.96) with a slope of 276 K
(Tc).[27,28] As DH becomes more negative (stronger interac-
tions), DS tends to decrease due to the lower freedom in the
system, and vice versa.[29] This compensation between en-
thalpy and entropy has been repeatedly observed for the as-
sociation between a receptor and a ligand for a great
number of systems.[27b,30]
Kinetic data: Using magnetisation transfer techniques, we
have measured the rate constants kD and kI for Equa-
tions (1) and (2) at several low temperatures (see the Sup-
porting Information). From these data we have derived the
activation parameters for the forward and reverse reactions,
as given in Table 6.
It is known that the optimum packing coefficient in en-
capsulation complexes is 55%.[25] Thus, the van der Waals
volume of CH3I (59 Š3, Table 1, entry 4) may be used to
roughly estimate the volume of the cavity in 1a·1a. The
result, approximately 110 Š3, is intermediate between the
values for the tennis ball (50–
55 Š3)[4a] and the tetraureido-
calix[4]arene dimers (210 Š3)[26]
reported by Rebek.
Table 6. Activation parameters for the forward and reverse encapsulation reactions shown in Equation (1) (in
CDCl3) and Equation (2) (in [D8]toluene).
Forward reaction
Reverse reaction
Some of the entropy data in
Table 5 also support the initial
situation of an empty capsule
in [D8]toluene [Eq. (2)]. Thus,
in this solvent, DS is strongly
negative for the encapsulation
of CH2Cl2 and CH3I, while for
the same guests in CDCl3 DS is
[a]
¼[a]
¼[b]
¼
[a]
[a]
¼[a]
¼[b]
¼
[a]
Solvent
Guest
EA
DH
DS
DG
EA
DH
DS
DG
298
298
CDCl3
CDCl3
CDCl3
CH3NO2
CH2Cl2
CH3I
CH3NO2
CH2Cl2
CH3I
56.6
54.0
70.7
56.1
51.1
61.1
54.5
51.9
68.4
54.0
49.0
58.9
ꢀ27.6
ꢀ43.5
2.7
ꢀ46.9
ꢀ63.5
ꢀ51.5
62.7
64.9
67.6
68.0
67.9
74.2
56.0
56.8
78.4
61.1
61.4
69.2
53.9
54.7
76.1
59.0
59.2
66.9
ꢀ26.4
ꢀ23.2
32.2
ꢀ14.2
ꢀ4.75
ꢀ8.67
61.8
61.6
66.5
63.2
60.6
69.5
[D8]toluene
[D8]toluene
[D8]toluene
[a] In kJmolꢀ1. [b] In JKꢀ1 molꢀ1
.
1566
www.chemeurj.org
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 1559 – 1569

Capsules in Solution
FULL PAPER
¼
The activation energies (EA and DH ) for Equations (1)
and (2) increase in the order CH2Cl2 <CH3NO2 <CH3I for
the forward reaction, while for the reverse process
CH2Cl2ffiCH3NO2 !CH3I. The trend is the same in both sol-
vents, again discarding the solvent–guest interactions as a
major differentiating factor. All the activation enthalpies
been observed. A gating mechanism is proposed for the
guest exchange.
Experimental Section
¼
PGSE NMR diffusion measurements: The PGSE NMR diffusion mea-
(DH ) are high and positive, probably because a partial or
AHCTREUNG
total rupture of the hydrogen-bonded ring of urea molecules
is necessary for the inclusion/exclusion of the guests.
AHCTREUNG
were performed on a 500 MHz Bruker AVANCE spectrometer, equipped
with a microprocessor-controlled gradient unit and a multinuclear inverse
probe with an actively shielded Z-gradient coil. The measurements were
carried out without spinning. The shape of the gradient pulse was rectan-
gular, and its strength was varied automatically during the course of the
experiments. The D values were determined from the slope of the regres-
The activation entropies for the forward reactions in
[D8]toluene [Eq. (2)] are strongly negative, as expected for a
process in which two species, host and guest, are brought to-
¼
gether. In CDCl3, the strong differences in the DS for
sion line ln
(I/Io) versus G2, according to Equation (3) in which I/Io =ob-
U
CH3NO2 and CH2Cl2, both negatives, and CH3I, which is
positive, suggest some change in the mechanism of the pro-
cess, due to the different properties of the guests.
served spin echo intensity/intensity without gradients, G=gradient
strength, D=delay between the midpoints of the gradients, D=diffusion
coefficient, g=gyromagnetic ratio and d=gradient length. The calibra-
tion of the gradients was carried out by means of a diffusion measure-
ment of HDO in D2O (DHDO =1.9”10ꢀ9 m2 sꢀ1).[36]
The inclusion/exclusion of the guest from the capsule may
take place by: 1) a dissociation–recombination mechanism,
in which a complete dissociation of the capsule takes place,
or 2) a gating mechanism, in which a temporary gate is
opened by partial dissociation of the capsule, allowing the
guest to come in or out. Interestingly, the exchange of 1a by
1c in a solution of homodimers 1a·1a, 1c·1c and heterodi-
mer 1a·1c is slow on the NMR time scale at room tempera-
ture,[31] while the inclusion/exclusion of a guest in 1a·1a is
fast. These observations point to a gating mechanism for the
guest exchange,[32] as was also found for Rebekꢀs tennis
ball,[4e] softballs[33] and cylindrical capsules.[34] In contrast,
the exchange of benzene in tetraureidocalix[4]arene dimers
requires capsule dissociation.[4c]
ꢀ
ꢁ
ꢀ
ꢁ
I
Io
d
2
ln
¼ ꢀðgdÞ G2 Dꢀ
D
ð3Þ
3
The precise temperatures were determined by introducing a thermocou-
ple inside the bore of the magnet, before and after the measurements.
For the experiments at 300 K the airflow was disconnected, to avoid per-
turbations of the mechanical stability of the sample. In the experiments
at 233 K, the onset of convection currents due to the cooling N2 flow was
avoided by the use of two commercially available coaxial NMR tubes,
separated by air, and kept concentric by a Pyrex spacer.[37] The inner
tube had an internal diameter of 1.96 mm and an external diameter of
2.97 mm. A standard 5 mm vessel was used for the outer tube. The values
reported in Tables 3 and 4 are the average of three different measure-
ments, which yielded D values within a maximum of ꢂ1.5% of the re-
1
ported one. All the measurements were carried out by using the H reso-
nances. Typically, 14–20 points were used for the regression analysis and
the experimental time was between approximately 30 min and 3 h. A re-
laxation delay of 5 s was routinely used. All of the observed data leading
to the reported D values afforded lines whose correlation coefficients
were above 0.999. Based on our experience of diffusion over several
years, we propose an experimental error of ꢂ2% for the D values and a
standard deviation of ꢂ0.1 Š for the hydrodynamic radii.
For the measurements in CDCl3 and [D8]toluene the gradient length, d,
was set to 1.75 ms and the diffusion delay was approximately 68, 118 or
168 ms. The number of scans was usually 16 to 28. The only exceptions
were the measurements on the signals of the solvent or guests, where
shorter diffusion delays, approximately 27 ms, were used, to compensate
for the higher mobility of these small molecules. Eight scans were
enough for these intense signals. At 213 K, due to the smaller volume of
the sample, the number of scans was between 40 and 100. The d values
were higher than those at room temperature (3–5 ms), to compensate for
the decreased mobility (weaker intensity decay) of the molecules at low
temperature. Typically, the diffusion delays were 36, 65.5, 90, 164 or
213.5 ms. In [D6]DMSO, the high viscosity of the solvent was also com-
pensated by a longer gradient length (2–3.5 ms).
Conclusions
Tris(3-ureidobenzyl)amines 1a form self-assembling dimeric
aggregates 1a·1a in CDCl3 and [D8]toluene, in which organ-
ic molecules of adequate size and shape, such as CH3CN,
CH3NO2, CH2Cl2, CH3I, CH2BrCl, CH2Br2, CHCl3 and
1
C6H6, can be accommodated. Variable temperature H and
1H,1H-ROESY NMR experiments have been used to inves-
tigate these 1a·1a·G assemblies. Thus, new resonances for
capsule and guest in 1a·1a·G have been detected upon addi-
tion of the guests to solutions of 1a·1a in CDCl3 and
[D8]toluene. PGSE diffusion measurements have been used
to further investigate the solutions in which G=CH3NO2,
CH2Cl2 and CH3I. The results show that capsule and guest
in 1a·1a·G form a kinetic unit in solution.
Thermodynamic and kinetic parameters for the encapsula-
tion of CH3NO2, CH2Cl2 and CH3I into 1a·1a in CDCl3 and
[D8]toluene have also been determined. Similar trends in
the thermodynamic parameters were found for the encapsu-
lation of each guest in both solvents, pointing to the affinity
between guest and 1a·1a being dictated mainly by the elec-
tronic, size and shape complementarity between guest and
cavity. The thermodynamic data also support the initial sit-
uation of an empty capsule in [D8]toluene. Interestingly, a
compensation between enthalpic and entropic effects has
ROESY experiment: The ROESY spectrum was measured on a 1.1”
10ꢀ2 m solution of 1a·1a in CDCl3 in the presence of a 30-fold excess of
CH2Cl2, by using a 600 MHz Bruker AVANCE spectrometer equipped
with a multinuclear inverse probe. A spin-lock pulse of 400 ms was used.
The number of scans per increment was 80, and 240 experiments were ac-
quired in the second dimension. Total experimental time was approxi-
mately 8 h.
Modelling: All molecular mechanics calculations were carried out by
using the AMBER* force field as implemented within Maestro/Macro-
Model 8.1. Standard potentials and atomic charges, as provided by the
Chem. Eur. J. 2007, 13, 1559 – 1569
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1567

A. Pastor, P. S. Pregosin et al.
AMBER* force field, were employed without modifications. AMBER*
and OPLAA force fields produced essentially the same results in related
structures. Calculations were initially performed under vacuum and then
in chloroform (GB/SA solvation model). Most complex structures were
virtually identical under both conditions. Energy minimizations were con-
ducted over 500 iterations on a Silicon Graphics Origin 2000 Computer.
Minimized structures were then subjected to conformational searches
with 5000-step Monte Carlo multiple minimum simulations. All confor-
mations within 15 kJmolꢀ1 of the computed global minimum were stored,
and the representative lowest-energy structure was analysed.
Chem. Int. Ed. 2005, 44, 1000–1002; c) J. Rebek, Jr., Angew. Chem.
2005, 117, 2104–2115; Angew. Chem. Int. Ed. 2005, 44, 2068–2078.
[4] a) N. Branda, R. Wyler, J. Rebek, Jr., Science 1994, 263, 1267–1268;
b) J. Kang, J. Rebek, Jr., Nature 1996, 382, 239–241; c) O. Mogck,
M. Pons, V. Bçhmer, W. Vogt, J. Am. Chem. Soc. 1997, 119, 5706–
5712; d) R. Meissner, X. Garcias, S. Mecozzi, J. Rebek, Jr., J. Am.
Chem. Soc. 1997, 119, 77–85; e) T. Szabo, G. Hilmersson, J. Re-
bek, Jr., J. Am. Chem. Soc. 1998, 120, 6193–6194; f) M. O. Vysotsky,
I. Thondorf, V. Bçhmer, Angew. Chem. 2000, 112, 1309–1312;
Angew. Chem. Int. Ed. 2000, 39, 1264–1267; g) F. Hof, C. Nuckolls,
S. L. Craig, T. Martín, J. Rebek, Jr., J. Am. Chem. Soc. 2000, 122,
10991–10996.
[5] a) M. Alajarín, A. Lꢂpez-Lµzaro, A. Pastor, P. D. Prince, J. W.
Steed, R. Arakawa, Chem. Commun. 2001, 169–170; b) M. Alajarín,
A. Pastor, R.-A. Orenes, J. W. Steed, J. Org. Chem. 2002, 67, 7091–
7095; c) M. Alajarín, A. Pastor, R.-A. Orenes, J. W. Steed, R. Ara-
kawa, Chem. Eur. J. 2004, 10, 1383–1397; d) M. Alajarín, A. Pastor,
R.-A. Orenes, E. Martínez-Viviente, P. S. Pregosin, Chem. Eur. J.
2006, 12, 877–886.
[6] a) D. S. Lawrence, T. Jiang, M. Levett, Chem. Rev. 1995, 95, 2229–
2260; b) L. J. Prins, D. N. Reinhoudt, P. Timmerman, Angew. Chem.
2001, 113, 2446–2492; Angew. Chem. Int. Ed. 2001, 40, 2382–2426.
[7] a) E. O. Stejskal, J. E. Tanner, J. Chem. Phys. 1965, 42, 288–292;
b) P. Stilbs, Prog. Nucl. Magn. Reson. Spectrosc. 1987, 19, 1–45;
c) W. S. Price, Annu. Rep. NMR Spectrosc. 1996, 32, 51–142; d) C. S.
Johnson, Jr., Prog. Nucl. Magn. Reson. Spectrosc. 1999, 34, 203–256.
[8] a) L. Frish, S. E. Matthews, V. Bçhmer, Y. Cohen, J. Chem. Soc.
Perkin Trans. 2 1999, 669–671; b) L. Frish, M. O. Vysotsky, S. E.
Matthews, V. Bçhmer, Y. Cohen, J. Chem. Soc. Perkin Trans. 2 2002,
88–93; c) L. Avram, Y. Cohen, Org. Lett. 2002, 4, 4365–4368; d) L.
Avram, Y. Cohen, J. Am. Chem. Soc. 2002, 124, 15148–15149; e) L.
Avram, Y. Cohen, Org. Lett. 2003, 5, 1099–1102; f) L. Frish, M. O.
Vysotsky, V. Bçhmer, Y. Cohen, Org. Biomol. Chem. 2003, 1, 2011–
2014; g) Y. Cohen, L. Avram, L. Frish, Angew. Chem. 2005, 117,
524–560; Angew. Chem. Int. Ed. 2005, 44, 520–554.
Rate constants: The rate constants have been determined with the help
of selective inversion-magnetisation transfer experiments (see the Sup-
porting Information for an extended explanation). For the analysis of the
results we have used the program CIFIT.[38] The relaxation delay was
equal to five times the T1 of the signals of interest. Typically, 17 or 22
data points were used for the fit, with the variable time delay increased
from 1 ms to 4 s. Total experimental times were approximately 2–4 h. For
the solutions of the samples in CDCl3, the NH resonance at approximate-
ly 8.3 ppm (from 1a·1a·G) was selectively inverted, and the evolution of
this peak and the NH resonance at approximately 8.0 ppm (from
1a·1a·CDCl3) was followed. The number of scans per experiment was
four or eight. In [D8]toluene, the concentration of the empty capsule
1a·1a was very low, or not detectable. Consequently, in this solvent the
resonances from the free and encapsulated guest were used for the mag-
netisation transfer experiment. Inversion was carried out on the former.
Due to the longer T1 values of the involved peaks, only four scans per ex-
periment were acquired.
The rate constants were determined at temperature intervals from 5 K.
The range of temperatures investigated depended on the dynamic behav-
iour of the system (the rate constant had to be measurable, and the
NMR signals resolved). These intervals were as follows: in CDCl3: 238–
273 K for G=CH2Cl2, 263–293 K for G=CH3I and 243–273 K for G=
CH3NO2; in [D8]toluene: 238–273 K for G=CH2Cl2, 258–293 K for G=
CH3I and 233–268 K for G=CH3NO2.
Association constants: The association constants were determined from
[9] a) S. Forsꢃn, R. A. Hoffman, J. Chem. Phys. 1963, 39, 2892–2901;
b) R. A. Hoffman, S. Forsꢃn, Prog. Nucl. Magn. Reson. Spectrosc.
1966, 1, 15–204; c) J. J. Led, H. Gesmar, J. Magn. Reson. 1982, 49,
444–463; d) A. D. Bain, D. A. Fletcher, Mol. Phys. 1998, 95, 1091–
1098.
1
the integrals of the separate H resonances for capsule and guest (see the
Supporting Information). The number of scans was eight and the relaxa-
tion delay was equal to five times the T1 of the guest. The average value
from two different determinations was taken. The same temperatures as
those used for determining the rate constants were investigated.
[10] The resonances for 1a·1a·CDCl3 in this mixture were assigned by
comparison with the spectrum of 1a·1a at 213 K in CDCl3 (in the
absence of guest).
1
ꢀ
[11] The assignment of the N H protons (see Scheme 1) to the reso-
nance at higher frequency, approximately 8 ppm, was based on the
Acknowledgements
X-ray structure of 1a·1a (d
A
E
1
This work was supported by the Swiss National Science Foundation, the
ETH Zurich, the Ministerio de Educaciꢂn y Ciencia (MEC) of Spain
(project CTQ2005-02323/BQU), and the Fundaciꢂn Sꢃneca-CARM
(project PI-1/00749/FS/01). P.S.P. thanks the Swiss National Science Foun-
dation, and the ETH Zurich for financial support. E.M.-V. thanks the
Fundaciꢂn Sꢃneca-CARM for a grant. Both E.M.-V. and A.P. thank the
MEC and the University of Murcia for their Ramón y Cajal contracts.
R.-A.O. is grateful to the Fundaciꢂn CajaMurcia for a fellowship.
[13] The 1H NMR measurements were conducted at 300 (C2D2Cl4 and
[D8]toluene), 243 (C2D2Cl4) and 213 K ([D8]toluene).
[14] a) It is difficult to determine whether these dimers are truly empty
or filled by molecules from atmospheric gases; b) we have not been
able to observe a signal corresponding to H2O encapsulated inside
1a·1a.
[15] These experiments were conducted by addition of a 20- to 100-fold
excess of guest G (0.09–0.67m) to a solution of 1a·1a in [D8]toluene
(3–6 mm).
[1] a) J. Rebek, Jr., Chem. Soc. Rev. 1996, 255–264; b) M. M. Conn, J.
Rebek, Jr., Chem. Rev. 1997, 97, 1647–1668; c) J. de Mendoza,
Chem. Eur. J. 1998, 4, 1373–1377; d) J. Rebek, Jr., Acc. Chem. Res.
1999, 32, 278–286; e) C. A. Schalley, Adv. Mater. 1999, 11, 1535–
1537; f) D. R. Turner, A. Pastor, M. Alajarín, J. W. Steed, Struct.
Bonding (Berlin) 2004, 108, 97–168.
[2] a) F. Hof, S. L. Craig, C. Nuckolls, J. Rebek, Jr., Angew. Chem. 2002,
114, 1556–1578; Angew. Chem. Int. Ed. 2002, 41, 1488–1508;
b) L. C. Palmer, J. Rebek, Jr., Org. Biomol. Chem. 2004, 2, 3051–
3059.
[16] For all the guests investigated, the maximum difference in d(1H) be-
tween the encapsulated guest in CDCl3 and [D8]toluene is 0.35 ppm
(at 213 K). When G=CHCl3 and C6H6 these resonances of G are
overlapped with resonances from 1a·1a and 1a·1a·G, and thus, are
not observable.
[17] Aromatic solvent-induced shift (ASIS). For instance, see: M.
Ahmad, L. Phillips, J. Chem. Soc. Perkin Trans. 2 1977, 1656–1661.
[18] These NOE contacts are not present in the ROESY spectrum of 1a
in [D6]DMSO in which the compound exists as a non-assembled
monomer (Supporting Information).
[3] a) F. Hof, J. Rebek, Jr., Proc. Nat. Acad. Sci. USA 2002, 99, 4775–
4777; b) A. Lützen, Angew. Chem. 2005, 117, 1022–1025; Angew.
1568
www.chemeurj.org
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 1559 – 1569

Capsules in Solution
FULL PAPER
[19] The Stokes–Einstein equation is D=(kBT)/
G
[24] a) J. Canceill, L. Lacombe, A. Collet, J. Am. Chem. Soc. 1986, 108,
4230–4232; b) J. Canceill, M. Cesario, A. Collet, J. Guilhem, L. La-
combe, B. Lozach, C. Pascard, Angew. Chem. 1989, 101, 1249–1251;
Angew. Chem. Int. Ed. Engl. 1989, 28, 1246–1248.
[25] S. Mecozzi, J. Rebek, Jr., Chem. Eur. J. 1998, 4, 1016–1022.
[26] B. C. Hamann, K. D. Shimizu, J. Rebek, Jr., Angew. Chem. 1996,
108, 1425–1427; Angew. Chem. Int. Ed. Engl. 1996, 35, 1326–1329.
[27] Tc is the compensation temperature, see: a) R. Lumry, S. Rajender,
Biopolymers 1970, 9, 1125–1227; b) K. N. Houk, A. G. Leach, S. P.
Kim, X. Zhang, Angew. Chem. 2003, 115, 5020–5046; Angew. Chem.
Int. Ed. 2003, 42, 4872–4897.
the diffusion coefficient, kB is the Boltzmann constant, T is the tem-
perature in Kelvin, h is the viscosity of the solution at the tempera-
ture T (for diluted solutions, the viscosity of the solvent can be
used), and rH is the hydrodynamic radius (the radius of a hypotheti-
cal hard sphere that diffuses with the same speed as the particle
under examination). For instance, see: J. T. Edward, J. Chem. Educ.
1970, 47, 261–270. The viscosities have been taken from C. L. Yaws,
Chemical Properties Handbook, McGraw-Hill, New York, 1999,
with the exception of the viscosity for toluene at 213 K, which is er-
roneous in that reference and has been taken from W. Fieggen, H.
Gerding, Recl. Trav. Chim. Pays-Bas 1970, 89, 236–244. It is accept-
ed that the factor 6 in the Stokes–Einstein equation is not valid for
small species that are comparable in size to the solvent molecules.
[28] a) J. D. Dunitz, Chem. Biol. 1995, 2, 709–712; b) M. S. Searle, M. S.
Westwell, D. H. Williams, J. Chem. Soc. Perkin Trans. 2 1995, 141–
151; c) E. Grunwald, C. Steel, J. Am. Chem. Soc. 1995, 117, 5687–
5692.
The limits for this condition are not clear, but in general when rsolute
rsolvent ꢃ ꢄ2, the substitution of 6 by 4, or its multiplication by an em-
pirical factor f<1 has been proposed. For instance, see: A. Gierer,
K. Wirtz, Z. Naturforsch. A. 1953, 8, 532–538; A. Spernol, K. Wirtz,
Z. Naturforsch. A. 1953, 8, 522–532; H. C. Chen, S. H. Chen, J.
Phys. Chem. 1984, 88, 5118–5121; J. T. Edward, J. Chem. Educ.
1970, 47, 261–270; M. Ue, J. Electrochem. Soc., 1994, 141, 3336–
3342. To be consistent and facilitate comparison, we have used the
Stokes–Einstein equation with the factor 6. However, for the small
solvent and guest molecules, we show in parentheses in Tables 3 and
4 the radii calculated by using the empirical equation proposed by
[29] It cannot be disregarded that the correlation could be due to the in-
terrelated errors in DH and DS, as pointed out in ref. [27b].
[30] a) R. G. Chapman, J. C. Sherman, J. Am. Chem. Soc. 1998, 120,
9818–9826; b) M. V. Rekharsky, Y. Inoue, Chem. Rev. 1998, 98,
1875–1917; c) E. L. Piatnitski, R. A. Flowers II, K. Deshayes, Chem.
Eur. J. 2000, 6, 999–1006.
[31] When the capsules 1a·1a and 1c·1c are mixed in solution, the het-
A
See ref. [5b].
[32] a) T. Szabo, B. M. OꢀLeary, J. Rebek, Jr., Angew. Chem. 1998, 110,
3606–3609; Angew. Chem. Int. Ed. 1998, 37, 3410–3413; b) X.
Wang, K. N. Houk, Org. Lett. 1999, 1, 591–594.
[33] J. Santamaría, T. Martín, G. Hilmersson, S. L. Craig, J. Rebek, Jr.,
Proc. Natl. Acad. Sci. USA 1999, 96, 8344–8347.
[34] S. L. Craig, S. Lin, J. Chen, J. Rebek, Jr., J. Am. Chem. Soc. 2002,
124, 8780–8781.
[35] a) P. S. Pregosin, E. Martínez-Viviente, P. G. A. Kumar, Dalton
Trans. 2003, 4007–4014; b) P. S. Pregosin, P. G. A. Kumar, I. Fernµn-
dez, Chem. Rev. 2005, 105, 2977–2998.
[36] H. J. V. Tyrrell, K. R. Harris, Diffusion in Liquids, Butterworths,
London, 1984.
Chen et al. f=1/[1+{0.695(rsolv/rsolute)
2.234}].
A
[20] For elongated molecules the hydrodynamic radius can be considered
to be approximately 85% of the rotational radius, according to a
suggestion from Mattison et al. in: K. W. Mattison, U. Nobbmann,
D. Dolak, Am. Biotechnol. Lab. 2001, 19, 66–67. For 1a·1a, the ro-
tational radius estimated by measuring distances between atoms in
the X-ray structure is approximately 10.5 Š. 85% of this value
would be 9.0 Š.
[21] a) The small difference between the rH values at 300 and 213 K can
be due to errors in the viscosities used in the Stokes–Einstein equa-
tion; b) no empirical equations have been suggested to modify the
Stokes–Einstein equation for small molecules at low temperatures;
c) the absence of convection effects at low temperature has been
checked by repeating the diffusion measurements with different dif-
fusion times, as explained in ref. [37]. The perfect agreement and re-
producibility between the different samples provides an additional
check on the quality of our low temperature measurements.
[37] E. Martínez-Viviente, P. S. Pregosin, Helv. Chim. Acta 2003, 86,
2364–2378.
[38] Developed by Alex D. Bain, McMaster University, Hamilton, Ontar-
io, Canada, www.chemistry.mcmaster.ca/bain/.
[22] h .
(DMSO) at 300 K=1.9075”10ꢀ3 Kgsꢀ1 mꢀ1
A
[23] The calculated structure for 1b·1b·CH2Cl2 is in good agreement
with its solid-state structure determined by X-ray diffraction (see
ref. [5b]).
Received: August 30, 2006
Published online: November 9, 2006
Chem. Eur. J. 2007, 13, 1559 – 1569
ꢁ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1569