Thermodynamics of supramolecular naphthalenediimide nanotube formation: The influence of solvents, side chains, and guest templates

Article abstract of DOI:10.1021/ja2088647

Amino-acid functionalized naphthalenediimides self-assemble into hydrogen-bonded supramolecular helical nanotubes via a noncooperative, isodesmic process; the self-assembly of ordered helical systems is usually realized through a cooperative process. This

Full text of DOI:10.1021/ja2088647

ARTICLE
pubs.acs.org/JACS
Thermodynamics of Supramolecular Naphthalenediimide Nanotube
Formation: The Influence of Solvents, Side Chains, and Guest
Templates
Nandhini Ponnuswamy,† G. Dan Pantos-,*,†,‡ Maarten M. J. Smulders,† and Jeremy K. M. Sanders*,†
†University Chemical Laboratory, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, United Kingdom
‡Department of Chemistry, University of Bath, Bath BA2 7AY, United Kingdom
S Supporting Information
b
ABSTRACT: Amino-acid functionalized naphthalenediimides
self-assemble into hydrogen-bonded supramolecular helical nano-
tubes via a noncooperative, isodesmic process; the self-assembly of
ordered helical systems is usually realized through a cooperative
process. This unexpected behavior was rationalized as a manifesta-
tion of entropyÀenthalpy compensation. Fundamental insights
into the thermodynamics governing this self-assembly were ob-
1
tained through the fitting of the isodesmic model to H NMR
spectrometry and circular dichroism spectroscopy measurements. Furthermore, we have extended the application of this
mathematical model, for the first time, to quantitatively estimate the effect of guests, solvents, and side chains on the stability of
the supramolecular nanotube; most significantly, we demonstrate that C60 acts as a template to stabilize the nanotube assembly and
thereby substantially increase the degree of polymerization.
’ INTRODUCTION
addition to the growing chain. The formation of chiral, helical
supramolecular structures is often associated with a cooperative
growth mechanism as the number of interactions between indivi-
dual monomer units increases when the first turn of the helix is
closed.4 Cooperative helical systems have been reported pre-
viously in biological polymers, with notable examples being the
nucleated formation of actin6À8 and tubulin.9 Some synthetic
examples of such systems include the supramolecular polymer-
ization of perylenediimide chromophores,10 amphiphilic hexa-
peri-hexabenzocoronenes,11,12 cyclotriveratrylene derivatives
equipped with dendrons,13,14 dendronized dipeptides15 and
oligo(p-phenylene vinylenes) equipped with ureidotriazine self-
complementary quadruple hydrogen bonding units.16,17
We describe here the thermodynamics governing the self-
assembly of amino-acid functionalized naphthalenediimide (NDI)
supramolecular helical nanotubes in CHCl3 and 1,1,2,2,-
tetrachloroethane (TCE) solutions. We have used mathematical
models for supramolecular polymerizations to show that, con-
trary to our expectation, the supramolecular polymerization of
NDI helical nanotubes occurs via a nonÀcooperative, isodesmic
mechanism. We have calculated the association constants leading
to the formation of the supramolecular nanotube and, for the first
time, have extended the application of the isodesmic model to
quantify the effect of stabilizing and destabilizing guests, solvents
and side chains on the stability of the nanotube. A surprising role
for entropyÀenthalpy compensation in nanotube formation is
also discussed.
The self-assembly of a quasi-one-dimensional supramolecular
polymer is described either via an isodesmic or a cooperative
mechanism.1À5 In an isodesmic mechanism, the addition of each
monomer to the growing chain is governed by a single equilib-
rium constant, Ke. This implies that there is a broad distribution
in the degree of polymerization, with substantial numbers of long
objects being only obtained either at high concentrations or low
temperatures, or for high association constants. In contrast, the
cooperative mechanism is characterized by an activation step,
governed by equilibrium constant, K2, which precedes chain elon-
gation with equilibrium constant Ke. If K2 , Ke, the polymeri-
zation becomes cooperative and is reminiscent of an actual phase
transition. This implies a sharp change in the degree of aggrega-
tion above which there is enhanced propensity for monomer
Previously, we have reported that amino-acid functionalized
NDIs form supramolecular nanotubes (diameter =12.4 Å) in
aprotic solvents of medium polarity such as CHCl3 and TCE: the
monomers are held together by COOH hydrogen-bonded dime-
rization between the i and i + 1 monomers and also by nonÀ
classical CH O hydrogen bonds between the i and i + 3
3 3 3
monomers.18 These nanotubes have been characterized by cir-
cular dichroism and X-ray diffraction methods (Figure 1) and were
found to act as receptors for fullerenes,19,20 condensed aromatic
systems,21 and quaternary ammonium ions.22 To date, we have
only reported a qualitative description of the binding events in
the formation of the nanotube and hostÀguest complexes: these
have been difficult systems to study due to the unknown length
Received: September 26, 2011
Published: November 18, 2011
r
2011 American Chemical Society
566
dx.doi.org/10.1021/ja2088647 J. Am. Chem. Soc. 2012, 134, 566–573
|

Journal of the American Chemical Society
ARTICLE
Figure 3. Amino-acid functionalized NDIs used in this study.
Figure 1. (A) Space-filling and (B) stick-representation highlighting
the H-bonding pattern (dashed lines) of the X-ray structure of 1a
obtained from CH2Cl2; the trityl groups have been removed for clarity.
(Color scheme: gray, C; cyan, H; red, O; blue, N; yellow, S). (C)
Absorption (dashed line) and CD (solid line) spectra of a 5.0 Â 10À4 M
solution of 1a in CHCl3.
Figure 4. (A) Effect of temperature on the chemical shift of the NDI
protons; [1a] = 0.9 Â 10À3 M in TCE. (B) Concentration-dependent
chemical shift of the NDI protons of 1a (blue dots), 1b (red dots), and
1c (black dots) at 300 K in CHCl3.
S-trityl protected cysteine functionalized NDIs 1(aÀc)
were chosen for this study due to their enhanced solubility in
chlorinated solvents compared to other amino-acid functionalized
NDIs (Figure 3).
The side chains of the NDI monomer can be arranged in the
syn or the anti conformation with respect to the naphthalene core
(Figure 2a). The consideration of synÀanti equilibrium is relevant
because the NDI nanotube can only form through the hydrogen-
bonded self-assembly of syn conformers. XÀray diffraction of
crystals of 1(a,b) grown from solvents of differing polarities and
molecular modeling studies have shown that the syn conformer is
unexpectedly preferred over the anti conformer in chlorinated
solvents, presumably due to favorable side chain interactions
(see SI).
Figure 2. Schematic representation of the equilibria between (A) anti
and syn side chain conformations of the NDI monomer and (B) possible
NDI aggregates in CHCl3 or TCE solution.
The NDI monomer might be expected to aggregate via π-
stacking and/or hydrogen bonds.27,28 In CHCl3, the formation of
π-stacked polymers is unfavorable on both electronic and steric
grounds: NDI has an electron deficient aromatic core and the
association constant for the stacking of two NDI units in CHCl3
has been reported by Iverson and Cubberley to be less than
distribution of the nanotube host, the undefined stoichiometry
of the interaction with guests, and the low chloroform solubility
of C60. However, recent developments in the analysis of self-
assembly mechanisms of supramolecular polymerizations have
allowed us to investigate the mechanism of formation of the NDI
nanotubes and quantitatively estimate the effect of hostÀguest
interactions.1À5
1 M .
À1 29 Additionally, the steric hindrance provided by the amino
acid side chains destabilizes the formation of any aggregate
formed by πÀπ interactions. Thus, the formation of supramo-
lecular polymers by π-stacking of NDI units can be safely ignored
when considering the possible equilibria of NDIs in solution.
With respect to hydrogen-bonded structures, both NDI con-
formers can self-assemble via COOH dimerization to form a
random supramolecular polymer, but only the poly-syn-NDI can
further fold to form the supramolecular nanotube (Figure 2b).
Concentration- and temperature-dependent 1H NMR experi-
ments were performed to gain insight into the self-assembly of 1a
in solution. The single NDI resonance around 8.6 ppm (from the
aromatic protons shown in red in Figure 4) in 1H NMR spectra
(500 MHz) of compound 1a dissolved in CHCl3 or TCE
is highly dependent on both concentration and temperature.
’ RESULTS AND DISCUSSION
Thermodynamics of the Formation of the Nanotube. In
order to investigate the relative stability of amino-acid functionalized
NDI supramolecular nanotubes, we considered all possible
supramolecular NDI aggregates present in solution as being
in equilibrium with each other (Figure 2). This equilibrium is
effectively a dynamic combinatorial library23À26 in which the
NDI building blocks are connected through reversible hydrogen
bonds to give a mixture of species under thermodynamic control.
567
dx.doi.org/10.1021/ja2088647 |J. Am. Chem. Soc. 2012, 134, 566–573

Journal of the American Chemical Society
ARTICLE
Figure 6. (A) Isodesmic fit of the concentration-dependent 1H NMR
chemical shift of 1a at 300 K in CHCl3. (B) Calculated distribution of
material over different nanotube lengths (number of aggregated mono-
mers, in short N-mers) at 300 K (black trace) and 273 K (red trace).
Figure 5. (A) 1H NMR spectrum of the NDI aromatic protons (shown
in red) of a 0.9 Â 10À3 M solution of 1a in TCE at different
temperatures. (B) CH O hydrogen-bonding between NDI i and
3 3 3
i + 3 in the nanotube, as observed in the crystal structure.
This resonance becomes more shielded by around 0.15 ppm on
decreasing the temperature (from 378 to 293 K, Figure 4a) and by
around 0.25 ppm on increasing the concentration (from 1.0 Â
10À6 M to 0.7 M, Figure 4b, blue dots). Significantly smaller shifts
(less than 0.03 ppm) in this proton resonance were observed in 1b,
which cannot form hydrogen bonds (Figure 4b, red dots) and in
monotopic derivative 1c which can only form a hydrogen-bonded
dimer (Figure 4b, black dots). These observations are in accor-
dance with 1a forming a higher order, hydrogen-bonded supra-
molecular structure in solution.
Lowering the sample temperature below 298 K results in
splitting of the broad singlet at 8.6 ppm into two peaks, which
further resolve into doublets at temperatures below 265 K as
shown in Figure 5a. The splitting of the NDI peak at lower
temperatures supports the idea that NDIs predominantly form a
well-defined and well-ordered hydrogen-bonded supramolecular
structure with two different environments for the NDI protons.
This is entirely consistent with the repetitive arrangement of
NDI building blocks held as a nanotube, as shown in Figure 5b. In
a random hydrogen-bonded polymer (see Figure 2), the NDI
aromatic protons should experience an effectively symmetric en-
vironment and consequently appear as a singlet in the 1H NMR
spectrum; by contrast, in the nanotube, the geometry imposed by
For further evidence of an isodesmic mechanism, the self-
association of 1a was also studied using a Dil plot31 developed
by Hunter and Anderson, which provides a more accurate
analysis for dilution experiments. In a Dil plot, the concentration
switching window cR, defined as the factorial increase in ligand
concentration required to change the bound/free receptor ratio
from 1:10 to 10:1, can be used as a macroscopic measure of
cooperativity.32 In other words, cR is a measure of the sharpness
of the boundÀfree transition. A value of log cR = 3 represents
isodesmic behavior and any deviation from this value indicates
cooperativity. Fitting of the concentration-dependent 1H NMR
shifts of 1a to the Dil plot gave the value of log cR = 3.06, which
confirmed that the selfÀassociation of 1a is dominated by the
isodesmic mechanism (see SI). Fitting of the above data to an
isodesmic model afforded an association constant, K = (2.6 (
0.5) Â 103 MÀ1 for the formation of nanotubes of 1a at 300 K
(Figure 6a) in CHCl3 (see SI for fitting).
The association constant for the formation of the nanotube
was also calculated using concentration-dependent circular di-
chroism (CD) spectroscopy. The NDI monomer has a very low
intrinsic CD signal, but upon forming the nanotube in CHCl3 the
CD signal (λmax at 381 nm) intensifies dramatically. The intensity of
the CD signal is therefore proportional to the proportion of NDI
monomers held in the nanotubular aggregate.33 An isodesmic
fit of a plot of the CD intensity vs concentration yielded the as-
sociation constant for the formation of the nanotube, in quali-
the proposed additional CÀH O hydrogen bonds desym-
3 3 3
metrizes the NDI aromatic protons by placing them in different
chemical environments, resulting in the splitting of their 1H
resonances. Lowering the temperature slows down the exchange
between the nonaggregated NDI monomers and those in a
nanotube, thus providing a window on the dynamics of nanotube
formation in solution.
1
tative agreement with that obtained by the H NMR method
described above. However, due to technical constraints, concen-
trations only up to 1.0 Â 10À3 M were accessible with this tech-
nique and as a result the fitting of the data was less precise
(K = (4.3 ( 1.2) Â 103 MÀ1 at 300 K; see SI).
The CÀH O hydrogen bonds also place the aromatic
3 3 3
proton of the i NDI unit in the carbonyl shielding region30 of
the i + 3 NDI unit (Figure 5b). Since the observed NDI chemical
shift is an average of the chemical shift of the NDI monomer,
random polymer, and the nanotube, an increase in the propor-
tion of material present as nanotubes, either at high concentra-
tions or low temperatures, results in the shielding of the NDI
aromatic proton. The very small upfield shift observed in 1c is
probably due to the formation of a hydrogen-bonded dimer.
The dependence of the chemical shift of the NDI aromatic
protons on concentration was analyzed in order to shed light on
the thermodynamics of nanotube formation. A smooth sigmoidal
curve devoid of any sharp slope changes, suggestive of an isodesmic
1
The temperature-dependent H NMR experiments using an
automated variable temperature NMR program are more accu-
rate and require less material than the concentration-dependent
1H NMR experiments.5 These experiments were carried out in
TCE (a solvent comparable to CHCl3 in terms of its hydrogen
bonding abilities), allowing experiments at higher temperatures.
1
The data obtained from the temperature-dependent H NMR
(Figure 4a) also revealed an isodesmic self-assembly mechanism
for 1a and yielded an association constant for the formation of
nanotube K = (2.9 ( 0.02) Â 103 MÀ1 at 300 K. The analysis also
provided the melting temperature (where the degree of aggrega-
tion is equal to 0.5) of the nanotube TM = 320 K, the molar
enthalpy of nanotube formation ΔH = À65 kJ molÀ1 and change
in entropy ΔS = À150 J molÀ1 KÀ1, for a solution of 0.9 Â 10À3 M.
The association constant obtained from the temperature-dependent
1
self-assembly mechanism, was observed in the plot of the H
NMR resonance for the NDI aromatic proton of 1a in CHCl3
at 300 K at different concentrations (Figure 4b, blue dots).10
568
dx.doi.org/10.1021/ja2088647 |J. Am. Chem. Soc. 2012, 134, 566–573

Journal of the American Chemical Society
ARTICLE
increases when the first turn of the helix is closed. The lack of
significant cooperativity, in spite of the presence of additional
interactions after the formation of one helical turn in the nano-
tube, is probably a manifestation of entropyÀenthalpy compensa-
tion: this is the widely documented empirical observation that the
increased enthalpic benefits of tighter binding generally results in
increased entropic costs due to more restricted relative motion.35À40
The two effects can cancel each other out because the energetic gain
from the formation of noncovalent bonds is, at room temperature,
comparable to the entropic losses that oppose them.
In the case of the NDI nanotubes, the enthalpic cooperativity
obtained by the formation of additional putative CH O hy-
3 3 3
drogen bonds between i and i + 3 NDI units and by interactions
between the solvent and the NDI units is almost completely
compensated by entropic costs due to higher ordering of the side
chains, as a result of which no significant cooperativity in free
energy is observed (see SI).
Figure 7. Schematic representation of the effect of mono- and di-
COOH units on association constant (Kdim and K1dim respectively) of
COOH hydrogen-bonded dimerization.
Having elucidated the mechanism of formation of NDI
nanotubes, we then applied the isodesmic model to explore the
effect of side chain interactions on the stability of the nanotube.
Effect of Side Chain Interactions on Nanotube Stability. As
these nanotubes are held together primarily by hydrogen bond-
ing between carboxylic acid groups, they can be disassembled by
base-induced deprotonation.41 AcidÀbase titration experiments
led us to conclude that, among the amino acid derivatives we have
studied, S-trityl protected cysteine functionalized NDIs (1a)
formed the most stable nanotube in solution, while ε-NHBoc
protected lysine (2) and isoleucine (3) functionalized NDIs
formed less stable nanotubes. Molecules 2 and 3 were therefore
chosen for studying the effect of side chains on the stability of the
nanotube. Chirooptical studies on equimolar concentrations of
1a, 2, and 3 in chloroform gave a relatively higher CD signal
intensity for 1a, also suggesting that the proportion of monomer
held as a nanotube was relatively higher in 1a. These experiments
gave a qualitative scale of the relative strengths of the nanotubes
derived from different NDIs but failed to give a quantitative
measurement or comparison of association constants for the
formation of these nanotubes.
analysis is very close to that obtained from concentration-
dependent analysis, and is experimentally much easier to obtain;
this establishes the utility of temperature-dependent 1H NMR to
calculate association constants for this type of hydrogen-bonded
system.5 The enthalpy and entropy contributions to the Gibbs
free energy changes for aggregation show that formation of the
nanotube is enthalpy driven and entropically disfavored. This is
expected, given that nanotube formation depends on hydrogen
bonds between NDI monomers that greatly restrict conforma-
tional freedom.
The number average degree of polymerization DPN according
to the temperature-dependent isodesmic model was calculated as
5.3 at 273 K and 2.2 at 300 K for a solution of 1a of 0.9 Â 10À3 M
in TCE. A simulation of the exponential distribution7 of species
in a 0.9 Â 10À3 M solution with the above DPN values showed
that, at 273 K, nanotubes up to 9 helical turns, corresponding to
28 aggregated NDI monomers, were present in the solution,
albeit at low concentrations (Figure 6b, see SI).
In a further exploration of the growth mechanism, we com-
pared the dimerization constant of COOH hydrogen bonds in
the NDI monomer to the equilibrium constant for the formation
of the nanotube, which should be similar in an ideal isodesmic
system. To estimate the dimerization constant for the formation
of COOH hydrogen-bonded dimers, a two state monomerÀ
dimer binding model was used to fit the concentration-dependent
1H NMR data of monotopic molecule 1c. A value of (2.1 ( 0.2) Â
102 MÀ1 was obtained at 300 K, which is an order of magnitude
lower than the equilibrium constant for the formation of nano-
tube of 1a.34 Much of the large apparent difference in association
constant obtained for 1c and 1a can be accounted for by a
statistical factor of 4 that distinguishes association of a molecule
containing one (1c) vs two (1a) COOH functional groups
as illustrated in Figure 7. Considering this factor of 4 in the
dimerization constant renders the cooperativity present in
molecule 1a to be only σ = Kdimer/Knanotube = 0.48, which is
probably too small to be observed experimentally and thus the
experimental data fits closely to an isodesmic mechanism
(simulated graph in SI).
We therefore applied the above analytical approach to the
variable temperature 1H NMR data of 2 and 3 in order to calculate
their association constants. In the case of 2, the equilibrium
in solution could be complicated since the presence of a Boc
protecting group could in principle lead to the competitive
formation of polymers held by CONH hydrogen bonds; how-
ever literature precedents indicate that this process is not very
likely to occur.42
Fitting the isodesmic model to the temperature-dependent 1H
NMR chemical shift of the NDI aromatic proton of 2 and 3 for a
0.9 Â 10À3 M and 1.0 Â 10À3 M TCE solution, respectively,
resulted in similar values for association constants, K = (1.26 (
0.02) Â 103 MÀ1 and K = (1.09 ( 0.06) Â 103 MÀ1 at 300 K
(Figure 8). The higher association constant obtained for mol-
ecule 1a compared to 2 and 3 is in agreement with our earlier
qualitative observations; this may be due to (1) differences in
solvation and/or creation of a more nonpolar environment in
sterically hindered 1a, resulting in the strengthening of COOH
hydrogen-bonded dimers, (2) stabilizing van der Waals’ interac-
tions between the trityl groups of i and i + 3 NDI units in the
nanotube, (3) competition for COOH groups to form hydrogen-
bonded dimers with amide groups present in the Boc side chain
of 2, and/or (4) the presence of a β-substituent in 3, may lead to
higher steric repulsions between the side chains.
The conclusion that the growth of NDI nanotubes could be
effectively explained by an isodesmic mechanism was unexpected,
since the formation of helical structures is usually cooperative
because the number of attractive interactions between monomers
569
dx.doi.org/10.1021/ja2088647 |J. Am. Chem. Soc. 2012, 134, 566–573

Journal of the American Chemical Society
ARTICLE
Figure 10. (A) Plot of association constant (K) for the formation of
nanotube, calculated for a 1.0 Â 10À3 M solution of 1a vs the percentage
of cyclohexane in TCE. (B) Plot of the CDmax signal intensity (mdeg)
for a 1.2 Â 10À4 M solution of 1a vs the percentage of cyclohexane in
TCE at 300 K (error bars with respect to error in CD intensity on
multiple runs).
Figure 8. (A) Amino-acid functionalized NDIs used for studying the
effect of side chains on the stability of the nanotube. (B) Comparison of
change in association constant with temperature for 1a, 2, and 3.
hydrogen bonds, offsetting the loss of the trityl groups which appear
to have a stabilizing effect through solvophobic interaction.
These results emphasize the significance of the side chain
interactions in the formation of the nanotube. Nanotubes formed
from 1a containing the bulky trityl side chain are more stable
than the ones derived from 2 and 3 containing an alkyl side chain,
mainly due to solvophobic effects. However in the case of molecule
5 compared to 4, the external stabilization obtained by the nonÀ
covalent cross-linking of the i and i + 3 NDI monomer through
the hydrogen bonding of thiourea functional groups in the NDI
side chain and via πÀπ interactions between the naphthyl moieties,
results in the enhanced stabilization of the helical core.
Figure 9. (A) AminoÀacid functionalized NDIs used for crossÀlinking
the side chains. (B) Comparison of the temperature-dependent associa-
tion constants for 4 and 5.
The solvophobic interactions leading to the superior stabiliza-
tion of 1a compared to 2 and 3 led us to investigate the role of
soluteÀsolvent interactions in the formation of the nanotube.
Effect of Solvent on the Stability of the Nanotube. Solva-
tion isanimportantfactor indeterminingtheself-assemblybehavior
of a system. Hunter et al. have carried out in-depth investigations
of the effects of solvent competition on intermolecular hydrogen
bonding interactions between neutral molecules in a range of
solvents.46 Their results clearly demonstrate that in noncompe-
titive solvents such as cyclohexane and carbon tetrachloride,
hydrogen bonds can be effectively stronger by a factor of 100
compared to partially competing solvents such as chloroform.
In order to increase the effective strength of the hydrogen
bonds in the NDI nanotube, cyclohexane was added to a solution
Kuroda et al. have recently shown that covalent cross-linking
of a single side chain of a helical peptide can stabilize the entire
helical structure.43 Based on similar cross-linking principles, we
envisaged that a thiourea functional group on the NDI side chain
could result in the additional stabilization of the nanotube via
additional hydrogen-bonded cross-linking of i and i + 3 NDI
monomers. For this purpose, molecule 4, which is N-desym-
metrized with trityl protected cysteine on one side and ε-NHBoc
protected lysine on the other, was synthesized using a reported
literature protocol.44,45 The Boc protecting group in molecule 4
was removed under mild acidic conditions that do not affect
the trityl group. The resultant amino group was then coupled
with 1-naphthyl isothiocyanate to give the desired product 5
(Figure 9a).
1
of 1a in TCE, using CD and variable temperature H NMR to
monitor the effect (see SI). Isodesmic fitting of temperature-
dependent chemical shift of the NDI aromatic proton yielded the
association constant for the formation of nanotube at different
ratios cyclohexane:TCE, which when plotted against the per-
centage of cyclohexane in TCE, showed an exponential increase
in association constant with increasing amount of cyclohexane
(Figure 10a). The association constant of a 1.1 Â 10À3 M solution
of 1a in 6:4 cyclohexane:TCE mixture at 300 K was found to be
(59.16 ( 0.02) Â 103 MÀ1 which is 20 times higher than in pure
Chirooptical studies of molecules 4 and 5 in TCE show
that both have a characteristic nanotube CD fingerprint (see SI).
1
Comparison of the H NMR properties of 5 with a model
compound butylÀnaphthyl thiourea in TCE at 300 K shows that
the thiourea is hydrogen-bonded and the naphthyl protons are
shielded, suggestive of a πÀπ interaction (see SI).
1
The isodesmic fit of the variable temperature H NMR data
of a 1.0 Â 10À3 M, TCE solution of molecule 5 resulted in K =
(14.07 ( 0.02) Â 103 MÀ1 at 280 K (Figure 9b). Comparison of
the association constants for molecules 4 (K = (4.32 ( 0.04) Â
103 MÀ1) and 5 at 280 K shows that the nanotubes formed by
molecule 5 are indeed stronger by a factor of 6, in agreement with
our expectations. However, comparing 5 with 1a we found that
the two nanotubes are characterized by similar association constants.
This can be explained if the introduction of the thiourea moiety has a
stabilizing effect on the nanotube formation due to intermolecular
1
TCE (DPN = 8.5 vs 2.2, respectively). H NMR experiments
at higher concentrations of cyclohexane in TCE could not be
performed due to the low solubility of 1a.
The effect of cyclohexane on the strength of the nanotube was
also studied using CD and UVÀvis spectroscopies. The intensity
of the CD signal on addition of cyclohexane to a 1.2 Â 10À4 M
solution of 1a in TCE, keeping the concentration of 1a constant,
initially increases, reaching a maximum at approximately 65%
570
dx.doi.org/10.1021/ja2088647 |J. Am. Chem. Soc. 2012, 134, 566–573

Journal of the American Chemical Society
ARTICLE
Figure 11. (A) 1H NMR spectrum of the NDI aromatic protons of a
0.9 Â 10À3 M solution of 1a in TCE in the presence of C60 at different
temperatures. (B) Top view of a space filling model of the nanotube, as
found in the molecular modeling studies.47
Figure 12. (A) Isodesmic fit of the temperature-dependent shifts of
1a + C60. (B) Distribution of percentage of material over different nano-
tube lengths of 1a (black trace) and 1a + C60 at 273 K (red trace).
approximately 7 times higher than that obtained for the forma-
tion of “empty” 1a nanotubes, signifying a positive templation and
stabilization of the nanotube by C60. The analysis also provided
the melting temperature of the “C60 filled” nanotube TM = 338 K
(19 °C higher than “empty” 1a nanotubes), the molar enthalpy of
“C60-filled” nanotube formation ΔH = À87 kJ molÀ1 and change
in entropy ΔS = À205 J molÀ1 KÀ1. The DPN value at 273 K for a
0.9 Â 10À3 M 1a solution in TCE was calculated as 5.3 for
“empty” 1a nanotube and 15.7 for “C60-filled” 1a nanotube. A
simulation of the exponential distribution of species (vide supra)
in a 0.9 Â 10À3 M solution with the above DPN values shows the
increase in the length of the nanotube on templation by C60
(Figure 12b). C60 clearly stabilizes the well-ordered environment
shown in Figure 5b due to the good match between the solvo-
phobic surface of guest and the inner walls of the host nanotube.
Comparison of the enthalpy and entropy contributions suggests
that the increase in enthalpy for the formation of “C60-filled”
nanotube results in the formation of more strongly bound nano-
tubes but at the cost of increased restriction of relative motion as
reflected in larger entropic costs, leading to partial entropyÀ
enthalpy compensation (see SI).
cyclohexane in TCE, and then decreases (Figure 10b). The addition
of cyclohexane also leads to very slight broadening and blue-shift
(∼2 nm) of the absorption spectrum (see SI). These observa-
1
tions were paralleled in H NMR experiments, leading to the
conclusion that the effective hydrogen bond strength is increas-
ing with the concentration of cyclohexane in TCE up to 65%.
Above this concentration the change in soluteÀsolvent interac-
tions leads to the destabilization of the nanotube and/or forma-
tion of other hydrogen-bonded structures with reduced interaction
between the chromophores, causing a decrease in the CD intensity33
(Figure 10b) and broadening of the 1H NMR signals (see SI).
These solvent-dependent experiments thus highlight the im-
portance of soluteÀsolvent interactions, in addition to COOH
hydrogen bonds in the formation of nanotubes. Addition of a
noncompeting hydrogen-bonding solvent, such as cyclohexane,
increases the effective strength of hydrogen bonds as predicted
but at the cost of reduced soluteÀsolvent interactions. This
experiment led us to further explore the effect of guests en-
capsulated within the nanotube cavity primarily by solvophobic
interactions, on the stability of the nanotube.
1H NMR analysis of “C60-filled” nanotubes of 2 and 3 in TCE
also showed positive templation of the nanotube by C60. How-
ever, the association constant at 300 K for the formation of
“C60-filled” nanotubes for 2 and 3 is only 2 times higher than
for “empty” nanotubes compared to the 7 times stabilization
achieved in the case of 1a (see SI). This may be due to enhanced
stabilization of the nanotube via van der Waals interactions
between the trityl groups of i and i + 3 NDI units in 1a, as a
consequence of formation of tighter nanotubes.
Template Effect of Guests. Our group has previously re-
ported that nanotubes can readily encapsulate C60 in chloroform,
probably via a solvophobic process, increasing the solubility of
C60 in this solvent by 16-fold.2 In our initial report, we were
however unsuccessful in the measurement of association con-
stants for the formation of the nanotube and for the formation of
hostÀguest complexes due to the undefined stoichiometry of the
interaction and problems arising from the low solubility of C60.
Our new approach, using isodesmic fitting of the temperature-
1
dependent H NMR chemical shift of NDI aromatic proton in
NDI nanotubes can also act as size selective receptors for
ammonium ions in chloroform solution.21 Furthermore, in a com-
petition experiment with C60 and ammonium ions, this dynamic
system showed a tendency to form mixed complexes in which
both guests are present simultaneously in the nanotube cavity,
apparently driven by the interaction between ion pair and C60.22
the presence and absence of guests to calculate thermodynamic
parameters provides a tool for better understanding of the formation
of these hostÀguest complexes.
1
The H NMR spectrum of 1a in TCE remains unchanged
upon C60 complexation, indicating that the nanotubular structure
is preserved (CD experiments confirm this observation, see SI).
The loss of symmetry of the NDI resonance (see Figure 4a) as a
consequence of slow exchange occurs at 303 K for “C60 filled”
nanotube, 20 °C higher than the “empty” nanotubes in an equally
concentrated solution. Presumably the increased stabilization of
the nanotube arises from a slowing down of the nanotube dis-
sociation as a result of C60ÀNDI interaction (Figure 11).
The data obtained from the temperature-dependent 1H NMR
of 1a in the presence of C60 could successfully be fitted to an
isodesmic model, resulting in K = (20.9 ( 0.5) Â 103 MÀ1 for the
formation of nanotubes at 300 K (Figure 12a). This value is
1
To further understand this system, we carried out a H NMR
investigation of 1a in the presence and absence of ammonium
ions in a manner similar to the C60 experiments. Isodesmic fit of
the temperature-dependent data of 1a in the presence of tetra-
butylammonium chloride (N+Bu4ClÀ) yielded K = (1.37 (
0.03) Â 103 MÀ1 for the formation of nanotubes at 300 K. The
analysis also provided the melting temperature of the “ion-pair-
filled” nanotube TM = 320 K, the molar enthalpy of nanotube
formation ΔH = À45 kJ molÀ1, and change in entropy ΔS =
À89 J molÀ1 KÀ1. The slight reduction of association constant com-
pared to “empty” nanotubes of 1a may be due to the destabilization
571
dx.doi.org/10.1021/ja2088647 |J. Am. Chem. Soc. 2012, 134, 566–573

Journal of the American Chemical Society
ARTICLE
nanotube in solution from 5.3 to 15.7 at 273 K. This corresponds
to an increase in the effective melting temperature (TM) by 20 °C.
The investigation of the weak CH O nonclassical hydrogen-
3 3 3
bonds and side chain interactions on the stability of the nanotube
gave insights into the limitation of this nanotubular system. Due
to the close match between between entropic loss and enthalpic
gain, a substantial improvement in ΔG is difficult to achieve. In
conclusion, this study illustrates the delicate nature of the NDI
nanotube by highlighting the fine balance between entropy and
enthalpy that governs its self-assembly process.
’ ASSOCIATED CONTENT
S
Supporting Information. Experimental procedures; cha-
b
racterization of 4 and 5, including their 1H and 13C NMR spectra;
crystallographic data; fitting of isodesmic model and calculation
of thermodynamic parameters. This material is available free of
charge via the Internet at http://pubs.acs.org.
Figure 13. (ÀΔH) vs (ÀTΔS) at 300 K for all the NDI molecules used
in this study.
of the hydrogen bonds by the ion pairs, or to the formation of
competing non-nanotubular complexes, resulting in the forma-
tion of shorter nanotubes which results in the observed increase
in entropy.
’ AUTHOR INFORMATION
Corresponding Author
g.d.pantos@bath.ac.uk; jkms@cam.ac.uk
These results represent a successful extension of the isodesmic
model, for the first time, to supramolecular polymerizations tem-
plated by the formation of hostÀguest complexes. The thermo-
dynamic parameters of the hostÀguest complexes obtained from
the fitting were in agreement with results obtained from previously
published experiments and enabled a quantitative understanding
of their stability. Stabilizing guests such as C60 were found to shift
the equilibrium of the NDI toward the formation of nanotubes
(Figure 2) through favorable solvophobic interactions while
destabilizing guests such as N+Bu4ClÀ resulted in the increase
of nonaggregated monomers.
’ ACKNOWLEDGMENT
We thank Dr. J. E. Davies for collecting X-ray data, Professor
C. A. Hunter, and F. B. L. Cougnon for helpful discussions. We
thank EPSRC (J.K.M.S.), Gates Cambridge Trust and St. John’s
College (N.P.), The Netherlands Organization for Scientific
ResearchÀNWO (M.M.J.S.), and Pembroke College, University
of Cambridge, and University of Bath (G.D.P.) for funding.
Summary of Thermodynamic Observations. All the results
described above can be summarized graphically in a plot of ÀΔH
vs ÀTΔS as shown in Figure 13. A linear correlation is obtained
between ÀΔH and ÀTΔS, implying that the increase in enthalpy
provided by either side chain or hostÀguest interactions is
compensated to a large extent by the increase in entropic penalties
due to the formation of tighter nanotubes. As a consequence of
this entropyÀenthalpy compensation, a substantial increase in
ΔG is not observed. This close match between enthalpy and
entropy shows the delicate nature of this supramolecular nano-
tube scaffold.
’ REFERENCES
(1) de Greef, T. F. A.; Smulders, M. M. J.; Wolffs, M.; Schenning, A.
P. H. J.; Sijbesma, R. P.; Meijer, E. W. Chem. Rev. 2009, 109, 5687–5754.
(2) Chen, Z.; Lohr, A.; SahaÀM€oller, C. R.; W€urthner, F. Chem. Soc.
Rev. 2009, 38, 564–584.
(3) Martin, R. B. Chem. Rev. 1996, 96, 3043–3064.
(4) Zhao, D.; Moore, J. S. Org. Biomol. Chem. 2003, 1, 3471–3491.
(5) Smulders, M. M. J.; Nieuwenhuizen, M. M. L.; de Greef, T. F. A.;
van der Schoot, P.; Schenning, A. P. H. J.; Meijer, E. W. Chem.—Eur. J.
2010, 16, 362–367.
(6) Oosawa, F.; Kasai, M. J. Mol. Biol. 1962, 4, 10–21.
(7) Sept, D.; McCammon, J. A. Biophys. J. 2001, 81, 667–674.
(8) Wegner, A.; Engel, J. Biophys. Chem. 1975, 3, 215–225.
(9) Flyvbjerg, H.; Jobs, E.; Leibler, S. Proc. Natl. Acad. Sci. U.S.A.
1996, 93, 5975–5979.
’ CONCLUSIONS
By treating the hydrogen-bonded equilibria of the NDI
nanotube as a dynamic combinatorial system, we have elucidated
its thermodynamics of formation and have shown for the first
time how its hostÀguest interactions can be quantified. We have
observed that the formation of the nanotube is best described by
an unexpected isodesmic mechanism, rationalized as a conse-
quence of entropyÀenthalpy compensation. To the best of our
knowledge, the NDI nanotube represents the first synthetic
supramolecular polymer where this phenomenon has been ob-
served and analyzed.
The analysis also revealed a significant templation effect by
C60 on the formation of the nanotube. Encapsulation of C60 led
to 7-fold increase in the effective association constant for the
formation of the nanotube, resulting in the increase of DPN of
(10) Kaiser, T. E.; Stepanenko, V.; W€urthner, F. J. Am. Chem. Soc.
2009, 131, 6719–6732.
(11) Hill, J. P.; Jin, W.; Kosaka, A.; Fukushima, T.; Ichihara, H.;
Shimomura, T.; Ito, K.; Hashizume, T.; Ishii, N.; Aida, T. Science 2004,
304, 1481–1483.
(12) Jin, W.; Yamamoto, Y.; Fukushima, T.; Ishii, N.; Kim, J.; Kato,
K.; Takata, M.; Aida, T. J. Am. Chem. Soc. 2008, 130, 9434–9440.
(13) Peterca,M.;Percec,V.;Imam,M.R.;Leowanawat,P.;Morimitsu,K.;
Heiney, P. A. J. Am. Chem. Soc. 2008, 130, 14840–14852.
(14) Percec, V.; Imam, M. R.; Peterca, M.; Wilson, D. A.; Heiney,
P. A. J. Am. Chem. Soc. 2009, 131, 1294–1304.
(15) Percec, V.; Dulcey, A. E.; Balagurusamy, V. S. K.; Miura, Y.;
Smidrkal, J.; Peterca, M.; Nummelin, S.; Edlund, U.; Hudson, S. D.;
Heiney, P. A.; Duan, H.; Magonov, S. N.; Vinogradov, S. A. Nature 2004,
430, 764–768.
572
dx.doi.org/10.1021/ja2088647 |J. Am. Chem. Soc. 2012, 134, 566–573

Journal of the American Chemical Society
ARTICLE
(16) Jonkheijm, P.; van der Schoot, P.; Schenning, A. P. H. J.; Meijer,
E. W. Science 2006, 313, 80–83.
(17) Beljonne, D.; Hennebicq, E.; Daniel, C.; Herz, L. M.; Silva, C.;
Scholes, G. D.; Hoeben, F. J. M.; Jonkheijm, P.; Schenning, A. P. H. J.;
Meskers, S. C. J.; Phillips, R. T.; Friends, R. H.; Meijer, E. W. J. Phys.
Chem. B 2005, 109, 10594–10604.
(18) Pantos-, G. D.; Pengo, P.; Sanders, J. K. M. Angew. Chem., Int. Ed.
2007, 46, 194–197.
(19) Pantos-, G. D.; Wietor, J.-L.; Sanders, J. K. M. Angew. Chem., Int.
Ed. 2007, 46, 2238–2240.
(20) Wietor, J.-L.; Pantos-, G. D.; Sanders, J. K. M. Angew. Chem., Int.
Ed. 2008, 47, 2689–2692.
(21) Tamanini, E.; Ponnuswamy, N.; Pantos-, G. D.; Sanders, J. K. M.
Faraday Discuss. 2010, 145, 205–218.
(22) Tamanini, E.; Pantos-, G. D.; Sanders, J. K. M. Chem.—Eur. J.
2010, 16, 81–84.
(23) Lehn, J. M. Chem. Soc. Rev. 2007, 36, 151–160.
(24) Corbett, P. T.; Leclaire, L.; Vial, L.; West, K. R.; Weitor, JÀL.;
Sanders, J. K. M.; Otto, S. Chem. Rev. 2006, 106, 3652–3711.
(25) Herrmann, A. Org. Biomol. Chem. 2009, 7, 3195–3204.
(26) Ladame, S. Org. Biomol. Chem. 2008, 6, 219–226.
(27) Bhosale, S. V.; Jani, C. H.; Langford, S. J. Chem. Soc. Rev. 2008, 37,
331–342.
(28) Sakai, N.; Mareda, J.; Matile, S. Chem. Commun. 2010, 46,
4225–4237.
(29) Cubberley, M. S.; Iverson, B. L. J. Am. Chem. Soc. 2001, 123,
7560–7563.
(30) McConnell, H. M. J. Chem. Phys. 1957, 27, 226–229.
(31) Hunter, C. A.; Anderson, H. L. Angew. Chem., Int. Ed. 2009, 48,
7488–7499.
(32) Hunter and Anderson have also shown that this analysis for
the ligandÀreceptor interactions can be extended to the study of self-
assembling supramolecular polymers.
(33) Bullheller, B. M.; Pantos-, G. D.; Sanders, J. K. M.; Hirst, J. D.
Phys. Chem. Chem. Phys. 2009, 11, 6060–6065.
(34) A value of K = (4.3 ( 0.3) Â 102 MÀ1 was obtained at 300 K for
reference compound phenylacetic acid (see SI).
(35) Calderone, C. T.; Williams, D. H. J. Am. Chem. Soc. 2001, 123,
6262–6267.
(36) Dunitz, J. D. Chem. Biol. 1995, 2, 709–712.
(37) Krishnamurthy, V. M.; Bohall, B. R.; Semetey, V.; Whitesides,
G. M. J. Am. Chem. Soc. 2006, 128, 5802–5812.
(38) Hunter, C. A.; Tomas, S. Chem. Biol. 2003, 10, 1023–1032.
(39) Williams, D. H.; Stephens, E.; O’Brien, D. P.; Zhou, M. Angew.
Chem., Int. Ed. 2004, 43, 6596–6616.
(40) Williams, D. H.; O’Brien, D. P.; Bardsley, B. J. Am. Chem. Soc.
2001, 123, 737–738.
(41) Stefankiewicz, A. R.; Tamanini, E.; Pantos-, G. D.; Sanders,
J. K. M. Angew. Chem. Int. Ed 2011, 50, 5725–5728.
(42) Steiner, T Angew. Chem., Int. Ed. 2002, 41, 48–76.
(43) Ousaka, N.; Sato, T.; Kuroda, R. J. Am. Chem. Soc. 2009, 131,
3820–3821.
(44) Pengo, P.; Pantos-, G. D.; Otto, S.; Sanders, J. K. M. J. Org. Chem.
2006, 71, 7063–7066.
(45) Tambara, K.; Ponnuswamy, N.; Hennrich, G.; Pantos-, G. D.
J. Org. Chem. 2011, 31 (9), 272–278.
(46) Cook, J. L.; Hunter, C. A.; Low, C. M. R.; Perez-Velasco, A;
Vinter, J. G. Angew. Chem., Int. Ed. 2007, 46, 3706–3709.
(47) AM1 semiempirical level using GAMESS package; the starting
geometries for the nanotube and fullerenes were obtained from X-ray
data for the free nanotube and the free fullerene and convergence
criterion was 0.01 kcal ÅÀ1 molÀ1
.
573
dx.doi.org/10.1021/ja2088647 |J. Am. Chem. Soc. 2012, 134, 566–573