Supramolecular Naphthalenediimide Nanotubes

  • Nandhini Ponnuswamy
  • Artur R. Stefankiewicz
  • Jeremy K. M. Sanders
  • G. Dan Pantoş
Part of the Topics in Current Chemistry book series (TOPCURRCHEM, volume 322)


Amino acid functionalized naphthalenediimides (NDIs) when dissolved in chloroform form a dynamic combinatorial library (DCL) in which the NDI building blocks are connected through reversible hydrogen bonds forming a versatile new supramolecular assembly in solution with intriguing host–guest properties. In chlorinated solvents the NDIs form supramolecular nanotubes which complex C60, ion-pairs, and extended aromatic molecules. In the presence of C70 a new hexameric receptor is formed at the expense of the nanotube; the equilibrium nanotube – hexameric receptor can be influenced by acid–base reactions. Achiral NDIs are incorporated in nanotubes formed by either dichiral or monochiral NDIs experiencing the “sergeants-and-soldiers” effect.


Circular dichroism Fullerenes Host–guest Ion-pairs Sergeants-and-soldiers Supramolecular chemistry Supramolecular polymers 

1 Introduction

Dynamic combinatorial chemistry is defined as combinatorial chemistry under thermodynamic control; that is, in a dynamic combinatorial library (DCL), all constituents are in equilibrium with each other [1, 2, 3, 4, 5, 6]. This requires the interconversion of library members into one another through a reversible chemical process, which can involve covalent bonds or non-covalent interactions such as hydrogen bonds. The composition of the library is determined by minimising the free energy of the whole system, which often is dominated by the thermodynamic stability of each of the library members under the particular conditions of the experiment. If a particular library member can be stabilised, either by binding to an external template, mutual interactions with other library members or due to a change in the DCL conditions, its free energy is lowered and consequently, in general, the equilibrium shifts towards its formation.

Hydrogen bonds between neutral partners in solution typically have energies between 0 kJ/mol and 20 kJ/mol [7] and have a preference for a linear arrangement of the three atoms involved (X−H···A angle ~180°). Due to its labile nature, equilibrium is often reached rapidly. Reinhoudt, Timmerman and co-workers were the first to describe a DCL based on hydrogen-bonded assemblies: using combinations of donor and acceptor hydrogen-bonded motifs, they built complex superstructures held together by an impressive number of hydrogen bonds [8, 9]. Rebek and co-workers have also utilised multiple hydrogen-bonding interactions to generate dynamic libraries composed of closed and spherical capsules [10, 11].

Amino acid functionalized naphthalenediimides (NDIs) when dissolved in chloroform form a DCL in which the NDI building blocks are connected through reversible hydrogen bonds forming a versatile new supramolecular assembly in solution with intriguing host–guest properties. These studies were prompted by the observation of an unusual arrangement of NDIs in a crystal structure and this chapter summarises the results and insights obtained to date.

2 Synthesis

Microwave dielectric heating is a mild and efficient method for the one-pot and stepwise synthesis of symmetrical and N-desymmetrised NDI derivatives of amines and α-amino acids. For the synthesis of symmetrical NDI derivatives, the reaction is carried out at 140 °C for 5 min in a dedicated microwave reactor in a pressure-resistant reaction vessel. Using this method, the products are obtained in high yield and purity after a simple aqueous workup; the acid-labile protecting groups (trityl, benzyl, tert-butyl, Boc and Pmc) are stable under the mild reaction conditions and undesired self-condensation side-products of amino acid esters, such as dipeptides, diketopiperazines and higher oligomers, are not observed. In the case of unprotected tyrosine and serine, the reaction is completely selective for the formation of the symmetrical imide without any trace of ester formation [12].

The method described above and outlined in Scheme 1 is particularly suitable for the synthesis of symmetrically substituted NDIs but originally it was of limited value for the synthesis of naphthalenemonoimide (NMI) and N-desymmetrised NDI derivatives. This is due to the difficulty of selective imide formation in a cross-conjugated dianhydride containing two equivalent electrophilic sites such as 1,4,5,8-naphthalenetetracarboxylic dianhydride (NDA). For all the aliphatic amines and amino acids tested, carrying out the reaction for 5 min at 140 °C only led to the formation of a 1:2:1 statistical mixture of dianhydride:monoimide:diimide (Table 1).
Scheme 1

Synthesis of amino acid derived symmetrical NDIs

The selective and efficient synthesis of NMI and stepwise synthesis of N-desymmetrised NDI derivatives was achieved through the development of a stepwise synthetic procedure which involves heating the reaction mixture at a lower temperature (40 °C or 75 °C) for 5 min prior to the final heating stage at 140 °C (Scheme 2). Molecular modelling at ab initio level (direct SCF, 6-31G**) provided a reasonable explanation for the selectivity observed in favour of NMI over a statistical mixture on lowering the temperature. Lowering the temperature allows for the selective nucleophilic attack of the amino group on one of the anhydrides of NDA to generate NMI (and its open form precursors). The subsequent increase in temperature provides enough energy for the second nucleophilic attack on the NMI anhydride and the final dehydration to generate NDI.
Scheme 2

General protocol for the microwave synthesis of NMIs and desymmetrised NDIs

For alkyl amines, a direct correlation between the steric bulk at the α-carbon and the yield of the reaction was found: amines attached to a secondary carbon gave higher yields than amines connected to a tertiary carbon, while amines connected to a quaternary carbon led only to the formation of an amide-carboxylic acid intermediate, rather than the corresponding imide. In the case of amino acids whose α-carbons are tertiary, a lower temperature was surprisingly required for high NMI selectivity in the first step (40 °C instead of 75 °C). This was explained by the presence of the COOR group, which assists in the collapse of the tetrahedral intermediate precursor to the imide formation. The amino acid derived NMIs were obtained as a mixture of open and closed forms due to the addition of triethylamine in the reaction. At high temperatures this promotes the formation of hydroxide ions, which causes ring opening of the unreacted anhydride moieties leading to the formation of open by-products. This process however is not detrimental to the synthesis of N-desymmetrised NDIs (Tables 2 and 3).
Table 3

Some N-desymmetrised NDIs derived from condensing the achiral amino acids shown with the NMI derived from S-trityl cysteine [14]


Amino acid


Yield (%)


Open image in new window




Open image in new window




Open image in new window



A higher degree of selectivity in favour of NMI was obtained when the amino acid derivatives contained aromatic side chains: trityl (100% selectivity), benzyl (70–90% selectivity) and alkyl (<60% selectivity). Also the amino acid esters gave higher selectivity towards NMI than the corresponding amino acids. This trend was rationalised by considering the solubility of the amino acid derivative and its reactivity towards NDA in DMF at room temperature. The NDA is insoluble in DMF at room temperature, but is rapidly solubilized by an amino acid containing aromatic side chains, which is itself soluble in DMF. It was proposed that the dissolution of NDA in DMF is probably due to π–π interactions between its extended aromatic core and the amino acid aromatic side chains. This enhanced solubility of NDA in DMF leads to a 1:1 mixture of NDA and the amino acid in solution and hence promotes the selective formation of the NMI. In the cases where sonication and heating were required in order to dissolve the reagents completely, a strong preference for the formation of the NDI was observed. The N-desymmetrised synthesis was primarily used to synthesise monochiral NDIs for their use in “sergeant-and-soldiers” experiments (see below).

3 Solid State Characterisation of α-Amino Acid Functionalised Naphthalenediimides

Single crystals of NDIs 1, 38 suitable for X-ray analysis were obtained by slow evaporation or vapour diffusion of solvents ranging from low polarity, such as benzene and dichloromethane, to the polar acetonitrile, acetone and dimethyl sulfoxide. The molecular structures of 1, 38 as well as crystallisation conditions are summarised in Table 4.
Table 4

X-ray structures and crystallisation conditions of molecules 1, 38

X-ray structure

Amino acid



Nanotube (Yes/No)

Open image in new window





Open image in new window





Open image in new window





Open image in new window









Open image in new window





Open image in new window





Open image in new window





NDI 1 (entry 1 in Table 4) stands out by being the only derivative of this set that forms a supramolecular nanotube in the solid state. This compound crystallises from CH2Cl2 in the trigonal P31 space group in which the amino acid side chains adopt a syn geometry with respect to the NDI plane (Fig. 1).
Fig. 1

Side and top views of the crystal packing of l

The syn geometry of 1 allows three S-trityl groups of three different molecules to interdigitate and the carboxylic groups of 1 to dimerize through two strong intermolecular hydrogen bonds (O1′H···O2″, 2.61 Å, 161° and O1″H···O2′, 2.63 Å, 161°). Closer inspection of the crystal structure reveals that 1 assembles in a hydrogen-bonded nanotubular supramolecular structure, in which NDI cores i and i + 3 are coplanar, forming the walls of the nanotube (Fig. 2a). This structure appears to be reinforced by two weak C−H···O hydrogen-bond interactions (C10H···O2, 3.16 Å, 142° and C4H···O4, 3.15 Å, 143°) per unit, formed between NDIs i and i + 3 (Fig. 2b).
Fig. 2

(a) Top view of the molecule 1 indicating H-bonding interactions observed in the self-assembled nanotube. The arrow indicates the direction of the nanotube assembly. (b) Schematic representation of the proposed non-classical H-bonding interactions between the NDIs units

In this arrangement, the angle between the central N1–N2 axis of the NDI core and the central axis of the nanotube is 60°, whereas the distance between two sequential aromatic cores is on average 4.8 Å. The nanotube has an average inner diameter of 12.4 Å and it contains diffuse electron density attributed to disordered water molecules.

The importance of the amino acid side chain in the solid state packing can be readily observed by comparing the crystal structures of 1 and 4 (Fig. 3). Both derivatives crystallise from CH2Cl2, but unlike 1, compound 4 crystallises in the anti geometry in the orthorhombic C222 space group. The intermolecular contacts between molecules of 4 are similar to those present in the crystal lattice of 1: the dimerization between the carboxylic groups through two strong intermolecular hydrogen bonds (O2″H···O1″ distance: 2.63 Å) is also supported by two weak C−H···O hydrogen-bond interactions (C4H···O4, 3.17 Å) per unit (Fig. 4).
Fig. 3

Side and top views of the hydrogen-bonded nanotubular structure formed by 1. Most of the hydrogen atoms, water molecules and CH2Cl2 molecules have been removed for clarity [15]

Despite having the same H-bonding interactions, the crystal packing view revealed significant differences between 1 (Fig. 1) and 4. NDI 4 forms a helical polymeric structure via carboxylic acid dimerization (Fig. 5a); however, in contrast to 1, no porous architecture was observed (Fig. 5c). The difference in solid state packing behaviour between NDIs 1 and 4 as well as the opposite arrangement of the side chains (syn and anti, respectively) is directly related to the intermolecular interactions between the amino acid side chains. In the case of 4, the flexible and relatively short alkyl chains of l-isoleucine form a tightly packed and compact structure in the solid state.
Fig. 4

View of solid state structure of 4 highlighting two types of H-bonding interactions (a) COOH dimerisation and (b) weak CH···O interactions

Fig. 5

(a) Side view of the helical hydrogen bonded structure of 4 in the solid state. Top views of the crystal packing of 4 capped sticks (b) and space filling representation (c)

NDI 5 is an eloquent example of the delicate balance between the intermolecular forces that determine crystal packing. Analysis of the X-ray structures of 5 crystallised from three different solvents – acetone, acetonitrile and benzene – reveals that the solid state arrangement of this l-phenylalanine methyl ester derivative of NDI is governed by π–π interactions and complemented by a set of CH···O hydrogen bonds (Fig. 6). In this arrangement, irrespective of the solvent of crystallisation, the π–π distance between the phenyl groups and the NDI core is in the range 3.38–3.44 Å, while the 4-CH···OCH3 hydrogen bond distance is 3.24–3.30 Å.
Fig. 6

Side view of the two molecules of 5 indicating π–π interaction between aromatic units of the adjacent NDIs and the CH···OCH3 hydrogen bonds from acetonitrile (similar interactions were observed in the crystals of 5 obtained from acetone and benzene; see the text for details)

The crystallisation solvent influences the packing of these NDI stacks as can be seen in Fig. 6. In acetonitrile, separate stacks of NDI 5 are interconnected by weak C−H···O hydrogen-bond interactions (CH···O, 2.72–2.83 Å, and C···O, 3.16–3.28 Å, Fig. 7a). Slightly different behaviour was observed in the crystal lattice of 5 crystallised from benzene and acetone; these two solvents yielded similar packing arrangements. The crystal packing is less dense than that obtained from acetonitrile, and consists of alternating NDI stacks with solvent molecules trapped in between. Benzene does not disturb π–π contacts and is not directly involved in these intermolecular interactions (Fig. 7b), suggesting that in the present case the solvent has limited influence on the final molecular packing. This observation was confirmed when crystals obtained from acetone had a similar packing structure.
Fig. 7

The crystal packing of 5 crystallised from acetonitrile (a), acetone (b) and benzene (c) showing solvent molecules intercalated between two stacks of NDIs. The hydrogen atoms have been removed for clarity; the acetone molecules in (b) are disordered over two sites

NDI derivatives 3 and 6 also contain electron rich aromatic units in their structure, and have been crystallised from acetonitrile and dimethyl sulfoxide, respectively. In contrast to 5, both of them possess free carboxylic acid groups, which, along with the aromatic units, were expected to influence the solid-state packing. As depicted in Fig. 8, both of these structures crystallise in a regular network of π-stacked architectures. The distances between aromatic units involved in π–π interactions are similar for both molecules and range from 3.53 Å to 3.55 Å. In the case of 3 this arrangement is further reinforced by intermolecular hydrogen bond interactions between carboxylic acid groups (OH···O distance: 2.51 Å). Molecule 6 presents a more complex structure with three separate intermolecular interactions taking place: π–π interactions between aromatic groups (NDI and 4-hydroxyphenyl, 3.53 Å), and two distinct hydrogen bonding interactions (Fig. 8b). Both of them involve carboxylic acid groups, which interact on one site with a solvent molecule (DMSO, OH···O distance: 2.57 Å) and on the other site with an OH group of a tyrosine moiety from a neighbouring NDI (OH···O distance: 2.67 Å).
Fig. 8

(a) Side view of the crystal packing of 3 crystallised from acetonitrile indicating π–π and hydrogen bonding interactions between adjacent molecules in the solid state; (b) Side view of the crystal packing of 6 crystallised from DMSO indicating π–π interactions between aromatic units of the adjacent NDIs and hydrogen bonding interactions between carboxylic acids and solvent molecules

Crystals of achiral amino acid derived NDI 8 suitable for X-ray diffraction study were obtained from slow diffusion of water in DMSO. In this hydrogen bond acceptor solvent, the NDIs are arranged in a π-stacked structure, with the NDI cores n and n + 1 tilted at ca. 30° with respect to each other, while n and n + 2 are parallel to each other (dihedral angle 0.6°, Fig. 9).
Fig. 9

Part of the unit cell of 8 showing tilt of NDI n + 1 with respect to parallel NDIs n and n + 2

A derivative that stands apart in this analysis is NDI 7, in which the carboxylic acids (or esters) have been replaced by amide groups, thus allowing the possibility of a new type of hydrogen bonding interaction. The crystal structure of molecule 7 was obtained by slow evaporation of an acetone solution and the analysis showed a complex network of molecules connected via a large number of hydrogen bonds.

Each NDI 7 is connected with four distinct molecules via strong hydrogen bonding interactions between the amide units (Fig. 10, NH···O distance: 2.85 Å). Even though there are no pendant aromatic units and solvent molecules are not directly involved in the interactions between the molecules, no tubular object could be observed. This might be because the amide moieties of 7 have the s-trans geometry and they adopt anti geometry with respect to the NDI plane, thus preventing formation of the nanotube. The latter factor seems to be critical for the formation of the supramolecular nanotube in the solid state as only derivative 1 with syn geometry of the amino acid side chains was found to assemble into the tubular architecture. This also suggests that there is a very narrow line between formation of a nanotube and other architectures in the solid state whereby modifications of the chemical structure of the amino acid, crystallisation solvent and the introduction of new hydrogen bonding units do not result in formation of porous structure like that observed for derivative 1.
Fig. 10

(a) Crystal structure of molecule 7 and its hydrogen bonding interactions with four adjacent NDI molecules in the solid state. (b) Top view of the crystal packing of 7 indicating a complex network of hydrogen bonding interactions between amide units of 7

4 Nanotube Characterisation

The helical supramolecular nanotubes of NDI 1, observed in the solid state, were identified in CHCl3 and 1,1,2,2-tetrachloroethane (TCE) solution by means of circular dichroism (CD) and NMR spectroscopies, and were further studied using molecular modelling.

4.1 Solution State Characterisation

The first evidence for the presence in solution of NDI nanotubes of 1 came from CD spectroscopy, which measures the optical rotation of circular polarised light by chiral molecules and therefore is an indicator of chirality. A molecule with a chiral centre usually produces a small intrinsic CD signal; however, if the molecule can aggregate to form a chiral supramolecular species, then a much stronger CD signal can be generated, as the entire structure expresses chirality. In chloroform, 1 had an intense CD signal at 383 nm, corresponding to the absorbance of the naphthalene core. By contrast, the corresponding methyl ester derivative of 1 (l-1 ester) was CD silent at this wavelength. The l,l-enantiomer of 1 (l-1) derived from the corresponding l-amino acid forms P-helices with a positive CD signal, while the d,d-enantiomer (d-1) forms M-helices with a negative CD signal [15].

CD spectroscopy provided the compelling evidence that the nanotubes are held together by hydrogen bonding: addition of a competing hydrogen bonding solvent such as methanol reduced the CD signal dramatically to what would be expected for the intrinsic chirality of a non-aggregated NDI monomer (Fig. 11). Similar CD results were also obtained with other amino acid derivatives of NDI (25) and mono-chiral NDI derivatives (1215) [14].
Fig. 11

Comparison of CD signal for d-1, l-1, l-1 + methanol and l-1-ester [15]

In 1H-NMR spectroscopy, the NDI aromatic protons appear as a broad singlet at 8.65 ppm in CDCl3. The presence of this broad signal, rather than two doublets as would be expected from the structure depicted in Fig. 2, was rationalised by the dynamic nature of the nanotube: rapid exchange of NDI units out of and into the nanotube averages the two non-equivalent sites (Fig. 2b). When the 1H NMR spectra were recorded at 263 K the expected lack of symmetry is clearly visible (Fig. 12) [16].
Fig. 12

1H-NMR spectrum of the NDI aromatic protons of a 1.0 × 10−3 M solution of 1 in TCE at different temperatures

A similar splitting pattern of the NDI 1H peaks could also be observed at room temperature for an N-desymmetrised monochiral NDI molecule, 12. Figure 13 compares the naphthyl region of the 1H-NMR spectra of 12 in 5% MeOD in CDCl3 (in which nanotubes cannot form) and in pure CDCl3 (in which they can). In the methanolic solution, molecule 12 has a C2 axis of symmetry intersecting the two nitrogen atoms and therefore the naphthyl protons are observed as two doublets coupling to one another. In pure CDCl3 solution, these protons split into four doublets, indicative of the asymmetric environment of the NDI.
Fig. 13

Comparison of the naphthyl proton signals observed in the 1H-NMR spectrum of molecule 12 in (a) CDCl3 + MeOD: NDI core contains two distinct sets of protons (blue and red) and (b) CDCl3: nanotube formation further differentiates the blue and red NDI protons as non hydrogen-bonded (H*) and hydrogen-bonded (H) [17]

4.2 Majority Rules Study

Some chiral self-assembling systems are able to incorporate enantiomers of opposite chirality into the supramolecular structure because the chiral centre is relatively remote from the moiety that is involved in self-assembly [18, 19]. By contrast, NDI monomers are highly discriminating when it comes to the formation of chiral nanotubes: one enantiomer cannot be incorporated into a helix made of the other enantiomer because the directionality created at the chiral centre is critical to self-assembly. Therefore, when mixed, the two enantiomers self-sort, as shown by a “majority rules” study [20, 21, 22]. Equimolar TCE solutions of chiral NDIs l-1 and d-1 were titrated with small quantities of the opposite enantiomer solution and the CD trace recorded for each mixture. The resulting graph (Fig. 14) of the CDmax points plotted against percentage of the starting NDI present was approximately linear, suggesting simple dilution with no interaction between the two enantiomers.
Fig. 14

Datasets from majority rules studies involving mixtures of l-1 and d-1 [17]

4.3 Theoretical Studies of the CD of NDI

In order to correlate the solid state and solution phase structures, molecular modelling using the exciton matrix method was used to predict the CD spectrum of 1 from its crystal structure and was compared to the CD spectrum obtained in CHCl3 solutions [23]. The matrix parameters for NDI were created using the Franck–Condon data derived from complete-active space self-consistent fields (CASSCF) calculations, combined with multi-configurational second-order perturbation theory (CASPT2).

The minimal active space needed to describe the electronic structure of the NDI moiety includes the five occupied and five unoccupied π-orbitals of the naphthalene core and four lone pair orbitals of the carbonyl groups. The 5π[4n]5π active space contains 14 electrons and electronic transitions arise from seven states. Only two of the seven states, 1 1B2u and 1 1B3u, show transitions in the region of interest between 320 and 420 nm. Other transitions have no effect on the bands in this region and hence were not considered. The main features in the experimental absorbance spectrum were reproduced using the most intense Frank–Condon transitions (Fig. 15). The calculated spectrum (dashed lines) showed a red shift of 9 nm relative to the experiment, which may be due to the representation of each transition by only two charges, and also due to the neglect of other transitions.
Fig. 15

Experimental spectrum from solid state (dashed line) and calculated spectrum of a heptamer of molecule 1 (solid line). Upper panel: Absorbance spectrum, lower panel: CD spectrum [23]

The dependence of the calculated CD spectrum on the angle between two adjacent NDI planes was also studied (Fig. 16). The intensity of the CD signal decreases uniformly for greater angles, until no interaction is observed for coplanar chromophores. If the angle is increased further, a change in the sense of the helix occurred which inverted the signs of the bands, similar to a reversal of chirality at the α-carbon.
Fig. 16

Dependence of the calculated CD spectrum on the angle between the NDI planes of two monomers [23]

In order to analyse the effect of the nanotube length on the CD spectrum, the modelling was applied to several oligomers of molecule 1 (Fig. 17). In general, the intensity of the spectra increases with the number of monomers added to the structure. The distance between the chromophore centres across the tube core is about 12 Å and the interaction between such monomers (i and i + 2) is responsible for the increase in going from a dimer to a trimer. As further monomers are added, the increase in CD intensity gradually becomes less pronounced, as expected (Fig. 17).
Fig. 17

Spectra calculated for oligomers of molecule 1 (left). Change in intensity of the three most intense bands in the calculated spectra depending on the length of the oligomer (right) [23]

4.4 The “Sergeants-and-Soldiers” Effect

The “sergeants-and-soldiers” effect was first proposed in the field of polymer chemistry in the 1960s and was named by M.M. Green and co-workers in 1989 [24]. Since that time, E.W. Meijer and others have broadened the scope of the concept to take in many facets of supramolecular as well as polymer chemistry [18]. In a system displaying “sergeants-and-soldiers” behaviour a chiral derivative, the “sergeant”, imposes its chirality on a structure formed mainly out of achiral derivatives, the “soldiers”. In the case of NDI nanotubes, investigating the possibility of a “sergeants-and-soldiers” effect requires the synthesis of NDIs derived from achiral amino acids. Glycine is perhaps the most obvious choice, but the resulting NDI had previously been found to be highly insoluble in organic solvents. Several different achiral amino acids were therefore employed to synthesise a range of achiral NDIs. The NDIs derived from S-trityl cysteine (1 and 1-ester), or N-Boc lysine (2 and 2-ester) were employed as the sergeant, while achiral derivatives 811 were used as soldiers (Table 1).

In order to test the “sergeants-and-soldiers” effect in the NDI nanotube system, solutions of 1 and 8 (2.1 × 10−4 mol dm−3 each) were prepared in TCE. Chiral amplification and propagation experiments were performed by addition of 1 to a solution of 8 and addition of 8 to solution of 1, respectively. As a control, both experiments were then repeated substituting 8 with a solution of 1-ester. The overall concentration of NDI moieties therefore remains constant throughout the additions. The collated CDmax points plotted against percentage of 1 present indicate that 8 is incorporated into nanotubes formed of NDI 1, thus producing a more intense CD signal when compared to the control experiment (Fig. 18). Maximum amplification of the CD signal (2.5-fold) is observed at around 70% of 8.
Fig. 18

Plotted CDmax (383.5 nm) for all experiments. Concentration of all NDIs = 2.1 × 10−4 mol dm−3 [17]

A further experiment with the same protocol was used to test that the “sergeants-and-soldiers” behaviour was not unique to 1. Separate solutions of both enantiomers of 2 (l-2 and d-2) were each diluted with 8 and with their methyl ester derivatives (l-2-ester and d-2-ester) as control, thus leading to the formation of mixed nanotubes composed of “sergeants” 2 and “soldiers” 8 (Fig. 19).
Fig. 19

Plotted CDmax points for mixtures of l-2/8 (filled squares) or l-2/l-2-ester (open squares); and d-2/8 mixtures (filled circles) or d-2/d-2-ester (open circles). Concentration of all NDIs = 2.1 × 10−4 mol dm−3 [17]

Of the four achiral NDIs studied, only derivatives 8 and 10 acted as “soldiers”, while derivatives 9 and 11 were not incorporated into supramolecular nanotubes. This suggests that the superstructure of the heterogeneous nanotubes is subtly different to that of the homogenous 1 nanotube. The difference is likely to be due to replacement of the α-proton with a significantly bulkier group. The steric repulsion of two alkyl groups on 8 is balanced by the cyclopropyl ring holding them tightly together, meaning the N–C–C(O) angle in 8 is within the range of those of 1 and 2, both of which readily form nanotubes (Fig. 20). However, the added rigidity imposed by the cyclopropyl moiety forces 8 to take up conformations that are slightly different to the minimum energetic conformation for the chiral amino acid-derived NDIs (which present the α-proton in the most sterically crowded position).
Fig. 20

Ring strain in NDI 8 counteracts the repulsion of the alkyl carbons to bring the N–C–C(O) angle back within the range of an NDI derived from achiral amino acid with only one R group (1, 2). The lesser ring strain in 10 produces a reduced version of this effect, while it is not present at all in unstrained 9 or 11. The degree of difference in the N–C–C(O) angle is exaggerated for clarity in the diagrams. Where crystal structures of the NDIs were not available, estimates have been made by averaging the values found in crystal structures of related compounds in the CSD [25]

The importance of the cyclopropyl strain in 8 is demonstrated by the fact that achiral NDIs 9 and 11 derived from achiral amino acids with reduced steric strain were observed to have no significant “sergeants-and-soldiers” activity. In achiral NDI 10, the ring strain is lower than in 8 but higher than in both 9 and 11, and as expected it acts as a “soldier”, albeit inferior to 8 (Fig. 21). This supports the idea that bond angle must be key to the molecule’s suitability as a soldier: it is the only significant molecular distinction between NDIs 8, 9 and 10, yet accounts for the variation in their properties.
Fig. 21

Example data sets for NDIs 8 (black filled circles) and 10 (grey filled circles) acting as soldiers to 1 as sergeant. Control data using 1-ester as inactive soldier is also plotted (open circles). Concentration of all NDIs = 2.1 × 10−4 mol dm−3

As can be seen in Fig. 21, solutions of chiral NDIs 1 containing small percentages of 8 tend to show stronger CD signals than 100% chiral solutions of the same total NDI concentration. The effect is small but reproducible, and must be the outcome of subtle geometrical changes in the size of the exciton coupling between adjacent chromophores, thus leading to an increased CD signal [17].

The “sergeant-and-soldier” experiment was applied in a system that contained a monochiral NDI 1214, and the achiral derivatives 8, 9 and 10 [14]. This is an extreme case of the “sergeant-and-soldier” experiment in which one chiral amino acid residue controls the assembly of a mixed NDI nanotube by imposing its chirality upon at least three other achiral centres. The experiments were carried out using a modified protocol in which the concentration of the “soldier” solution was doubled compared to the “sergeant” solution (Fig. 22). These experiments showed that the monochiral 12 is a better sergeant than 13, which is in agreement with the results obtained in the “sergeant-and-soldier” experiments using achiral and chiral NDIs.
Fig. 22

“Sergeant-and-soldier” behaviour of monochiral NDI 12 [14]

5 C60 Encapsulation

The nanotubular cavities (mean diameter: 12.4 Å) were found to be effective hosts for C60 molecules (van der Waals radius: 10.3 Å), and were capable of solubilising C60 in solvents such as chloroform, where the fullerene has poor solubility. UV–vis, CD, 13C-NMR and molecular modelling were used to characterize this NDI nanotube-C60 host–guest complex [26].

5.1 UV–Vis Spectroscopy

A chloroform solution of 1 when left to stand over solid C60 shows a drastic colour change from pale yellow to brown within a few minutes. This is different from the expected summation of colours for a saturated solution of C60 in chloroform (pale purple) and the initial pale yellow solution (Fig. 23, inset). The visible region of the absorption spectrum (Fig. 23, inset) is marked by the appearance of a broad band centred at 452 nm (identified with *), in addition to the absorbance bands characteristic of C60. This broad band is usually associated with C60 films and aggregates and has been attributed to interactions between fullerenes, implying that fullerenes inside the nanotube are in close proximity, forming a one-dimensional C60 array.
Fig. 23

UV–vis trace of C60, 1, 1 + C60 and the C60 contribution to 1 + C60 in CHCl3. The concentration of 1 for the measurement of the visible spectrum (400–800 nm) was 40 times higher than the UV region [26]

The uptake of C60 corresponds to the increase in absorbance at 258 and 328 nm in the absorption spectrum of 1 + C60 (Fig. 23). Comparison of a solution of 1 + C60 with a saturated solution of C60 in chloroform showed that the C60 concentration increased 16-fold in the presence of NDI nanotubes (1). In contrast with these results, the methyl ester of 1, which is unable to form hydrogen-bonded supramolecular nanotubes, did not enhance the solubility of C60 in chloroform, supporting the thesis that the C60 molecules are complexed in the inner nanotubular cavity. The increase in absorbance at 258 nm also led to an estimate of [NDI]/[C60] stoichiometry, revealing that an average of 3.6 NDI units were encapsulating one C60 molecule. Similar results were also obtained with other amino acid derivatives of NDI.

5.2 CD Spectroscopy

In the UV region, where the NDI chromophore absorbs, the CD signal of the complex is broadly unchanged from that of the nanotube alone, indicating that the structure is preserved. In the visible region, where the host is silent, a weak induced circular dichroism (ICD) signal at 595 and 663 nm is observed, suggesting that C60 also experiences the helicity of the environment (Fig. 24). It is not clear whether this is a direct sensing of the amino acid chirality, or whether the nanotube actually has a long-range chiral supercoiled structure.
Fig. 24

Comparison of the CD spectra of 1 and 1 + C60 for the l and d enantiomers [26]

5.3 13C-NMR Experiments

Further evidence for the uptake of C60 came from 13C NMR spectroscopy, in which the signal for C60 is shifted upfield by over 1.4 ppm upon complexation by NDI nanotubes (Fig. 25), indicative of a shielding effect due to the proximity of the NDI aromatic units (and possibly also C60–C60 proximity). The retention of high symmetry of the fullerene upon binding within the nanotube indicates that spinning of the fullerene is fast on the 13C NMR timescale.
Fig. 25

The 13C NMR signal in CDCl3 of C60, in the absence and presence of the nanotubes (left). Looking down the axis of a space filler model of the nanotube (centre). Looking laterally at a ball and stick model of the helix, as found in the molecular modelling studies (right). AM1 semi-empirical level using Hyperchem 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 Å−1 mol−1

13C NMR experiments in TCE-d2 confirmed that this functional behaviour of the nanotubes is also present in hybrid nanotubes composed of a mixture of chiral 1 and achiral 8 [14]. The C60 uptake is lower for the hybrid nanotubes with increasing amounts of incorporated achiral NDIs. A similar effect was observed for the monochiral NDI, in which nanotubes containing the cyclopropyl side chain 12 are better receptors for C60 than the analogues formed from 13. These observations highlight the importance of the rigidity in the NDIs derived from achiral amino acids.

5.4 Self-Sorting of Nanotubes

Complexation of C60 also remarkably demonstrated the self-sorting of nanotubes of opposite helicity. A 1:1 mixture of l-1 and d-1 is capable of encapsulating C60 as shown by UV–vis spectroscopy: the [NDI]/[C60] ratio of 3.9 matches (within experimental error) that is obtained for optically pure samples of either l-1 or d-1, and the 13C signal of C60 is shielded by 1.4 ppm, indicating that nanotubes are still present (even though they are inevitably invisible by CD). This quantitative self-sorting of the two helical nanotube enantiomers shows that one NDI enantiomer cannot be incorporated into a helix made of the other enantiomer as the directionality created at the chiral centre is critical to self-assembly and is in agreement with the “majority-rule” experiments (see above).

6 Self-Assembled C70 Receptors

As described above, the NDI nanotubes can complex a “string” of C60 molecules inside the tubular cavity. However, C70, leads to complete destruction of the nanotubes by templating the formation of a new discrete receptor. The disassembly of the nanotube was immediately evident from the change in the CD spectrum of a solution of NDI on the addition of C70 (Fig. 26a) [27]. Equally striking differences were observed in the 1H NMR spectra obtained from solutions prepared in CDCl3 (Fig. 26b). The radically different CD and 1H NMR spectra and the fact that derivatives 1 and 2 give similar CD and NMR signatures in the presence of C70 is indicative of the formation of a new supramolecular structure. 1H NMR titration experiments (supported by CD data) led to the conclusion that six NDIs interact with one molecule of C70, thus forming a hexameric receptor.
Fig. 26

(a) CD spectra of 3 in the absence and in the presence of C70 at [3] = 3.86 × 10−4 M in CHCl3; (b) 1H NMR spectra of empty nanotube and C70 receptor of 3 in CDCl3

The 1H NMR spectrum (Fig. 26b) of nanotubes of 3 (3N) shows a high degree of symmetry experienced by the NDI units, as opposed to the signature of the C70 receptor, which shows four doublets for the aromatic protons: three of these have a similar chemical shift, while one is shifted downfield by 0.78 ppm. Similarly, there are two signals for the α-protons of the C70 receptor rather than one in the nanotube. 13C NMR spectra of C70 in the absence and in the presence of NDI indicate clear shift differences with roughly equal effects (2.0–2.4 ppm) on all five inequivalent carbon signals. The four aromatic signals observed in the 1H NMR spectrum of the C70 receptor might arise, at first sight, from four different situations (Fig. 27b–e): four inequivalent protons on a single NDI molecule (b), four equivalent protons on each of four inequivalent NDI molecules (c) or two pairs of protons on two inequivalent NDI molecules (d, e). Situations (c) and (d) can be ruled out, as they would give rise to four different signals of the α-protons rather than two. Situation (e) can also be ruled out, as it would result in four singlets for the aromatic protons. The number and multiplicity of the signals in the 1H NMR spectrum of 3 + C70 clearly point to the situation pictured in Fig. 28b. Therefore, all NDI molecules are equivalent, yet none of them lie on a symmetry element of the assembly.
Fig. 27

(a) Symmetry elements of an NDI derivative: symmetry plane σ (red) and twofold axis (blue); (be) different combinations of symmetry elements and number of inequivalent NDI molecules that give rise to four aromatic signals [27]

Fig. 28

Possible arrangements of NDI 12 in the C70 receptor. The single α-proton signal means either A or B are possible, but C can be eliminated

A detailed symmetry analysis based on spectroscopic data showed that a 6:1 stoichiometry with a D3 point-group symmetry is the only plausible structural class of the C70 receptor. Thus, at the “poles” of the C70 receptor, three NDI molecules have to bind to each other in a C3-symmetrical manner, requiring angles of 120o (Fig. 28). Weak CH···O hydrogen bonding to an imide or carboxylic acid carbonyl group could explain the chemical shift change of more than 0.7 ppm for one aromatic proton in the C70 receptor (Fig. 26b). Indeed, such an arrangement can give rise to a favourable trimeric interaction mode (Fig. 28), accounting for formation of one trimeric “half” of the receptor. In the proposed arrangement, consistent with D3 point symmetry for the complex, one carboxylic residue per NDI unit would remain free for carboxylic acid dimerization with the NDI counterpart of the other hemisphere, at the equator of the C70 receptor.

In chloroform, the monochiral NDIs were found to form the C70 receptor in the presence of an excess of C70. However, unlike 1, they did not form the receptor exclusively, rather the equilibrium position between receptor and nanotube is different for each monochiral NDI, as indicated by 1H NMR experiments. As expected, 12 NDI was the most efficient at forming the hexameric capsule with 70% of the material being incorporated, while only 30% of 14 forms the receptor. Figure 28 shows a cartoon representation of the proposed geometry of the C70 receptor including the possible orientations for an N-desymmetrised NDI [17].

Only one α-proton signal is seen in the 1H NMR spectrum of the C70 receptor formed from 12, which means that all the chiral ends of the NDIs are in the same environment. It is unclear whether the single α-proton signal of the monochiral component is associated with the equatorial or axial position, but this demonstrates that the arrangement of NDIs in the receptor is ordered rather than random (Fig. 28).

7 Proton-Driven Switching Between C60 and C70 Receptors

Morphological switching between nanotube, hexameric receptor and monomers is readily achieved by simple protonation-deprotonation reactions. This system can be described as a library of dynamic, size selective fullerene receptors whose structure and recognition properties depend on the position of the acid/base equilibrium (Fig. 29) [28]. Unexpected differences in the sensitivity to base-induced dissociation of the nanotubes derived from different amino acids were also uncovered.
Fig. 29

Schematic representation of the proton-driven cyclic morphological switching between NDI monomers, C60 and C70 receptors (the clock indicates the slow kinetics of C60 uptake)

Chiro-optical studies were carried out in chloroform solution with four structurally diverse hydrogen-bonded nanotubes 14 using triethylamine (TEA) and methanesulfonic acid (MSA) as base and acid triggers. CD spectra of a chloroform solution of 2 (red trace) after sequential additions of base and acid are shown in Fig. 30. Addition of 1 equiv. of base caused a dramatic decrease of the CD signal intensity (green trace), attributed to the dissociation of the nanotube by the breaking of hydrogen bonds between NDI components. Subsequent addition of a stoichiometric amount of acid re-established the original spectrum (blue trace). This process is reversible, as demonstrated by the essentially complete recovery of the CD over several cycles (Fig. 30, inset). Comparison of these measurements for all four NDI derivatives revealed that the amount of base required to dissociate the nanotube architecture depends on the nature of the amino acid side chain. The nanotubes of 1, 3 and 4, all with apolar substituents, required more base (4, 2 and 2 equiv. per NDI, respectively) for complete dissociation than that needed for dissociation of 2 (1 equiv.), with a polar side chain. This may be a consequence of differences in solvation and/or creation of a more non-polar environment, which would raise the pKa in a manner that is reminiscent of carboxylic groups in enzyme active sites [28]. It is not clear whether removal of, on average, one proton per (COOH)2 link (which would still allow connection via a single, charge-assisted hydrogen bond) leads to the dissociation of the nanotubes or whether both protons need to be removed. In all cases, the nanotubes re-assembled when MSA was added to neutralise the base.
Fig. 30

CD spectra of a CHCl3 solution of 2 (7 × 10−4 M, red trace) in the presence of one equivalent of TEA (green trace) and an additional 1 equiv. of MSA (blue trace). Inset: demonstration of the reversibility of the base–acid driven switch between the nanotube and free NDI components [28]

Investigations were performed to examine whether the C60 and C70 guests have any influence on the host’s resistance to base-induced dissociation. While acid/base experiments in the presence and absence of C60 produced essentially identical dissociation and re-association results, C70 behaved quite differently. Thus, the C70 receptor was formed by the addition of C70 to a solution of 1 in dry chloroform (ratio of NDI to C70 was 6:1, Fig. 31a, black trace). The addition of 1 equiv. of TEA to the 1 + C70 complex leads to the disassembly of the C70 receptor and the formation of a supramolecular nanotube, as indicated by the CD spectrum showing a characteristic positive signal at 383 nm (Fig. 31a, red trace). This remarkable morphological switching reveals a difference in stability of the H-bonding arrays in the C70 capsule and the nanotube, the latter being particularly stable when derived from 1. In the cysteine case, 1, 1 equiv. of base (per NDI) is sufficient to destroy the C70 receptor but not the nanotube. Presumably, a partially deprotonated nanotube may co-exist in solution with deprotonated NDIs and free C70. The subsequent addition of a further 3 equiv. of TEA results in the decrease and finally disappearance of the characteristic nanotube CD signal at 383 nm (Fig. 31a, violet trace). The reversibility of the processes was confirmed by stepwise addition of equimolar amounts of acid, which first regenerated the nanotube, and then the C70 receptor (Fig. 31b, orange and blue traces, respectively). Furthermore, this proton-controlled morphological switching between supramolecular architectures strongly depends on the structure of the NDI component. Thus, addition of 1 equiv. of TEA to the C70 receptor involving 2 resulted in complete dissociation of the supramolecular architecture giving free NDI components, completely by-passing the nanotube phase.
Fig. 31

Evolution in the CD spectrum of a CHCl3 solution of 1 + C70 (7 × 10−4 M) after addition of (a) 4 equiv. of TEA and (b) 4 equiv. of MSA [28]

The nanotube, the C70 receptor and the uncomplexed NDI molecules have distinct 1H NMR spectral signatures, particularly in the aromatic region of the spectra, providing a clear window on the switching between these three architectures. Thus, starting with a solution of 3 + C70 (ratio of NDI to C70 was 6:1) and adding one equivalent of TEA, the C70 receptor peaks (9.5–8.4 ppm for the NDI core and 6.9, 6.1 ppm the α-protons) were replaced by two signals at 8.5 (NDI) and 5.8 ppm (α) characteristic of the nanotube structure (Fig. 32). Addition of another 1 equiv. of TEA resulted in dissociation of the nanotube to the free NDI molecules as indicated by the sharpening and downfield shifts of the two signals to 8.7 and 6.0 ppm, respectively. Reversibility was demonstrated by progressive addition of 2 equiv. of MSA which first reformed the nanotube, followed by the C70 receptor [29].
Fig. 32

Part of the 500 MHz 1H NMR spectra of 3 + C70 showing the acid–base driven reversible switching between the C70 receptor, nanotube and free NDI components in CDCl3 at 7 × 10−4 M [28]

To illustrate further the potential of this system, two fullerene guests were simultaneously employed in a competition experiment. The morphological switching was followed by 13C NMR spectroscopy and, as in the previous experiments, this showed preferential formation of the C70 complex over that of the C60/nanotube species in the absence of base (the four signals between 143 and 148 ppm are due to the complexed C70, Fig. 33a). Progressive addition of base caused, in the first instance, appearance of an additional signal at 142.9 ppm, characteristic of C60 within a nanotube as part of a ternary complex with triethylammonium [30] (Fig. 33b) followed by its significant amplification when more base was added (Fig. 33c). At this stage, both host–guest complexes were evident but the nanotube + C60 complex was dominant. Further addition of base first caused complete disappearance of C70 receptor signals and finally disassembly of the nanotube (Fig. 33d, e).
Fig. 33

Part of the 500 MHz 13C NMR spectra of 3 + C70 + C60 showing the acid–base driven reversible switching between the C70 receptor, nanotube-C60 based receptor and free NDI components in CDCl3 at 7 × 10−4 M

8 Complexation of Polyaromatic Hydrocarbons

Expanding our search to “fullerene fragments” we investigated the potential host–guest chemistry of the nanotubes with polyaromatic hydrocarbons shown in Fig. 34 [30].
Fig. 34

Polyaromatic guests used in this investigation. (IVI) were found to interact the nanotube while (VIIX) show no interaction

The change in colour upon addition of different guests to the preformed NDI nanotubes in CHCl3 was a good indicator of complexation for all the examples studied. Fluorene I and all the pyrene derivatives II–VI led to a dramatic colour change of the chloroform solution of 1 from pale yellow to deep red as illustrated in Fig. 35 (inset: for pyrene), indicative of strong NDI-pyrene donor-acceptor interactions. No significant colour change was observed when the larger analogues VII–X were added under similar conditions. These species were presumably too large to fit within the nanotube cavity. Control experiments with the corresponding NDI-methyl ester derivative 1-ester gave no colour change, regardless of the aromatic molecules present in solution, demonstrating that the complexation is a property of the supramolecular nanotube system rather than individual NDI monomers.
Fig. 35

CD and UV spectra of a 4:1 mixture of 1 and II in chloroform (inset: colour changes upon mixing the two solutions); [1] = 1.0 × 10−4 M in CHCl3 at 21 °C [30]

Chiro-optical studies confirmed the persistence of the NDI nanotubes in solution, indicating that the geometry of the supramolecular nanotubes is preserved upon complexation of pyrene (Fig. 36). An induced Cotton effect (ICD) at λmax = 440 nm corresponding to the charge transfer transition between II and the NDI nanotube was observed. Surprisingly, addition of small amounts of methanol (<1 vol.%), known to disassemble the nanotube, led to an increase of the ICD band, confirming that the host–guest complexation is driven by solvophobic interactions. Increasing the MeOH concentration above 1% led to the expected destruction of the supramolecular nanotubes as manifested in a decrease in the ICD band (Fig. 36).
Fig. 36

CD spectra of a solution of 1 and II in the presence of increasing quantities of MeOH. Inset: change in the CD intensity at 550 nm with increasing amounts of MeOH. (1:II in 4:1 ratio, [II] = 5 × 10−3 M in CHCl3 at 21 °C) [30]

In a mixture containing a 2:1 molar ratio of 1 and II the NDI proton signal of 1 and all the protons of II are shifted upfield by 0.09 and 0.08 ppm, respectively, while in a similar mixture containing 1-ester and II the same protons are shifted upfield by only 0.04 ppm (Fig. 37). This behaviour is consistent with the formation of an inclusion complex between 1 and II, where the pyrene molecules are shielded by the naphthalene aromatic cores of the NDIs, which in turn are shielded by the pyrenes complexed inside the nanotube. Spectrum d in Fig. 37 shows that, although there is some interaction between pyrene II and an individual NDI (1-ester), it is nonspecific and cannot explain the larger chemical shifts observed in the case of nanotube 1 and II. The 1H-NMR experiments were reflected in the 13C-NMR where both the NDI and the pyrene peaks were shifted upfield in the 1 + II complex vs the free species in solution.
Fig. 37

Aromatic region of the 1H-NMR spectra of 1 (a), 1-ester (e), II (c) and the corresponding 2:1 mixtures: 1 + II (b) and 1-ester + II (d). Colour coding: red = NDI protons of 1, blue = pyrene (II) protons. [1] = [1-ester] = 41 × 10−3 M in CDCl3 at 23 °C [30]

9 Complexation of Ion Pairs

The supramolecular nanotubes act as size selective receptors not only for polyaromatic hydrocarbons but also for ion pairs (Fig. 38) as demonstrated by changes in the 1H chemical shifts (Δδ) of the host signals [30].
Fig. 38

Sizes (in Å) of anions and cations (ion pairs) that interact with the NDI nanotubes [25]

A direct correlation between Δδ for the NDI protons and the dimensions of the ammonium ions was observed, leading to the conclusion that smaller ammonium ions are better guests than their larger analogues [30]. Solvophobic effects were also important in determining a guest’s affinity for the nanotube. For example the addition of XVI (least soluble of all the ammonium ions tested) to the nanotube solution resulted in the largest Δδ observed for both series (bromides and chlorides). Its behaviour is similar to that for C60 whereby the solubility of the guest increases in the presence of the host. The comparison of the Δδ values recorded for the two series of ion pairs revealed that the size of the anion is equally important in the complexation event (Fig. 39). Ammonium chlorides were more readily taken up by the nanotube than the corresponding larger bromides. The largest shift was observed upon addition of acetylcholine chloride (XIII.Cl), which is the smallest of the ion pairs tested.
Fig. 39

Comparison of the change in 1H NMR chemical shift of the NDI α proton, aromatic protons and the benzylic protons of the ammonium ion [30]

Proof for the complexation of the anion inside the nanotube’s cavity came from 19F-NMR spectra of the inclusion complex between 2 and XIV·BF4 which showed a downfield shift (ca. 0.41 ppm) of the fluorine atoms (Fig. 40). The magnitude of the shift observed for this system was comparable with the downfield shift observed for PF6 trapped inside a tetrahedral aromatic cage (0.64 ppm) [31].
Fig. 40

19F-NMR spectra of a solution of XIV.BF4 in the presence (red line) and in the absence (blue line) of 2. [2] = 20 mM, [XIV.BF4] = 5 mM at 23 °C in CDCl3 [30]

Competition experiments showed that C60 could be displaced from the nanotube by selected ion pairs, resulting in the formation of mixed C60-ion pair complexes [32]. The addition of 1 equiv. of XIV·Cl to a 10-mM solution of C60-encapsulated 2 resulted in the precipitation of solid C60 and a colour change of the supernatant from brown to light orange (Fig. 41). The displacement of C60 from the nanotubular cavity was confirmed using UV–vis and CD spectroscopy; the intensity of absorption and induced CD (ICD) signals at 593 and 660 nm (characteristic of C60) decreased dramatically upon addition of 1 equiv. of XIV·Cl and stayed constant when XIV·Br and XIV·I were added. These results demonstrate that the interaction of XIV·Cl with the nanotube is stronger than that of XIV·Br or XIV·I, as the latter are incapable of displacing C60 from the cavity of the host (Fig. 42)
Fig. 41

Schematic representation of the self-assembled nanotube and the competition experiment with C60 and ion pairs. Picture: photographical comparison of a solution of 2 + C60 in the presence of different ion pairs [32]

Fig. 42

UV–vis spectra of a 10 mM solution of 2 + C60 in the presence of 1 equiv. of different ion pairs [32]

Increasing the amount of XIV·Cl in a solution containing 2 + C60 complex also resulted in the rapid decrease of absorption at 452 nm (attributed to fullerene–fullerene interactions in a closed-packed one-dimensional array of C60 inside the nanotubular cavity), indicating that the encapsulation of ammonium ions led to partial disruption of the close-packed C60 array resulting in the formation of a mixed complex ion-pair/C60 host–guest complex, where the ion pair is intercalating between the fullerenes.

The formation of the mixed ion-pair/C60 host–guest complex is best illustrated by 13C-NMR spectroscopy. As described earlier, the 13C signal of C60 is shifted upfield by more than 1.4 ppm upon encapsulation by the nanotube due to proximal aromatic units (Fig. 43). The addition of XIV·Cl (0.5 and 1 equiv.) to a 10-mM solution of containing this C60 complex in CDCl3 resulted in a small but significant downfield shift and a decrease in the intensity of the 13C signal, demonstrating the release of C60 in solution. The presence of a C60 signal even after the addition of 1 equiv. of ion pair indicated that only partial displacement of C60 occurred; this is possible only if a mixed ion-pair/C60 complexation takes place. The downfield shift of the 13C signal of C60 towards the uncomplexed position represents further evidence for the disruption of the close packed array of C60. The presence of the ion pair inside the tubular cavity is confirmed by the expected upfield shift observed in the 1H NMR spectrum, for the cation’s benzyl protons.
Fig. 43

C6013C chemical variation upon addition of XIV·Cl [32]

The reverse experiments, in which C60 was added to solutions of XIV·Cl encapsulated 2, also led to the formation of the same nanotube–fullerene–ion pair mixed complex. UV and NMR experiments confirmed that C60 partially displaced ion pairs from the nanotube’s cavity. This is indicative of the dynamic nature of the systems and the propensity of these nanotubes to form mixed complexes.

10 Conclusions

The serendipitous discovery that amino acid derived NDIs form supramolecular nanotubes in the solid state led to the unravelling of a very rich vein of supramolecular assemblies in organic solvents. This chemistry highlights how a very simple and versatile building block can form hydrogen-bonded dynamic combinatorial libraries that allow the co-existence of small oligomers, nanotubes and hexameric capsules. Achiral and mono-chiral derivatives allowed us to discover how very small changes in their structure influenced incorporation into chiral supramolecular structures. The formation of a dynamic nanoreceptor whose morphology and recognition properties can be tuned by a simple acid–base equilibrium highlighted the importance of the amino acid side chains in the formation of these supramolecular structures. The remarkable ability of NDIs to self-assemble in a receptor for C60, C70, polyaromatic molecules and ion pairs is unprecedented and should inspire chemists to investigate other “simple” systems whose abundant supramolecular chemistry is yet to be discovered.


  1. 1.
    Reek JNH, Otto S (2010) Dynamic combinatorial chemistry. Wiley-VCH, WeinheimCrossRefGoogle Scholar
  2. 2.
    Lehn JM (2007) From supramolecular chemistry towards constitutional dynamic chemistry and adaptive chemistry. Chem Soc Rev 36:151–160CrossRefGoogle Scholar
  3. 3.
    Lehn JM (1999) Dynamic combinatorial chemistry and virtual combinatorial libraries. Chem Eur J 5:2455–2463CrossRefGoogle Scholar
  4. 4.
    Corbett PT, Leclaire L, Vial L, West KR, Weitor J-L, Sanders JKM, Otto S (2006) Dynamic combinatorial chemistry. Chem Rev 106:3652–3711CrossRefGoogle Scholar
  5. 5.
    Herrmann A (2009) Dynamic mixtures and combinatorial libraries: imines as probes for molecular evolution at the interface between chemistry and biology. Org Biomol Chem 7:3195–3204CrossRefGoogle Scholar
  6. 6.
    Ladame S (2008) Dynamic combinatorial chemistry: on the road to fulfilling the promise. Org Biomol Chem 6:219–226CrossRefGoogle Scholar
  7. 7.
    Hunter CA (2004) Quantifying intermolecular interactions: guidelines for the molecular recognition toolbox. Angew Chem Int Ed 43:5310–5324CrossRefGoogle Scholar
  8. 8.
    Crego Calama M, Hulst R, Fokkens R, Nibbering NMM, Timmerman P, Reinhoudt DN (1998) Libraries of non-covalent hydrogen-bonded assemblies; combinatorial synthesis of supramolecular systems. Chem Commun 1021–1022Google Scholar
  9. 9.
    Timmerman P, Vreekamp RH, Hulst R, Verboom W, Reinhoudt DN, Rissanen K, Udachin KA, Ripmeester J (1997) Noncovalent assembly of functional groups on calix[4]arene molecular boxes. Chem Eur J 3:1823–1832CrossRefGoogle Scholar
  10. 10.
    Hof F, Nuckolls C, Rebek J Jr (2000) Diversity and selection in self-assembled tetrameric capsules. J Am Chem Soc 122:4251–4252CrossRefGoogle Scholar
  11. 11.
    Wyler R, de Mendoza J, Rebek J Jr (1993) A synthetic cavity assembles through self-complementary hydrogen bonds. Angew Chem Int Ed 32:1699–1701CrossRefGoogle Scholar
  12. 12.
    Pengo P, Pantoş GD, Otto S, Sanders JKM (2006) Efficient and mild microwave-assisted stepwise functionalization of naphthalenediimide with α-amino acids. J Org Chem 71:7063–7066CrossRefGoogle Scholar
  13. 13.
    Tambara K, Ponnuswamy N, Hennrich G, Pantoş GD (2011) Microwave-assisted synthesis of naphthalenemonoimide and N-desymmetrized naphthalenediimides. J Org Chem 76:3338–3347CrossRefGoogle Scholar
  14. 14.
    Anderson TW, Pantoş GD, Sanders JKM (2011) Supramolecular chemistry of monochiral naphthalenediimides. Org Biomol Chem 9:7547–7553Google Scholar
  15. 15.
    Pantoş GD, Pengo P, Sanders JKM (2007) Hydrogen-bonded helical organic nanotubes. Angew Chem Int Ed 46:194–197CrossRefGoogle Scholar
  16. 16.
    Ponnuswamy N, Pantoş GD, Smulders MMJ, Sanders JKM (2012) Thermodynamics of Supramolecular Naphthalenediimide Nanotube Formation: The Influence of solvents, side-chains and guest templates. J Am Chem Soc doi:10.1021/ja2088647
  17. 17.
    Anderson TW, Pantoş GD, Sanders JKM (2010) The sergeants-and-soldiers effect: chiral amplification in naphthalenediimide nanotubes. Org Biomol Chem 8:4274–4280CrossRefGoogle Scholar
  18. 18.
    Palmans ARA, Meijer EW (2007) Amplification of chirality in dynamic supramolecular aggregates. Angew Chem Int Ed 46:8948–8968CrossRefGoogle Scholar
  19. 19.
    Smulders MMJ, Filot IAW, Leenders JMA, van der Schoot P, Palmans ARA, Schenning APHJ, Meijer EW (2010) Tuning the extent of chiral amplification by temperature in a dynamic supramolecular polymer. J Am Chem Soc 132:611–619CrossRefGoogle Scholar
  20. 20.
    Smulders MMJ, Stals PJM, Mes T, Paffen TF, Schenning APHJ, Palmans ARA, Meijer EW (2010) Probing the limits of the majority-rules principle in a dynamic supramolecular polymer. J Am Chem Soc 132:620–626CrossRefGoogle Scholar
  21. 21.
    Van Gestel J (2004) Amplification of chirality in helical supramolecular polymers. The majority-rules principle. Macromolecules 37:3894–3898CrossRefGoogle Scholar
  22. 22.
    Van Gestel J, Palmans ARA, Titulaer B, Vekemans JAJM, Meijer EW (2005) “Majority-rules” operative in chiral columnar stacks of C 3-symmetrical molecules. J Am Chem Soc 127:5490–5494CrossRefGoogle Scholar
  23. 23.
    Bulheller BM, Pantoş GD, Sanders JKM, Hirst JD (2009) Electronic structure and circular dichroism spectroscopy of naphthalenediimide nanotubes. Phys Chem Chem Phys 11:6060–6065CrossRefGoogle Scholar
  24. 24.
    Green MM, Reidy MP, Johnson RJ, Darling G, Oleary DJ, Willson G (1989) Macromolecular stereochemistry: the out-of-proportion influence of optically active comonomers on the conformational characteristics of polyisocyanates. The sergeants and soldiers experiment. J Am Chem Soc 111:6452–6454CrossRefGoogle Scholar
  25. 25.
    Cambridge Structural Database, v.5.30, 2009Google Scholar
  26. 26.
    Pantoş GD, Wietor J-L, Sanders JKM (2007) Filling helical nanotubes with C60. Angew Chem Int Ed 46:2238–2240CrossRefGoogle Scholar
  27. 27.
    Wietor J-L, Pantoş GD, Sanders JKM (2008) Templated amplification of an unexpected receptor for C70. Angew Chem Int Ed 47:2689–2692CrossRefGoogle Scholar
  28. 28.
    Harris TK, Turner GJ (2002) Structural basis of perturbed pKa values of catalytic groups in enzyme active sites. IUBMB Life 53:85–98CrossRefGoogle Scholar
  29. 29.
    Stefankiewicz AR, Tamanini E, Pantoş GD, Sanders JKM (2011) Proton-driven switching between receptors for C60 and C70. Angew Chem Int Ed 50:5725–5728CrossRefGoogle Scholar
  30. 30.
    Tamanini E, Ponnuswamy N, Pantoş GD, Sanders JKM (2009) New host–guest chemistry of supramolecular nanotubes. Faraday Discuss 145:205–218CrossRefGoogle Scholar
  31. 31.
    Glasson CRK, Meehan GV, Clegg JK, Lindoy LF, Turner P, Duriska MB, Willis R (2008) A new FeIIquaterpyridyl M4L6 tetrahedron exhibiting selective anion binding. Chem Commun 1190–1192Google Scholar
  32. 32.
    Tamanini E, Pantoş GD, Sanders JKM (2010) Ion pairs and C60: simultaneous guests in supramolecular nanotubes. Chem Eur J 16:81–84CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Nandhini Ponnuswamy
    • 1
  • Artur R. Stefankiewicz
    • 1
  • Jeremy K. M. Sanders
    • 1
  • G. Dan Pantoş
    • 2
  1. 1.Department of ChemistryUniversity of CambridgeCambridgeUK
  2. 2.Department of ChemistryUniversity of BathBathUK

Personalised recommendations