Jack K. Clegg*ab, Jonathan Cremersa, Andrew J. Hogbena, Boris Breinera, Maarten M. J. Smuldersac, John D. Thoburnad and Jonathan R. Nitschke*a
aDepartment of Chemistry, The University of Cambridge, Lensfield Road, Cambridge, CB2 1EW, UK. E-mail: jrn34@cam.ac.uk
bSchool of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane St Lucia, QLD 4072, Australia. E-mail: j.clegg@uq.edu.au
cBiomolecular Nanotechnology (BNT), MESA+ Institute for Nanotechnology, University of Twente, 7500AE Enschede, The Netherlands
dDepartment of Chemistry, Randolph-Macon College, Ashland, VA 23005, USA
First published on 5th October 2012
A new cationic Fe4L6 cage molecule was synthesised from 4,4′-diaminobiphenyl, 2-formylpyridine and iron(II). The cage exists as a system of interconverting diastereomers in solution. The system adapts to the addition of anionic guest molecules, expressing a new combination of diastereomers that synergistically bind the guest molecules. Not only do the cage diastereomers interconvert, the volume of the individual cages adapts physically through the rotation of bonds, providing a tailored binding pocket for the guest lined with hydrogen-bond donors. A model for the resulting complex network of species was developed that allowed the system to be fully described. The anion binding constants and the kinetics of both diastereomer interconversion and guest exchange were measured.
Adaptation is a feature displayed by living systems; biological processes and organisms change their behaviour in an environmentally dependent manner. For example, protein folding activates or deactivates enzymatic processes while organisms respond to climatic changes in order to survive.9 Chemical systems have also been demonstrated to adapt to external stimuli, either on the molecular level through physical changes,10,11 or on the system-wide level with the expression of chemically distinct receptors after the addition of templating guest molecules; amplification in dynamic combinatorial systems6,11 or allosteric binding12 are two such examples. There are, however, very few examples of synthetic systems that produce multiple responses to stimuli at the same time, a feature of biological systems.1,2,8 Here we present a new class of synthetic system, composed of interconverting diastereomers of a self-assembled cage molecule that adapts on the molecular and system-wide levels, to the presence of suitable guest molecules, producing synergistic anion binding.
Based on our experience13–15 with the design and synthesis of self-assembled metallo-supramolecular capsules16 containing reversible metal–imine bonds,17,18 we constructed a new anion-binding [Fe4L6]8+ cage that exists as a mixture of three diastereomers. Upon the addition of different anions, these cages adapt on the system-wide level, reorganising themselves to express a new combination of diastereomers to encapsulate the anionic guests; a higher proportion of guests are bound by the new mixture than would be possible with the original mixture, suggesting that the anion binding is a characteristic of the system rather than individual molecules. Not only do the cage diastereomers interconvert, the volume of the individual cages adapts physically through the rotation of bonds, providing a tailored binding pocket for the guest lined with hydrogen-bond donors.
Fig. 1 Self-assembly of the tetrahedral [Fe4L6]8+ cage complex, 1. (a) Six equivalents of 4,4′-diaminobiphenyl react with twelve equivalents of 2-formylpyridine and four equivalents of an iron(II) salt in acetonitrile at 50 °C overnight to produce 1. Suitable iron(II) salts include Fe(NTf2)2, Fe(OTf)2, Fe(ClO4)2·6H2O, Fe(PF6)2·6MeCN and Fe(BF4)2·6H2O producing [1]·8NTf2, [1]·8OTf·7H2O, [1]·8ClO4·12H2O, [1]·8PF6·3H2O and [1]·8BF4·3MeCN respectively in 80–90% yields after precipitation with diethyl ether and chromatography as appropriate. (b) Schematic view of the X-ray crystal structure of [1]·8NTf2·5MeCN·2H2O·Et2O. The complex crystallises as the S4 diastereomer (1-S4) with two Λ and two Δ metal centres. Average Fe–Fe distances are 12.5 Å and the calculated void volume is 65 Å3. (c) Schematic representation of the three diastereomers of 1. Purple and blue metal centres are of opposite stereochemical configuration, green lines represent anti-linkages between homochiral metal centres and red lines represent syn-linkages between heterochiral metal centers. 1-T has one magnetically equivalent environment per atom labelled in (a); 1-C3 has four such environments; 1-S4 has three such environments. |
Fig. 2 1H NMR spectra (400 MHz, CD3CN, 298 K) of 1 with different anions. For clarity, only peaks corresponding to T-symmetric diastereomers are labelled in red (see Fig. 1 for labelling scheme), blue dots indicate peaks corresponding to 1-S4, green dots indicate peaks corresponding to 1-C3. Where small peaks are unlabelled they correspond to less than 1% relative abundance and are within the error of the measurements. (a) [1]·8NTf2; 1-T, 1-S4 and 1-C3 are observed in a 32:49:19 ratio. (b) [1]·8OTf; 1-T, 1-S4 and 1-C3 are observed in a 32:49:19 ratio. (c) [PF6⊂1]·7PF6; 1-T, 1-S4 and 1-C3 are observed in a 59:29:15 ratio. (d) [ClO4⊂1]·7ClO4; 1-T, 1-S4 and 1-C3 are observed in a 68:8:24 ratio. (e) [BF4⊂1-T]·7BF4; only 1-T is observed. (f) [NO3⊂1-T]·7NTf2 formed from the addition of one equivalent of tetrabutylammonium nitrate to [1]·8NTf2. (g) [Br⊂1-T]·7NTf2 formed from the addition of one equivalent of tetrabutylammonium bromide to [1]·8NTf2. (h) [I⊂1-T]·7NTf2 formed from the addition of one equivalent of tetrabutylammonium iodide to [1]·8NTf2. |
The synthesis of the tetrahedral cage 1 was also possible when other iron(II) salts were employed. Thus, [1]·8OTf, [1]·8ClO4, [1]·8PF6 and [1]·8BF4 could be prepared from the corresponding iron(II) salts in similar yields to that of [1]·8NTf2. The 1H NMR spectra of [1]·8OTf, [1]·8ClO4 and [1]·8PF6 again displayed the presence of diastereomers, while within the limits of detection, [1]·8BF4 gave exclusively peaks corresponding to the T-symmetric diastereomer, 1-T (Fig. 2). In [1]·8OTf, the T:S4:C3 ratio was 32:49:19, the same as observed for [1]·8NTf2, however, for the perchlorate analogue the ratio was 64:8:28 and in [1]·8PF6 it was 59:26:15, suggesting that the system of cage molecules was adapting to the presence of different anions.
The 19F NMR spectrum of [1]·8OTf showed only one sharp peak consistent with unencapsulated OTf−, while [1]·8BF4 gave two peaks in a 1:7 ratio suggesting that a tetrafluoroborate anion had been encapsulated inside 1. Similarly, [1]·8PF6 produced peaks for both free and bound PF6−. The bound PF6− gave three signals, corresponding to hexafluorophosphate anions encapsulated in each of 1-T, 1-S4 and 1-C3 which were present in the same ratio observed in the 1H NMR. 1H–19F HOESY of [1]·8BF4 and [1]·8PF6 (see ESI†) and crystal structures of the perchlorate, tetrafluoroborate and hexafluorophosphate species (Fig. 3) confirmed encapsulation of these anionic guests within the tetrahedral host.
Fig. 3 X-ray crystal structures of host–guest complexes. The approximately T-symmetric host structures are shown as schematic representations and the encapsulated guests as space-filling representations. The ΔΔΔΔ-enantiomer of each cage is shown; both enantiomers are present in each crystal structure. (a) [BF4⊂1-T]·7BF4·5.25MeCN·0.25Et2O, average Fe–Fe separation 12.6 Å and calculated volume 69 Å3. (b) [ClO4⊂1-T]·4ClO4·3OTf·4MeCN·2Et2O, average Fe–Fe separation 12.6 Å and calculated volume 74 Å3. (c) [PF6⊂1-T]·7PF6·9MeCN·3Et2O, average Fe–Fe separation 12.7 Å and calculated volume 87 Å3. (d) [I⊂1-T]·3I·3.5OTf·0.5NTf2·2MeCN·Et2O, average Fe–Fe separation 12.6 Å and calculated volume 66 Å3. |
The addition of either PF6−, BF4− or ClO4− to a solution of either of the empty cages ([1]·8NTf2 or [1]·8OTf) produced new peaks and/or changes in the diastereomer ratios present in the 1H NMR (see Table 1), indicating a system-wide response to their presence. Further anionic and neutral guests were also screened for potential binding in CD3CN (see ESI†). Nitrate, iodide, bromide, chloride and hexafluoroarsenate were observed to bind inside 1, and each induced changes in the corresponding diastereomer ratios (Fig. 2); the addition of small organic guests and larger anions did not induce changes in 1's 1H NMR spectrum. Other anions including sulfate and phosphate resulted in a colour change and precipitation, suggesting decomposition of the cage.
Anion | Anion volume (Å3) | Ka (M−1) | Diastereomer ratio (T:S4:C3) | ||
---|---|---|---|---|---|
KTG | KSG | KCG | |||
a Determined by 19F NMR.b Binding not observed.c Determined by competitive titration with the PF6− adduct.d Addition of more than two equivalents of guest resulted in cage decomposition.e Determined by competitive titration with the Br− adduct. | |||||
NTf2− | 156 | 0 | 0 | 0 | 32:49:19 |
OTf− | 85 | 0 | 0 | 0 | 32:49:19 |
SbF6− | 85 | 0 | 0 | 0 | 32:49:19 |
AsF6− | 80 | 1.5a | b | b | 32:49:19 |
PF6− | 75 | 1.00(5) × 103 | 0.29(2) × 103 | 0.43(4) × 103 | 59:26:15 |
Cl−d,e | 24 | 2.5(6) × 105 | b | b | 100:0:0 |
Br−c,a | 28 | 5.6(11) × 105 | b | b | 100:0:0 |
ClO4−e | 55 | 1.1(3) × 106 | 0.18(2) × 106 | 0.81(4) × 106 | 64:8:28 |
NO3−d,e | 41 | 1.2(4) × 106 | b | b | 100:0:0 |
BF4−e | 53 | 2.2(5) × 106 | b | b | 100:0:0 |
I−e | 35 | 1.7(4) × 107 | b | b | 100:0:0 |
Not only are the empty diastereomers of 1 (1-T, 1-S4 and 1-C3) in equilibrium in solution, each may also bind an anionic guest (G), with the corresponding host–guest complexes ([G⊂1-T], [G⊂1-S4] and [G⊂1-C3]) also in equilibrium with each other. The resulting network of equilibria in a solution of 1 is represented schematically in Fig. 4a. The equilibrium constants and diastereomer ratios could be determined from NMR experiments and are given in Table 1 (see the ESI† for the derivation of the binding model).
Fig. 4 The network of species present in 1 after the addition of an anionic guest and calculated responses of the system. (a) Each of the three diastereomers of 1 (1-T, 1-S4 and 1-C3) are in equilibrium with each other; these relationships are governed by the equilibrium constants KTC, KTS and KSC. Encapsulating an anionic guest, G, by 1-C3 produces [G⊂1-C3], encapsulation of G by 1-T produces [G⊂1-T] and encapsulation of G by 1-S4 produces [G⊂1-S4]. These equilibria are governed by the equilibrium constants KCG, KTG and KSG, respectively. The host–guest complex diastereomers are also in equilibrium with each other with the relationships governed by the equilibrium constants K′TC, K′TS and K′SC. (b) Calculated (see the ESI†) change in the total fraction of each diastereomer upon addition of PF6−. 1-C3 stays almost constant, and 1-T increases at the expense of 1-S4. (c) Calculated (see the ESI†) change in the total fraction of each diastereomer upon addition of ClO4−. 1-C3 and 1-T increase at the expense of 1-S4. (d) Calculated (see the ESI†) change in the total fraction of each diastereomer upon addition of BF4−. 1-T increases at the expense of both 1-S4 and 1-C3. |
The information obtained from these experiments can be used with the model of equilibria (Fig. 4a) to produce a full description of the network of 1 and the way it responds to the influence of different external stimuli (anions). This description allows for speciation curves to be derived, which describe the relationships between each component of the system at varying guest concentrations. The diastereomeric adaptation of the system (measured as fractions of the total concentration of 1) upon the addition of various anions are given in Fig. 4b, 4c and 4d. In each case a different response to each anion is observed. The addition of PF6− to 1 (Fig. 4b) produced little change in the fraction of 1-C3, while 1-T increased at the expense of 1-S4. In contrast, upon the addition of ClO4− (Fig. 4c), both 1-T and 1-C3 increased and 1-S4 was disfavoured. Adding BF4− instead resulted in the exclusive formation of 1-T (Fig. 4d). The system is thus capable of responding in various ways depending on the size and shape of the added guest.
The largest anion to be encapsulated within 1 was hexafluoroarsenate which has a volume of 80 Å3, but was only weakly bound (Ka = 1.5 M−1). The larger hexafluoroantimonate (85 Å3) was not encapsulated. The binding constants obtained for the smaller anions were substantially higher (all greater than 103 M−1 affinity). In general, guests that have symmetry elements that match those of the T-symmetric host were bound more strongly.
The halides were found to bind in size order, with iodide producing the highest binding constant (1.7(4) × 107 M−1) of the guests explored; 1 is 30 times more selective for iodide than the other halides. Cage 1 represents the strongest 1:1 iodide binding host of which we are aware in either synthetic23 or natural systems.24 Indeed, this represents the strongest 1:1 host–guest binding for a metal–organic host that can be isolated in the absence of a guest.14,25
The combination of strong binding and adaptive diastereomeric response upon the addition of a guest leads to a significant enhancement of anion binding. In each of the cases where 1-T was expressed exclusively upon guest addition (BF4−, NO3− and halides), the quantity of guest encapsulated is amplified by more than 200% compared to the hypothetical guest binding of the system if the ratio of diastereomers remained at the pre-addition values of 32:49:19. The system of cages responds as a network to the addition of certain anionic guests, reconfiguring to synergistically bind anions in a manner that would not be possible without such a response.
The crystal structures of the host–guest complexes (Fig. 3) suggest that, in addition to electrostatics, the ability of 1 to bind anions originates from non-classical hydrogen bonding.26 In [BF4⊂1-T] and [ClO4⊂1-T] the tetrahedral guests are arranged such that each of their fluorine and oxygen atoms, respectively, point towards the faces of the cage to take advantage of three stabilising, symmetrically arranged hydrogen bonding (Ha⋯guest) interactions for each atom, providing a total of twelve interactions per anion in the range of 2.3–2.6 Å. In order to accommodate the perchlorate anion, which is slightly larger than tetrafluoroborate, the volume of the cage molecule expands from 69 Å3 to 74 Å3. This is achieved by adaptation of the cage through the expansion of the dihedral angle between the phenyl rings and the Nimine–Fe–Npyridyl chelate planes (that is, the rotation of the ligand phenyl ring around the C4–Nimine bond), which can be measured by the C3–C4–Nimine–Cimine torsion angle (φ). This expansion can occur while conserving idealised T-symmetry and maintaining the biphenyl dihedral at its optimum angle (55°).15
This rotation has several potential consequences. As φ approaches 0°, the phenyl ring and the pyridyl-imine approach coplanarity, bringing Ha and Hb (Fig. 1) away from the centre of the cage, increasing its cavity volume. As φ approaches 90°, the converse occurs; Ha and Hb move towards the centre of the cage, decreasing its volume. This flexibility allows cage 1-T to adapt to the size of a potential guest, while also allowing the Ha protons to be orientated to act as H-bond donors for a specific guest. The ideal angle for φ is approximately 69 ± 9°,15 which allows for Hb and Hc to avoid an energetically disfavourable transannular eclipse and also produces stabilising CH⋯π interactions between Hb and neighbouring phenyl rings.15,27 These stabilising interactions are confirmed by 1H NMR, with Hb experiencing significant shielding due to its proximity to the neighbouring phenyl rings (Fig. 2); the degree of the chemical shift change observed correlates with the strength of anion binding. A change in φ from its ideal angle is also associated with an enthalpic cost. As φ increases, stabilising CH⋯π interactions are broken. Additionally, as φ decreases, destabilising steric Hb–Hc clashes are introduced. The differences in binding constants between anions can be attributed to the degree to which the energetically disfavoured arrangements are adopted by 1-T in order to accommodate the guest with a favourable hydrogen-bonding arrangement. This hypothesis also explains the amplification of the 1-T diastereomer upon guest binding. Only this diastereomer can adopt both a symmetric ring of CH⋯π interactions and position its protons in a suitable manner to stabilise guest molecules without the energetic penalties arising from the steric proximity of Hb and Hc or between two neighbouring biphenylene Ha protons.15
Compared to tetrafluoroborate and perchlorate, hexafluorophosphate is both larger and does not have an ideal symmetry match to the cavity of 1-T. To accommodate this guest, the volume of the cage further expands to 87 Å3. In the crystal structure of [PF6⊂1-T] (Fig. 3c), four of the fluorine atoms point towards the faces, benefiting from hydrogen bonding interactions, and the remaining two point towards the vertices of the tetrahedron.
In contrast, iodide is significantly smaller than BF4−, ClO4− or PF6−, but it is also bound the most strongly by 1-T. In the X-ray structure, the 1-T cage was observed to enclose a volume of 66 Å3, only slightly larger than the volume of 65 Å3 observed in the structure of 1-S4 (1·8NTF2). The Ha protons are arranged to provide an almost spherical arrangement of twelve stabilising hydrogen bond donors around the guest, both matching its symmetry and providing an ideal binding pocket24,18, which is reflected in the strength of binding. The lower binding constants of the other halides can be rationalised by considering that smaller guests will increase the Ha⋯X− distances resulting in weaker interactions.
Further supporting the hypothesis relating guest binding affinity to hydrogen bonding and stabilising CH⋯π interactions, we observed no correlation between host–guest fill ratios and the strength of binding. In the solid state, BF4− occupies 80% of the available volume of 1-T, ClO4− occupies 75%, PF6− occupies 85% and I− occupies 53%, suggesting that in the presence of hydrogen bonding and electrostatic interactions, volume-occupation ratios are not a good predictor of guest binding affinity.28
To further investigate the ability of 1-T to adapt upon guest binding, we undertook semi-empirical calculations to correlate φ with the volume of 1-T. PM3 minimisations were performed with φ values of 0–90° (see ESI†). As expected the minimum volume (58 Å3) was found at 90°. At 0° a volume of 203 Å3 was obtained, with this structure representing the highest energy of those modelled. Thus, by changing φ the host is potentially capable of expanding its enclosed volume by 250%.
A volume of 203 Å3 suggests that larger anions such as, OTf−, NTf2− and SbF6− should be able to fit inside the cavity, however, they were not observed to do so. This can be explained by the energetic penalty (steric clashes and breaking CH⋯π interactions) that the cage must pay to adapt to guests with larger volumes. The magnitude of the larger anions' binding enthalpy is therefore not sufficient to overcome the energetic penalty of rearrangement, an effect which is exacerbated as the Ha protons rotate away from the cavity, resulting in less favourable arrangements of hydrogen bond donors and leaving only dispersion forces to stabilise any host–guest complex. It therefore appears to be more energetically favourable for the system of 1 to remain empty than adapt to bind any of these larger anions.
The kinetics of guest exchange and diastereomer interconversion were examined by NMR. The rate constants and the corresponding activation energies are summarised in Table 2. The measurement of these values was facilitated by three observations. Firstly, the exchange of guest and unencapsulated tetrafluoroborate molecules in [BF4⊂1-T]·7BF4, while slow on the T2 timescale, was fast enough to give rise to exchange peaks in 19F EXSY NMR experiments. This allowed the determination of the rates of tetrafluoroborate exchange.
System | Process | k (s−1) | ΔG‡ (kcal mol−1) |
---|---|---|---|
a Determined by 19F–19F EXSY NMR.b Determined by 19F NMR.c Estimated by the disappearance of free PF6− and represents a value for the inclusion of the guest into the total system of cages rather than a specific diastereomer.d Determined by 1H NMR. | |||
[BF4⊂1-T]7+a | kin | 0.020 ± 0.002 | 19.75 ± 0.06 |
kout | 0.148 ± 0.015 | 18.56 ± 0.06 | |
[PF6⊂1]7+b | kinc | 0.0006 ± 0.0001 | 21.81 ± 0.1 |
k′TC | 11 ± 1 × 10−6 | 24.19 ± 0.05 | |
k′TS | 16 ± 1 × 10−6 | 23.97 ± 0.04 | |
k′SC | 1.0 ± 1.0 × 10−6 | 25.6 ± 0.6 | |
k′ST | 44 ± 3 × 10−6 | 23.97 ± 0.04 | |
k′CS | 1.4 ± 1.0 × 10−6 | 25.4 ± 0.4 | |
k′CT | 42 ± 2 × 10−6 | 23.40 ± 0.03 | |
“Empty” [1]8+d | kTC | 4.7 ± 0.8 × 10−6 | 24.7 ± 0.1 |
kTS | 10.0 ± 0.6 × 10−6 | 24.25 ± 0.04 | |
kSC | 17 ± 2 × 10−6 | 23.93 ± 0.07 | |
kST | 8.6 ± 0.5 × 10−6 | 24.33 ± 0.03 | |
kCS | 77 ± 7 × 10−6 | 23.04 ± 0.05 | |
kCT | 19 ± 3 × 10−6 | 23.87 ± 0.09 |
Secondly, in the 19F NMR of 1·8PF6 all three diastereomers could be observed. Given that it crystallised exclusively as [PF6⊂1-T]·7PF6, we were able to follow the conversion of [PF6⊂1-T] to the equilibrium [PF6⊂1-T]:[PF6⊂1-S4]:[PF6⊂1-C] ratio of 59:26:15 over time, permitting the calculation of the rates of diastereomer interconversion for [PF6⊂1]. The rate constants for each process were found to be similar, although substantially slower than the rate of tetrafluoroborate exchange.
The addition of PF6− to a solution of the empty cage, 1·8NTf2, also allowed us to estimate the rate of uptake of this guest into the host complex by 19F NMR, which was found to be slower than the corresponding process observed in the case of tetrafluoroborate.
Thirdly, the empty cage, 1·8NTf2, was observed to crystallise exclusively as 1-S4 equilibrating to a 1-T:1-S4:1-C3 ratio of 32:49:19 when redissolved, a process that could be monitored over several days. The rate constants for this process are very similar to the rates found for [PF6⊂1], suggesting that a similar mechanism for interconversion of diastereomers occurred in the presence or absence of a guest.
The kinetic data provide mechanistic insights into the dynamics of the system. The large differences in rate constants between guest exchange and stereoisomerisation suggests that separate mechanisms are responsible for the two processes. The slow rates of diastereomer interconversion are consistent with either the Bailar twist or de-ligation mechanisms that have been proposed for related cages,15,20,21,29,30 while the much faster rates of guest exchange suggests a through-face process.29 The latter is consistent with the observation that the larger PF6− is encapsulated 33 times more slowly than BF4−.
Interestingly, the barrier to direct conversion of T to S4 is similar to that of T to C3 in both the cases of [PF6⊂1] and empty 1. These data are inconsistent with a Bailar twist. The activation energy for the concerted double Bailar twist (required for T to S4) should be twice that for a single Bailar twist (T to C3). Such a high barrier precludes the concerted double twist, yet the T to S4 stereoisomerisation process is observed. Therefore we conclude that diastereomer interconversion takes place by a deligation process as the enthalpic cost associated with a sequential deligation need not be greater than that for single deligation.
Footnote |
† Electronic supplementary information (ESI) available: Experimental details of the synthesis and characterisation of products, details of the X-ray structural determinations and titrations, derivations of the binding model, details of the kinetics studies, molecular modelling and volume calculations. See DOI: 10.1039/c2sc21486e |
This journal is © The Royal Society of Chemistry 2013 |