Lei
Fang
a,
Mark A.
Olson
a,
Diego
Benítez
b,
Ekaterina
Tkatchouk
b,
William A.
Goddard III
b and
J. Fraser
Stoddart
*a
aDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208-3113, USA. E-mail: stoddart@northwestern.edu; Fax: +1 (847) 491-1009; Tel: +1 (847) 491-3793
bMaterials and Process Simulation Center, California Institute of Technology, 1200 E. California Blvd., Pasadena, California 91125, USA
First published on 18th November 2009
Mechanically bonded macromolecules constitute a class of challenging synthetic targets in polymer science. The controllable intramolecular motions of mechanical bonds, in combination with the processability and useful physical and mechanical properties of macromolecules, ultimately ensure their potential for applications in materials science, nanotechnology and medicine. This tutorial review describes the syntheses and properties of a library of diverse mechanically bonded macromolecules, which covers (i) main-chain, side-chain, bridged, and pendant oligo/polycatenanes, (ii) main-chain oligo/polyrotaxanes, (iii) poly[c2]daisy chains, and finally (iv) mechanically interlocked dendrimers. A variety of highly efficient synthetic protocols—including template-directed assembly, step-growth polymerisation, quantitative conjugation, etc.—were employed in the construction of these mechanically interlocked architectures. Some of these structures, i.e., side-chain polycatenanes and poly[c2]daisy chains, undergo controllable molecular switching in a manner similar to their small molecular counterparts. The challenges posed by the syntheses of polycatenanes and polyrotaxanes with high molecular weights are contemplated.
Lei Fang | Lei Fang was born in Poyang, China, in 1983. He received both his BS (2003) and MS (2006) degrees in chemistry from Wuhan University, China while carrying out research under the supervision of Professor Yong-Bing He. During his graduate studies, Lei spent one and a half years at the Hong Kong Baptist University in the laboratories of Professor Wing-Hong Chan studying cholesterol-derived molecular sensors. Presently, he is pursuing his PhD in chemistry with Professor Stoddart at Northwestern University. During his PhD program, Lei has conducted research on interdisciplinary topics in the broad areas of nanoscience as well as organic and polymer chemistry. |
Mark A. Olson | Mark A. Olson received his BS degree in chemistry from Texas A&M University-Corpus Christi in 2005. During his undergraduate studies, Mark experienced a short stay at the California Institute of Technology in the laboratories of Dr Jack Beauchamp. Presently a fourth year graduate student, after spending his first three years in UCLA, Mark is now finishing up his PhD in organic chemistry under the tutelage of Professor Stoddart at Northwestern University. His focus is on the synthesis of exotic molecular switchable materials including bistable side-chain poly[2]catenanes, reconfigurable (Au, Pt, Pd, Ag) nanoparticle assemblies, and reprogrammable self-assembling polymer blends. |
Diego Benítez | Diego Benítez obtained his PhD in chemistry from the California Institute of Technology in 2005 conducting research in the laboratories of Robert H. Grubbs and William A. Goddard III. He then joined the laboratory of J. Fraser Stoddart at UCLA as a Research Associate. In 2009, he returned to the California Institute of Technology as the Director of Nanomaterials Technology in the Materials and Process Simulation Center. |
Ekaterina Tkatchouk | Ekaterina Tkatchouk obtained her BS degree in chemistry from the Universidad Nacional Autonoma de Mexico and her PhD in Materials Science and Engineering from the same institution. After working for some time in industry, she spent a year in the laboratory of Kendall N. Houk at UCLA as a UC MEXUS-CONACYT Postdoctoral Research Fellow. Since late 2007, she has been a Postdoctoral Scholar in the laboratory of William A. Goddard III where she conducts computational research on topics related to homogeneous catalysis and mechanically interlocked molecules. |
William A. Goddard III | William A. Goddard III obtained his BS Engr. from UCLA in 1960 and his PhD in Engineering Science and Physics from California Institute of Technology (Caltech) in Oct. 1964. Since Nov. 1964, he has been a member of the Chemistry faculty at Caltech where he is now the Charles and Mary Ferkel Professor of Chemistry, Materials Science, and Applied Physics. His current research interests include new methodology for quantum chemistry, reactive force molecular dynamics, mesoscale dynamics, statistical mechanics and electron dynamics. |
J. Fraser Stoddart | Fraser Stoddart received all (BSc, PhD, DSc) of his degrees from the University of Edinburgh, UK. Presently, he holds a Board of Trustees Professorship in the Department of Chemistry at Northwestern University. His research has opened up a new materials world of mechanically interlocked molecular compounds and, in doing so, has produced a blueprint for the subsequent growth of functional molecular nanotechnology. |
Fig. 1 Examples of natural and artificial macromolecular mechanically interlocked systems: (a) graphical representation of catenated circular DNA; (b) crystal structure of the bacteriophage HK97 capsid chainmail,6 with the subunits that are cross-linked into rings colored identically (reprinted with permission of the American Association for the Advancement of Science); (c) the construction of an interpenetrating polymer network at a conceptual level; (d) crystal structure of MOF-14 showing10 a pair of interwoven 3-D porous frameworks (reprinted with permission of the American Association for the Advancement of Science). |
Rapid advances in the synthesis of mechanically-interlocked molecules (MIMs) in recent times have enabled11 precise control of the architectures and topologies of the molecules; hence, it has enabled chemists to develop specifically desired functions based on these unique structures. For example, the application of switchable, mechanically-interlocked, small molecules has been widely demonstrated12–14 in solid-state electronic devices,12 mechanised nanoparticles,13 and nanoelectromechanical systems.14 From the perspective of an organic chemist, engineering mechanical bonds into macromolecular scaffolds, by employing organic/polymer synthetic protocols, is becoming an area of considerable contemporary interest. The importance and attraction of this field originated from the fact that (i) device fabrication of MIMs could benefit enormously from the macromolecular materials’ processability if the mechanically interlocked structures could be incorporated into polymer/dendrimer scaffolds, and (ii) intramolecular motion of mechanically interlocked units in a polymer/dendrimer network could induce accumulative, macroscopic property changes in the material itself. In fact, back in the early 1990s, soon after the field of MIMs began to blossom, the development of polyrotaxanes and polycatenanes—the two most commonly sought after mechanically interlocked macromolecules—had been identified by synthetic chemists as important and challenging targets in synthesis. The challenge is a combination of the difficulty in preparing MIMs themselves and the huge entropy cost in making high-molecular weight macromolecules. It has to be admitted that mechanically bonded macromolecules with precisely controlled structures are not yet producible on a gram scale, using routine synthetic procedures. Several comprehensive review articles have been published15–17 on mechanically bonded macromolecules, in addition to the extensive review literature11,18 now available on MIMs themselves.
This review describes our own efforts—covering the past two decades—to introduce mechanical bonds into macromolecules. Fig. 2 shows the graphical representations of some of these mechanically interlocked macromolecules—namely, (a) main-chain [n]catenanes, (b) bridged poly[2]catenanes, (c) pendant poly[2]catenanes, (d) side-chain poly[2]catenanes, (e) main-chain [n]rotaxanes, (f) linear poly[c2]daisy chains, as well as (g) mechanically interlocked dendrimers. Various approaches and synthetic strategies have been employed in synthesising these constitutionally and topologically diverse targets. In general, these synthetic strategies can be described under three categories: (I) The formation of the mechanical bonds spontaneously while constructing the macromolecular scaffold; (II) Coupling already-made MIMs on to polymer/oligomer/dendrimer scaffolds by covalent linking; (III) Incorporating mechanical bonds onto an already existing polymeric/oligomeric/dendritic scaffold.
Fig. 2 Architectures of the mechanically bonded macromolecules that are covered in this review: (a) main-chain [n]catenanes; (b) bridged poly[2]catenanes; (c) pendant poly[2]catenanes; (d) side-chain poly[2]catenanes; (e) main-chain [n]rotaxanes; (f) linear poly[c2]daisy chains; (g) mechanically interlocked dendrimers. |
Scheme 1 Assembly of the main-chain oligocatenanes 14+, 212+ and 320+. |
Fig. 3 Space-filling representations of the solid-state structures of (a) 14+, (b) 212+ and (c) 320+. |
In an attempt to uncover higher synthetic efficiencies in the synthesis of the main-chain [n]catenane, a one-pot self-assembly process of oligocatenanes, using a similar template-directed strategy, gave22 linear [2]-, [3]-, [5]-, and [7]catenanes in yields of 5, 14, 4, and 2%, respectively. In this reaction, a tetra-1,5-naphtho[76]crown-20 was employed as the macrocyclic template, while the cyclobis(paraquat-O,O′-diethylenehydroquinone) rings were formed along the DNP units in the templates, to generate the oligocatenated topologies in one pot. Since the [5]- and [7]catenanes isolated in this reaction could only be characterised by electrospray mass spectroscopy, their topologies and molecular structures remain uncertain. Higher oligocatenanes could not be detected and perhaps did not form on account of their entropically unfavourable nature. Although the use of ultrahigh pressure and the careful choice of templates greatly facilitates the self-assembly of the oligocatenanes, both the stepwise and the one-pot methods suffer from low yields, hampering the further development of higher molecular weight main-chain poly[n]catenanes.
It is apparent that a much more efficient synthetic strategy is necessary for the generation of higher oligo- or polycatenanes using a “bottom-up” approach. Recently, a success23 in highly efficient thermodynamic assembly of a [3]catenane has raised our hopes of making main-chain polycatenanes in high yields. In this reaction, the cyclobis(paraquat-4,4′-biphenylene) ring underwent dynamic nucleophilic substitution in the presence of iodide (I−) at high temperature (>80 °C) which allowed24 the ring to open and close reversibly. This dynamic process enabled formation of the thermodynamically stable product when a π-electron-rich dinaphtho[38]crown-10 macrocycle served as a template in solution, to afford the expected [3]catenane, in 91% yield! It is not difficult to believe that a high molecular weight polycatenane can be obtained by assembling cyclobis(paraquat-4,4′-biphenylene) with a carefully designed π-electron-rich macrocycle in a similar thermodynamically controlled reaction. In principle, under exchange conditions (I− and T > 80 °C) the system should behave as a step-growth dynamic covalent polymerisation.25 The driving force for the polymerisation will be the enthalpic release from the donor–acceptor catenation step. As with every dynamic polymerisation,26 a ring-chain equilibrium will be established and will be highly dependent of the concentration of the reactants. In the wake of these considerations, a reasonable approach would be to use exact molar amounts of reactants with saturated solutions of the highly soluble electron-poor and electron-rich macrocycles. Lower monomer concentrations might shift the ring-chain equilibrium towards cyclic oligomers, producing cyclic oligocatenanes.
Scheme 2 Structural formulae and cartoon representations of the bridged polycatenanes 4·8nPF6, 5·8nPF6, 6·4nPF6 and the pendant polycatenane 7·4nPF6. |
Scheme 3 (a) Synthesis and redox-driven switching of sidechain poly[2]catenane 8·4nPF6; (b) Graphical representation showing the switching process of the polymer 8·4nPF6. |
Fig. 4 (a) An early example of an α-cyclodextrin-based polyrotaxane, in which a polyethylene glycol chain acts36 as the thread and the terminal 2,4-dinitrophenyl groups serve as the stoppers; (b) the chemical structure of a β-cyclodextrin-threaded conjugated polyrotaxane which represents37 an insulated molecular wire. |
Formation of dynamic covalent bonds were employed40 in the “clipping” approach to afford main-chain oligorotaxanes. The reversible nature of the reactions introduces41 the prospects of “error checking” and “proof-reading” into synthetic processes where dynamic covalent chemistry operates. Since the formation of products occurs under thermodynamic control, product distributions depend only on the relative stabilities of the final products. The dynamic covalent chemistry involved in this system, is the highly efficient clipping42 of a diamine and a dialdehyde to form a [24]crown-8 diimine-containing macrocycle, templated by a dialkylammonium ion recognition site. In this event (Scheme 4a), oligomeric dialkylammonium stalks terminated by two bulky dimethoxybenzyl stoppers were used as the dumbbell components for the oligorotaxane and as the template for the thermodynamic controlled clipping reaction. A series of such dumbbell-shaped molecules (9a–9f·nPF6), containing n (n = 2, 3, 4, 6, 10, 14, respectively) dialkylammonium centres, were synthesised firstly by reductive amination with derivatives of benzylamine and benzaldehydes, then by protonation of the secondary amines followed by counterion exchange. The clipping reaction of 2,6-dipyridinedicarboxaldehyde and tetraethyleneglycol bis(2-aminophenyl)ether, templated by 9a–9d·nPF6, gave the thermodynamically stable [3]-, [4]-, [5]-, and [7]rotaxanes (10a–10d·nPF6) in nearly quantitative yields in MeNO2. When n = 10, 4-octyloxypyridinedicarboxaldehyde was used, in a 21-component self-assembly reaction in order to improve the solubility of the product 10e·nPF6. Although the assembly of the [11]rotaxane 10e·nPF6 was successful using the octyloxyl solublising group, unfortunately, when n = 14, the poor solubility of the dumbbell 9f·nPF6 (n = 14) once again put a limit on the template-directed synthesis of the [15]rotaxane, using this approach. On account of their readily hydrolysable imine bonds, the dynamic rotaxanes 10a–10c·nPF6 can be fixed by reducing with BH3·THF. The fixed [3]-, [4]-, and [5]rotaxanes 11a–11c·nPF6 were isolated pure in 77, 74, and 40% yields, respectively. For the higher rotaxanes 10d–10e·nPF6 (n = 6, 10), the reductions were unsuccessful on account of partial cleavage and dissociation of the macrocycles from the dumbbell compounds during the reductions.
Scheme 4 (a) The “clipping” approach to the dynamic oligorotaxanes 10a–e·nPF6, and the “fixed” oligorotaxanes 11a–c·nPF6. The architectures of these structures are shown as cartoon graphs on the right. The dynamic covalent bonds are highlighted in orange. (b) Molecular dynamics simulation for the folding of the oligorotaxane 1014+ into a random coil conformation. The backbone is shaded blue, while alternating rings (for visual ease) are colored red and orange. |
Meanwhile, a “threading-followed-by-stoppering” approach had also been developed (Scheme 5) for the synthesis of the π-donor–acceptor polyrotaxane 13·4nPF6. Such [n]rotaxanes are composed of a linear polymeric dumbbell-shaped component, containing 2n DNP units as the π-electron donors, and n CBPQT4+ cyclophanes as the π-electron acceptors. The template-directed “clipping” methodology used traditionally for the installation of CBPQT4+ rings on dumbbell-shaped components was not efficient enough to obtain a high coverage of rings encircling the polymer backbone. Instead, the “threading-followed-by-stoppering” method, employing Cu(I)-catalysed Huisgen 1,3-dipolar cycloadditions, proceeded43 with high efficiencies when it was applied in the kinetically controlled synthesis of [2]rotaxanes. In this context, the dumbbell component 12 was fed with 0.6 equivalents of CBPQT4+ rings relative to the total number of DNP units of 12 in DMF. After 24 h at room temperature, UV/vis absorption spectroscopy demonstrated that the slow threading process had gone to completion. The resulting pseudopolyrotaxane [12 ⊂ nCBPQT4+] could be converted to the polyrotaxane 13·4nPF6 by covalently attaching two bulky 2,6-diisopropylphenyl stoppers onto the chain ends using Cu(I)-catalysed Huisgen 1,3-dipolar cycloadditions. The Mn value of 13·4nPF6 reached up as far as 181 kDa with a PDI of 1.71, indicating that there were about 94 repeating units in each polymer chain on average. The polyrotaxane 13·4nPF6 adopts an alternating stacking pattern that is induced by secondary noncovalent bonding interactions between the CBPQT4+ rings and the alongside DNP units, an interaction which was confirmed by 1H NMR spectroscopy by comparing the polyrotaxanes with a simple folded [2]rotaxane model compound. Another piece of evidence for the folding of 13·4nPF6 came from GPC investigation which showed that the polymer dimensions of poly[n]rotaxane 13·4nPF6 were significantly smaller than that of the polymer thread 12, despite a near two-fold molecular mass increase.
Scheme 5 Structural formulae and graphical representations of the “threading-followed-by-stoppering” approach to the self-folding polyrotaxane 13·4nPF6. |
In the absence of experimental data, the prediction of the secondary structure of higher order donor–acceptor [n]rotaxanes (n > 10) is a challenging exercise that awaits a thorough analysis. Although in the solid-state, infinite stacks are clearly possible (Scheme 5), in solution the dynamic nature of the oligomers/polymers is likely to produce a distribution of π-stack lengths in equilibrium with each other and dictated by the enthalpy gain for the stacking being weighed against the entropy loss associated with a rigid rod-like molecule. Quantum mechanical calculations carried out on short donor–acceptor stacks with the constitution of 13·4nPF6 support the folded secondary structure anticipated44 to exist in the solid state. The entropic contributions, however, from unfolding are difficult to predict in such large molecules—especially in solution. Indirect data—1H NMR spectroscopy and hydrodynamic radii—on 134n+ revealed evidence for a folded structure. It is not easy, however, to establish the most abundant length of π-stacks—analogous to the conjugated critical length in polyacetylenes45—using NMR spectroscopy and light scattering alone. Molecular mechanical simulations—although attractive for their performance/cost for large systems—fail to describe the sandwiched, parallel, displaced and T-shaped geometries of stacked and interacting aromatics with the required accuracy to make their use applicable to large systems.
On the other hand, in systems such as 10n+ and 11n+ where [π⋯π] stacking is not a primary consideration, the secondary structures can be modeled using popular force-fields,46 such as the recent OPLS-2005. We have performed a combination of molecular dynamics (MD) and molecular mechanics (MM) minimisations using the OPLS force-field on a [15]pseudorotaxane to obtain (Scheme 4b) a low energy conformation of an analogue of 10n+. The starting geometry for the [15]pseudorotaxane, which was generated from the X-ray crystal structure44 of the monomeric [2]pseudorotaxane, was subjected to a 2000 ps MD simulation at 300 K until an equilibrium geometry was obtained. This analogue is predicted to adopt a random coil conformation similar to that of the dumbbell 9f14+, indicating that the installment of the polyether macrocycles on the dumbbell molecules exerts little significant effect on the secondary structure in such hydrogen-bond interaction-based systems, compared to the [π⋯π] stacking-based structure 134n+.
As an entropically unfavourable process, the assembly of a high molecular weight daisy chain polymer requires a high concentration of the monomer and a strong binding affinity between the complementary recognition sites. Although supramolecular [an]daisy chains, composed with two carefully designed heteroditopic monomers, were reported49 recently, the mechanically interlocked macromolecular [an]daisy chain still remains a challenging synthetic target. Instead, cyclic dimers ([c2]daisy chains) were easy to isolate in high yields and show remarkable stabilities both in solution and in the solid-state. On account of its C2h symmetry and linear intramolecular motion, the [c2]daisy chain topology has been employed50,51 in the construction of molecular actuators that have been designed to undergo contraction/extension molecular movements with the appropriate stimuli, e.g., redox and pH. Consequently, the polymerisation of [c2]daisy chains became an obvious target in attempts to create muscle-like materials (Fig. 2f). In this context, a bifunctional [c2]daisy chain was synthesised52 by employing ring-closing metathesis of a bisolefin-containing polyether templated by a dialkylammonium unit, on account of the strong hydrogen bonds formed between oxygen atoms in the macrocyclic component and the NH2+ cationic centre. This bisalkene-functionalised compound was subjected to acyclic diene metathesis polymerisation in the presence of Grubbs’ ruthenium catalyst, to afford a poly[c2]daisy chain with a molecular weight of 13 kDa.
Encouraged by the success in polymerising a [c2]daisy chain, we have synthesised53 a bistable [c2]daisy chain 14·6PF6 functionalised with dialkyne terminal groups (Scheme 6). The macrocycle is dibenzo[24]crown-8 (DB24C8) and it has two recognition sites—a dialkylammonium centre and a bipyridinium unit—appended to it. The mutual association of two of these components during templation prior to stoppering leads to the formation of the [c2]daisy chain. In MeCN, the DB24C8 ring encircles the dialkylammonium unit predominantly as a result of [N+–H⋯O] hydrogen bonding primarily. Since the binding preference of the DB24C8 ring can be reversed by deprotonation of the dialkylammonium ion with a non-nucleophilic base and restored by reprotonation of the resulting dialkylamine function with acid, the daisy chain molecule can be made to undergo acid/base controlled contraction/extension movements. Functionalised with bisalkynyl terminal groups, 14·6PF6 was subjected to an AA + BB type step-growth polymerisation with an appropriate diazide employing the highly efficient Huisgen type 1,3-dipolar cycloaddition.34 The resulting bistable poly[c2]daisy chain 15·6nPF6 has a molecular weight of 33 kDa and a polydispersity of 1.85. This polymer undergoes a similar quantitative, efficient and fully reversible switching process in solution, triggered by base/acid, i.e., DABCO and trifluoroacetic acid. Stop-flow kinetics measurements demonstrated that the extension/contraction movements of 15·6nPF6 are actually faster than those occurring in its monomeric counterpart 14·6PF6. Although the reason for this phenomenon is still not clear to us, the robust switchability of the polymer provides an avenue through which correlated molecular motions can lead to changes in macroscopic properties. In a similar system reported54 recently, the radius of gyration of a poly[c2]daisy chain polymer was observed to increase by 48% after expansion of the [c2]daisy chain monomer had been induced by external stimuli.
Scheme 6 Polymerisation of the bisfunctionalised [c2]daisy chain 14·6PF6 and the acid–base induced switching of the resulting poly[c2]daisy chain 15·6nPF6. Graphical representations of the switchable poly[c2]daisy chain architectures are shown. |
Not until the dynamic covalent chemistry mentioned42 above was employed, was the assembly of Fréchet-type dendrons into a mechanically interlocked dendrimer achieved57,58 in a nearly quantitative yield (Scheme 7). In order to construct the mechanically interlocked dendritic structure, a tripod molecule 16·3PF6, with three dialkylammonium unit containing arms, was used as the dendrimer core. 1H NMR spectroscopy reveals >90% conversion 5 min after the dendritic dialdehydes ([G1]–[G3]), the diamine, and 16·3PF6 were mixed in a 3:3:1 molar ratio in CD3CN or CD3NO2. This seven-component dynamic assembly process shows remarkable efficiency and the mechanically interlocked dynamic products 17a–17c·3PF6 are thermodynamically stabilised by numerous [N+–H⋯O] hydrogen bonds and [C–H⋯O] interactions. The formation of 17a–17c·3PF6 could be monitored not only by 1H NMR spectroscopy quantitatively, but also by electrospray ionisation mass spectrometry, since the molecular ions [17a]3+, [17b]3+, and [17c]3+ can be clearly identified. In common with the oligomeric rotaxanes 10a–10e·nPF6, the kinetically labile dendrimers 17a–17b·3PF6 ([G1]–[G2]) could be fixed by reduction (BH3·THF), followed by deprotonation to give the kinetically stable, neutral dendrimers 18a–18b·3PF6, in ∼80% yields. These reductions went smoothly with [G1]–[G2], but yielded a mixture of degraded dendrimers with the [G3] dendron, as a result of the massive steric hindrance of the [G3] dendron. In summary, by taking advantage of dynamic covalent chemistry,41 we have developed a practical way of making mechanically bonded dendrimers, in which the thermodynamically controlled seven-component self-assembly process proceeds quantitatively by overcoming the massive steric hindrance of three dendrons as large as the [G3] Fréchet-type wedge-shaped dendrons.
Scheme 7 Dynamic covalent assembly of the mechanically interlocked dendrimers 17a–17c·3PF6 and the fixed dendrimers 18a–18b·3PF6. |
This journal is © The Royal Society of Chemistry 2010 |