Jean-Claude
Chambron
*a and
Jean-Pierre
Sauvage
b
aICMUB (UMR CNRS No. 6302), Université de Bourgogne, 9 Avenue Alain Savary, BP 47870, 21000 Dijon, France. E-mail: jean-claude.chambron@u-bourgogne.fr; Fax: +33 (0)3 80 39 61 17; Tel: +33 (0)3 80 39 61 16
bISIS, Université de Strasbourg, 8 Allée Gaspard Monge, 67000 Strasbourg, France. E-mail: jpsauvage@unistra.fr; Fax: +33 (0)3 68 85 51 40; Tel: +33 (0)3 68 85 51 43
First published on 22nd October 2012
Topological constructions made from closed curves range from simple links to intricate knots and started to capture the chemists' attention in the early sixties. These mathematical objects result from particular embeddings of a single or a set of closed curves in the three-dimensional space that show an infinite variety of presentations. Simple catenanes, higher order interlocked macrocycles, and molecular knots can be synthesized via the metal template approach, just as simple macrocycles. However, this requires that rigid presentations with appropriate geometrical characteristics be identified prior to molecular design, and those selected for the metal-templated synthesis of some of these fascinating molecules are reviewed here.
![]() Jean-Claude Chambron | Jean-Claude Chambron has been a CNRS Research Director of the University of Burgundy at Dijon since 2001. He received his PhD from the University of Strasbourg with Prof. Jean-Pierre Sauvage and worked in his group as a CNRS Research Associate on topological chirality and porphyrin-incorporating catenanes and rotaxanes. His current research interests concern the induction of chirality by non-covalent interactions and the self-assembly of chiral molecular capsules by dynamic covalent chemistry. |
![]() Jean-Pierre Sauvage | Jean-Pierre Sauvage is presently a Professor Emeritus and a CNRS Research Director Emeritus of the University of Strasbourg as well as a Distinguished Visiting Scholar at Northwestern University. With his group, he has contributed to the fields of molecular topology (knots and catenanes), rotaxanes and molecular machines. He is also interested in artificial photosynthesis and multicomponent systems combining porphyrins and ruthenium or iridium complexes. He was awarded the Izatt-Christensen Award in Macrocyclic Chemistry in 1991, the Prelog Gold Medal, ETH-Zürich (Switzerland) in 1994, the RB Woodward Award in Porphyrin Chemistry in 2008 and the Japan Coordination Chemistry Award in 2009 as well as numerous Named Lectureships. |
In fact, knotting and intertwining are very common but non-trivial manual acts in nautical and weaving techniques. The latter could be even rendered mechanical by the use of looms as early as in the 18th century. Although the first molecules made with two intertwined macrocycles were designed and synthesized in the sixties,1 that is, more than 50 years ago, the fabrication of topologically complex molecules still remains a challenging task. As a matter of fact, so far only the simplest non-trivial knot (the trefoil knot)2–5 has been the subject of extensive investigations, not to mention the recent report of the covalent capture of the pentafoil knot.6
Several synthetic strategies‡ have been imagined and developed for making interlocked and knotted molecules, which all rely on template effects.7 The latter can be classified either as kinetic or as thermodynamic. Kinetic template effects operate in the course of a reaction when molecular precursor fragments take up well defined geometrical and/or topological arrangements (preorganisation), which foreshadow the geometry and/or the topology of the target molecule. By contrast, a thermodynamically templated reaction involves a mixture of species in equilibrium, the latter being displaced towards the formation of the most stable product. Strictly speaking, a template is an auxiliary chemical species (atom, ion, or molecule) that is engaged in bonding interactions with the precursors (kinetic template) or the final product (thermodynamic template), and that can be removed once the reaction is complete. However, the operation of an external species may not be required, the interactions take place directly between the precursors or within the product. Templating interactions include covalent and coordination bonds, hydrogen bonding, π–π donor–acceptor and van der Waals interactions, and solvophobic effects, which influence reaction enthalpies and entropies to various extents. The concept of transition metal templating, which relies on coordination bonds, was initially developed for making simple macrocycles.7 It turned out to lend itself wonderfully for the rational synthesis of topologically complex molecules, as advantage was taken not only from stereoelectronic but also from chemical properties of transition metal cations.8 In this “Perspective” article we shall show that the implementation of metal-templation in the synthesis of these exotic molecular systems results from the interplay between topology and geometry. We hope that this observation will help researchers interested in this topic to carry out the synthesis of more and more complex knotted and intertwined chemical structures.
![]() | ||
Fig. 1 Different presentations of the embedding of the circular graph in the three-dimensional space. Note that B and C are chiral and are enantiomorphs. |
A molecule can be considered as topologically complex when none of the projections in the plane of all possible presentations of its embedded graph can be drawn without at least a crossing point. Particular in this respect is the case of molecules, the graph of which cannot be drawn in the plane without crossing points. Such graphs are qualified as intrinsically non-planar, and the minimally non-planar graphs, that is, those displaying only one crossing point are Kuratowski's graphs K5, the complete graph built on five vertices, and K3,3, the complete bipartite graph on six vertices, three of which connect to each of the other three. Rigid presentations of embeddings of these graphs in the three-dimensional space that have been used for the design of molecules with intrinsically non-planar graphs are given in Fig. 2. Noteworthily, molecular representatives of either K5 or K3,3 non-planar graphs have been synthesized in the past.11,12
![]() | ||
Fig. 2 Presentations of remarkable embeddings of the Kuratowski's non-planar graphs (a) K5 and (b) K3,3 in the three-dimensional space. |
In the following discussion, we shall consider only molecules that are topologically complex according to the most general criterion, that is, molecules with intrinsically non-planar graphs will be ignored here. The simplest topologically complex molecules belong to either the family of links, which result from interlocking of rings, or the family of knots, which result from tying a string and linking its two ends. Common and minimal presentations of the prototypical members of these families of molecules are represented in Fig. 3.
![]() | ||
Fig. 3 The four minimal presentations of a Hopf link and the Solomon links, the Borromean rings, and the trefoil knot. Note that the Solomon links and the trefoil knot are chiral. Only one enantiomer is represented here. |
In the family of links, [2]catenanes (topological figure: Hopf links) are made from two interlocked rings, doubly-interlocked [2]catenanes (topological figure: Solomon links) show four crossings, and the Borromean rings are made of three interdependent rings, yet any two of them are not interlocked. Strictly speaking, [2]rotaxanes made of a ring encircling a dumbbell-like fragment are not topologically complex since unthreading could easily be carried out mentally by shrinking the voluminous groups attached at the ends of the dumbbell – an operation which is of course impossible from an experimental viewpoint – however if the stoppers are large enough to prevent ring unthreading under any physically-accessible conditions (which comes down to having an infinitely long string component), the corresponding molecules behave as [2]catenanes. Therefore [2]rotaxanes can be treated as [2]catenanes, and the link making the two components interdependent is qualified as mechanical.13
It is the family of knots that contains the most topologically complex figures, the trefoil knot (denoted 31) being the first of the series of prime knots. Therefore, it is not at all surprising that most of the research in this field has focussed on devising methods for making molecules displaying this topology.2–5
All these figures are realized chemically using macrocyclic organic components, or, in a few instances metallo-organic macrocycles that incorporate a metal in their atom sequence.
![]() | ||
Fig. 4 Presentations of remarkable embeddings in the three-dimensional space of selected links and knots. (a) Hopf links, (b) Solomon links, (c) Borromean rings, (d) Trefoil knot. |
![]() | ||
Fig. 5 Synthesis of a [2]catenane from linear (L1) and cyclic (L2) precursors incorporating bidentate chelates for complexation of a tetrahedral metal cation template. The disk-ended bold blue lines represent 1,10-phenanthroline chelates and their N-donor atoms. The grey sphere is the Cu+ cation and f is a function that reacts either inter- or intramolecularly. The metal coordination geometry is enhanced by dotted red lines, as well as in the subsequent figures. |
![]() | ||
Fig. 6 Synthesis of a [2]catenane via “active metal template” strategy from linear (L3) and cyclic (L4) precursors. The disk-ended blue bold lines represent a chelate and its donor atoms, and the grey sphere is Cu+. The functions f (terminal alkyne) and g (bromoalkyne) react at the metal to form the new C–C bond sequence shown in red. |
![]() | ||
Fig. 7 Rational synthesis of a doubly-interlocked [2]catenane. The disk-ended coloured bold lines represent 1,10-phenanthroline chelates and their N-donor atoms, the grey sphere is the Li+ cation, and f represents reactive functions, in that case the terminal olefin fragments of trisphenanthroline L5. |
Interestingly, other reported examples of Solomon rings were obtained by serendipity. All proceeded under thermodynamic equilibrium conditions, using reversible, either covalent or metal–ligand bonds. For example mixing dpp-based chelates carrying pyridyl end groups L6 with Cu+ and Pd(en)(ONO2)2 afforded quantitatively doubly-interlocked [2]catenane [Cu2L64Pd4(en)4]10+ rather than the anticipated [2]catenane [CuL62Pd2(en)2]5+ that would have resulted from capture of two pyridyl ligands of the same dpp chelate by [Pd(en)]2+ complex subunits (Fig. 8).22 Such a transition metal-locked [2]catenane was indeed obtained in 92% isolated yield by using extended 4-pyridyl-substituted dpp ligands, such as L7, in which the pyridyl end groups of the same dpp can effectively clamp Pd2+ cations.23 Remarkably, the structure of this doubly interlocked [2]catenane corresponds to presentation E (Fig. 4b) of the Solomon rings.
![]() | ||
Fig. 8 Formation of a doubly interlocked [2]catenane by simultaneous Cu(I) templation and capture by Pd(II) metal complex fragment. The disk-ended coloured bold lines represent 1,10-phenanthroline chelates and their N-donor atoms, the isolated disks pyridyl ligands, the grey sphere is the Cu+ cation, and the brown sphere Pd2+ = [Pd(en)]2+ metal complex fragment. |
Similarly, Puddephatt and co-workers had shown that breaking the organometallic polymer based on Au(I) complexation of a bis-propargyl derivative of bisphenol A (L8) with Ph2P(CH2)4PPh2 (dppb) led to a [2]catenane made of organometallic [(L8-2H)Au2(dppb)] macrocyclic subunits held together by aurophilic interactions in 68% yield (Fig. 9).24 Quite interestingly, just changing the two methyl groups of the bisphenol A subunit by a cyclohexyl substituent (in L9) switched the equilibrium to the formation of the organometallic doubly interlocked [2]catenane [(L9-2H)2Au4(dppb)4] in 72% yield. Note that this latter system corresponds to the classical presentation (A) of the Solomon rings.25 It is worth mentioning that shorter diphosphines led to organometallic macrocycles.
![]() | ||
Fig. 9 Formation of a doubly-interlocked [2]catenane (one enantiomer shown) against a [2]catenane using aurophilic inter-metallomacrocycle interactions. The brown spheres are Au+ cations, the black and coloured disks represent carbyne and phosphorus binding atoms, respectively. For a given diphosphine the outcome of the reaction depends on the chemical nature of the bridge connecting the carbyne ligands (L8vs.L9). |
Finally, Stoddart and co-workers reported the formation in the solid state of a doubly-interlocked [2]catenane resulting from mixed Cu(II)–Zn(II) Schiff-base condensation of diamine L10 incorporating the 2,2′-bipyridyl chelate, and 2,6-pyridine dialdehyde L11 (Fig. 10).26 These Solomon rings are constructed from two macrocycles incorporating alternating exo bipyridyl and endo pyridyl diimine chelates, respectively, so that each metal template cation has an octahedral coordination sphere made from a bidentate, a terdentate, and a monodentate triflate ligand. Therefore they also correspond to presentation A of the Solomon rings. The doubly interlocked catenane is one of the unexpected possible reaction products that is in equilibrium, in particular, with the corresponding Borromean rings.
![]() | ||
Fig. 10 Self-assembly of a doubly-interlocked [2]catenane (one enantiomer shown) by mixed Cu2+–Zn2+ metal templation. The coloured bold lines represent a 2,6-pyridine diimine tridentate chelate and a 2,2′-bipyridine bidentate chelate, with the coordinating N atoms represented as disks. The brown and grey spheres represent Cu2+ and Zn2+ cations, respectively. A sixth binding site (not represented) is occupied by a trifluoroacetate anion, which makes the coordination spheres of the metal cations actually octahedral. |
We note that Solomon links derived from presentation D (Fig. 4b) have been reported by Quintela et al. However they will not be further discussed here as their formation relies also on donor–acceptor π–π stacking interactions.27
![]() | ||
Fig. 11 Cu+-templated formation of the metallo-organic Borromean rings structure constructed from ligand L12. The grey sphere is the Cu+ cation, the disk-ended coloured bold lines represent the 2,2′-bipyridyl chelate and the isolated disks stand for the nitrogen atoms of the pyridyl end-groups of L12. |
Various attempts using rational approaches for the synthesis of Borromean rings featuring fully organic rings were undertaken. They all relied either on the “ring-in-ring” presentation H or the “rack” presentation J.29 In spite of these efforts however, only a one-pot metal template synthesis using Schiff-base condensation of L10 and L11, but in the presence of Zn(II), afforded the spectacular hexanuclear “Borromean rings” Zn(II) complex (Fig. 12) in 95% yield.30 This was generalized to the synthesis of the so-called Borromeates, using divalent metals such as Cu(II), Ni(II), Cd(II), Mn(II), and Co(II).30b This remarkable achievement constitutes another example of how variations (even quite subtle) in ligand structure or nature of the template metal cation can be inferred in the topological outcome of the reaction product. This is of course the case when the reactions are thermodynamically controlled. In the present case, the newly created bonds can be considered as “covalent dynamic” bonds.31
![]() | ||
Fig. 12 Zn2+-templated Schiff-base condensation of 2,2′-bipyridine L10 functionalized with primary amine end groups and pyridine-2,6-dicarboxaldehyde L11 affords nearly quantitatively the hexanuclear Borromean rings structure. The grey sphere is the Zn2+ cation, the coloured bold lines represent a 2,6-pyridine diimine tridentate chelate and a 2,2′-bipyridine bidentate chelate, respectively, with the coordinating N marked as disks. A sixth binding site (not represented) is occupied by a trifluoroacetate anion, which makes the coordination spheres of the metal cations actually octahedral. |
![]() | ||
Fig. 13 Cu(I)-templated synthesis of a trefoil knot based on the double helical presentation M in Fig. 4. The disk-ended blue bold lines represent 1,10-phenanthroline chelates and their N-donor atoms. The grey sphere is the Cu+ cation and f is the allyl function for cyclization by intramolecular metathesis reaction. |
The approach taken by Hunter and co-workers,4 that is the one relying on presentation K in Fig. 4d and an octahedral metal template, had been actually devised in the 1970's by Sokolov in a seminal paper dealing with molecular topology.32 It remained to those authors to design and synthesize the tris-bipyridyl ligand (L14) that would form the appropriate interlaced structure in the presence of Zn2+ cations (Fig. 14). Because of its demonstrated efficiency in numerous cyclization reactions, the Grubbs metathesis reaction was again employed as a means of covalent sealing of the intermediate open knot structure, and this afforded the knot in Fig. 14 in 31% yield. Up to 68% yield was obtained when allyl functions were used. Catalytic hydrogenation followed by demetallation with lithium sulfide afforded the template-free trefoil knot.
![]() | ||
Fig. 14 Zn2+ templated synthesis of a trefoil knot from tris-bipyridine ligand L14 carrying allyl groups at its extremities. The disk-ended blue bold lines represent 2,2′-bipyridine chelates and their N-donor atoms. The grey sphere is the Zn2+ cation and f the allyl function for cyclization by intramolecular metathesis reaction. |
The so-called triskelion approach of Siegel and co-workers also uses the C3-symmetric presentation K in Fig. 4d, however, instead of placing the template in the center of the topological figure in an octahedral environment, it creates pseudotetrahedral ones at its three crossing points.33 Strictly speaking, this approach does not entirely rely on transition metal templating, the bipyridine chelates being held by a central benzene ring that plays the role of a permanent organic template. The three arms of the triskelion L15 spontaneously take up a C3-symmetric, helical arrangement because of steric hindrance. This produces the three required crossings, which are further fixed by Cu(I) coordination (Fig. 15).
![]() | ||
Fig. 15 The triskelion approach to the molecular trefoil knot. The disk-ended blue bold lines represent 2,2′-bipyridine chelates and their N-donor atoms. The grey sphere is the Cu+ cation and f the propargyl function for cyclization by intramolecular Glaser reaction. |
Triple cyclisation by Glaser coupling reaction of the dangling extremities produces the knotted structure in 85% yield.34 The template metal could be removed by reaction with cyanide.
The most recent transition metal-templated route to a molecular trefoil knot uses the “active metal” template concept invented and developed by the group of Leigh for making rotaxanes and catenanes (see Fig. 6).20 As in the approach of the group of Hunter the authors have developed a molecular thread (L16) incorporating binding sites (a pyridine and two bipyridines) in programmed positions, which in the presence of Cu(I) gives the knotted structure shown in Fig. 16 in 24% overall yield after removal of the metal template with EDTA.5 Two metal templates are required: the first fixes a loop by defining a crossing point at the bipyridine chelates, the second, bound to the pyridine located inside the loop, reacts with the functionalized extremities of the molecular thread, which simultaneously defines the two other crossing points and covalently fixes the knotted structure. The cyclization reaction chosen was the azide–alkyne Huisgen cycloaddition, because it is catalyzed by the same metal cation (Cu+) as the one used as geometrical template.
![]() | ||
Fig. 16 Active metal templated synthesis of a molecular knot. The grey sphere is the Cu+ cation. The disk-ended blue bold lines represent the 2,2′-bipyridyl chelates and the isolated disk stand for the nitrogen atom of the pyridyl central fragment. The functions f and g stand for the azide and the terminal alkyne groups that are used in the Cu(I)-catalyzed azide–alkyne Huisgen cycloaddition. |
Although it is not the purpose of this Perspective article to discuss and analyse the fascination that interlocking or knotted objects have exerted and continue to exert on molecular chemists, we would like to mention that the synthetic work done on such compounds has partly been motivated by their beauty, in relation to their topological properties and to their shape. The trefoil knot, for example, has been the object of many discussions for many decades before it was made at the molecular level. The synthetic challenge has also stimulated synthetic teams to tackle the problem. Before starting an actual preparation of these molecules, to design a convincing strategy was of utmost importance. As highlighted in this review article, the topological presentation of the object to be made was and still is determining and represents the best source of inspiration for conceptually elaborating a route to the target species. Looking back to the past, what has been realised in the course of the last decade or so was difficult to imagine twenty or twenty-five years ago. This success is to a large extent due to the power of the transition metal-based approaches proposed and implemented by the various research teams working in the field of topological chemistry. In twenty years virtually all conceivable approaches to the molecular trefoil knot have been explored, and a first synthesis of the pentafoil knot has been recently proposed. However the Borromean rings resisted so far controlled, stepwise assembly and the borromeates could not be obtained in metal-free form. In spite of these difficulties, there is no doubt that the synthesis of more complex knots and catenanes than those known today will be achieved within the next few years. Beyond these achievements, another point concerns the applications of interlocked and knotted molecules. Mechanical bonds have been used now for more than fifteen years for designing, fabricating and actuating molecular switches and motors, and the field of supramolecular polymers based on interlocked components is currently being developed spectacularly. By comparison, with very few exceptions properties associated with intentionally made knotted structures have yet to be identified or imagined, and much work remains to be done in this field.
Footnotes |
† This article is included in the All Aboard 2013 themed issue. |
‡ The earliest claimed catenane formation was based on statistical threading. See: E. Wasserman, J. Am. Chem. Soc. 1982, 82, 4433. |
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2013 |