We describe a strategy for assembling different macrocycles onto a nonsymmetrical rotaxane thread in a precise sequence. If the macrocycles are small and rigid enough so that they cannot pass each other then the sequence is maintained mechanically, affording stereoisomerism in a manner reminiscent of atropisomerism. The method is exemplified through the synthesis of a pair of [3]rotaxane diastereomers that are constitutionally identical other than for the sequence of the different macrocycles on the thread. The synthesis features the iterative binding of different palladium(II) pyridine-2,6-dicarboxamide complexes to a pyridine ligand on the thread followed by their macrocyclization by ring-closing olefin metathesis. Removal of the palladium(II) from the first rotaxane formed frees the pyridine site to coordinate to a second, different, palladium(II) pyridine-2,6-dicarboxamide unit which, following macrocyclization, provides a multiring rotaxane of predetermined macrocycle sequence.

We report the synthesis of a [2]catenate using a square planar palladium(II) template, together with two isomers of the interlocked structure: a single tetradentate macrocycle that adopts a "figure of eight" conformation to encapsulate the metal and a complex in which the two macrocycles of the catenane are not interlocked. The three isomers can each be selectively formed depending on how the building blocks are assembled and cyclized. Olefin metathesis of both building blocks while they are attached to the metal gives the single large macrocycle in 77% yield. Cyclizing the monodentate unit prior to attaching both ligands to the metal gives the [2]catenate in 78% yield. Preforming the tridentate macrocycle produces a complex in two atropisomeric forms-threaded and nonthreaded-in a 2:3 ratio, which do not interconvert in dichloromethane at room temperature over 7 days. RCM of the nonthreaded atropisomer affords the complex with two noninterlocked macrocyclic ligands; RCM of the threaded atropisomer generates the topologically isomeric [2]catenate. Heating the acyclic atropisomers in acetonitrile provides a mechanism for their interconversion via ligand exchange, allowing the threaded:nonthreaded ratio to be varied from 2:3 to 8:1. All three fully ring-closed complexes were characterized unambiguously by 1H NMR spectroscopy and X-ray crystallography. As far as we are aware, this is the first time such a set of three formal topological and constitutional isomers has been described.

Dedicated to Sir J. Fraser Stoddart on the occasion of his 65th birthdayThe introduction [1] and subsequent spectacular growth [2][3][4][5][6][7][8] of template strategies [2] to molecules with multiple mechanically interlocked components (rotaxanes, [3, 4, 8] catenanes, [5,6,8] and Borromean rings [7] ) has revolutionized approaches to their synthesis. However, despite the many different template systems and innovative assembly processes developed thus far, a feature common to all current methods is that they each employ at least one binding site per macrocycle, that is, a minimum of nÀ1 templates per n interlocked components. Here we describe a strategy (Figure 1) for assembling multiring rotaxanes in which just a single ligation site is used to clip on as many macrocycles as required (and the length of the thread will permit), through the repetition of three simple steps. The method (Scheme 1) involves the repetitive coordination and macrocyclization of a tridentate ligand for palladium about a stoppered molecular thread (L1), sequentially forming rings of type H 2 L2 to iteratively [8b] generate [2]-, [3]-, and [4]rotaxanes. Overcoming the restriction of only one ring per template unit-especially the ability to re-utilize the template despite the proximity of a previously cyclized component-should aid the synthesis of increasingly complex higher order interlocked assemblies (for example, defined sequences of different rings on a molecular strand).The single template site iterative assembly of multiple ring rotaxanes utilizes a square-planar geometry Pd II -based "clipping" methodology previously developed [9] as a "classical" (that is, one macrocycle per binding site) template synthesis of rotaxanes and catenanes. The metal atom is first bound to a tridentate 2,6-pyridinedicarboxamide ligand (significantly, this involves deprotonation of the ligand amide groups [9] ). The resulting complex [L3Pd(CH 3 CN)][9] is coordinated to a monodentate pyridine unit on a thread such that subsequent macrocyclization by ring-closing metathesis (RCM) occurs around the thread to generate an interlocked architecture.[10]The key to extending this protocol so that further ligands can be cyclized around the template after the first is decomplexed, is the observation that whilst [L3Pd(CH 3 CN)] will readily exchange its labile coordinated acetonitrile molecule for a pyridine unit (see Scheme 1, step a), it does not undergo tridentate ligand exchange with protonated (that is, metalfree) versions of the 2,6-pyridinedicarboxamide system (for example, H 2 L4 and H 2 L5; Scheme 1, steps d and g). Thus, a pyridine group on a thread can repetitively be used to replace acetonitrile ligands of [L3Pd(CH 3 CN)], with the resulting complexes then being macrocyclized one at a time.To demonstrate the effectiveness of this strategy in a multi-ring rotaxane synthesis, a suitable thread L1 was prepared (see the Supporting Information) with a single pyridine unit as the only potential template site and sufficient space-provided by a C 11 al...

The starting point for the revolution in catenane and rotaxane synthesis that occurred during the last part of the 20th century was the realization by Sauvage and co-workers that metalligand coordination geometries could fix molecular fragments in three-dimensional space such that they were predisposed to form mechanically interlocked architectures through macrocyclization or "stoppering" reactions.[1] Efficient synthetic methods to rotaxanes were subsequently developed based on four-(tetrahedral) [2] , five-(trigonal bipyramidal and square pyramidal) [3] and, most recently, six-coordinate (octahedral) [4] metal templates (Figure 1). [5] One of the benefits of using specific coordination motifs for such assemblies is that the resulting interlocked ligands often do not permit other metal geometries in their binding site, which can consequently be exploited either to lock a metal in an unusual geometry for its oxidation state [6] or to bring about large-amplitude "shuttling" of the ligand components. [3,7] Here we show that threedimensional interlocked architectures can also be assembled from two-dimensional coordination templates by using steric and electronic restrictions to direct the synthesis in the third